ORCA – Online Research @ Cardiff

Thi s is a n O p e n Acc e s s d o c u m e n t d o w nlo a d e d fro m ORCA, C a r diff U niv e r si ty' s

in s ti t u tio n al r e p o si t o ry: h t t p s://o r c a. c a r diff.a c. u k/i d/ e p ri n t/ 1 5 4 4 9 0/

This is t h e a u t h o r’s v e r sio n of a w o r k t h a t w a s s u b mi t t e d t o / a c c e p t e d fo r

p u blic a tio n.

Cit a tio n fo r fin al p u blis h e d ve r sio n:

Jia n g, Yali, Yan g, G a n g, Li, H aijia n g a n d Z h a n g , Tia n 2 0 2 2. Kno wl e d g e d riv e n

a p p r o a c h fo r s m a r t b ri d g e m ai n t e n a n c e u si n g bi g d a t a m i ni n g. Auto m a tio n in
Co n s t r u c tio n 1 4 6 , 1 0 4 6 7 3. 1 0. 1 0 1 6/j. a u t c o n. 2 0 2 2. 1 0 4 6 7 3

P u blis h e r s p a g e : h t t p:// dx. d oi.o r g/ 1 0. 1 0 1 6/j. a u t c o n. 2 0 2 2. 1 0 4 6 7 3

Pl e a s e n o t e:
C h a n g e s m a d e a s a r e s ul t of p u blis hi n g p r o c e s s e s s u c h a s c o py-e di ti n g, fo r m a t ti n g
a n d p a g e n u m b e r s m a y n o t b e r efl e c t e d in t hi s v e r sio n. Fo r t h e d efi nitiv e v e r sio n of
t hi s p u blic a tio n, pl e a s e r ef e r t o t h e p u blis h e d s o u r c e . You a r e a d vis e d t o c o n s ul t t h e
p u blis h e r’s v e r sio n if yo u wis h t o ci t e t hi s p a p er.

This ve r sio n is b ei n g m a d e a v ail a bl e in a c c o r d a n c e wi t h p u blis h e r p olici e s. S e e

h t t p://o r c a . cf. a c. u k/ p olici e s. h t ml fo r u s a g e p olici e s. Co py ri g h t a n d m o r al ri g h t s fo r
p u blic a tio n s m a d e a v ail a bl e in ORCA a r e r e t ai n e d by t h e c o py ri g h t h ol d e r s .
1 Knowledge Driven Approach for Smart Bridge Maintenance Using Big Data
2 Mining

3 Yali Jianga, Gang Yang a, Haijiang Li b,*, Tian Zhang a, *

4 aCollege of Transportation Engineering, Dalian Maritime University, Dalian 116026, China

5 bSchool of Engineering, Cardiff University, Cardiff CF24 3AA, UK

6 * Corresponding authors: lih@cardiff.ac.uk (H.J. Li); saghb@126.com (T. Zhang);

8 Abstract: Life cycle bridge maintenance is highly complex and multi-disciplinary oriented,

9 different ICT technologies have been widely adopted, but the generated data and information

10 are often intensive, specific and isolated, it is very difficult to contribute effectively for holistic

11 bridge maintenance decisions. This paper investigates state-of-the-art methods used in

12 bridge maintenance, a total of 2732 papers were selected for visualization analysis and

13 323 papers were pinpointed for further critical review. The review informs that mindset

14 shifting from traditional and pre-digital, through data driven to knowledge-based

15 approach is required for bridge engineers to holistically understand multi-sources of

16 data and information to enable systematic thinking. The review further reveals the need

17 for a knowledge-driven approach that can leverage bridge maintenance big data to

18 provide smart holistic decisions, a novel knowledge-oriented framework and

19 methodology were proposed in the end with an aim to unify and streamline different

20 sources of data and information to facilitate new developments towards smart bridge

21 maintenance.

22

23 Keywords: Bridge maintenance; Multi-disciplinary; IoTs; Visualization analysis;

24 Holistic decision making; Knowledge-driven approach; Big Data ; BIM

25

26 1 Introduction

27 Bridges are essential for highway networks and play an important role in human

28 society [1,2]; they are facing ageing challenges involving factors, such as

1
29 environmental corrosion, vehicle overload and human-made hazards [3,4]. According

30 to the 2016 Canadian Infrastructure Report, 25% of the existing bridges are in poor or

31 very poor condition [5]. According to the 2017 American Society of Civil Engineering

32 report, approximately 240,000 bridges have exceeded their 50-year service life, and

33 over 56,000 bridges have been classified as structurally deficient [6]. In China, the total

34 number of dangerous bridges posing serious safety risks to human society was

35 approximately 70000 by the end of 2017 [2]. To restore the sub-standard bridges back

36 to perfect condition, ¥69.7 billion was invested in the renovation of 34,000 dangerous

37 bridges in China between 2016 and 2020 [7]. According to the UK government, the

38 number of substandard bridges has risen to more than 3,100 by January 2021 [8], and

39 the cost for proper maintenance for those sub-standard bridges was estimated to be

40 £1.16 billion by March 2022 [9]. Therefore, an effective maintenance procedure can

41 considerably reduce the cost [5], and this becomes increasingly pivotal [10].

42 Bridge maintenance tasks are complex and multi-disciplinary oriented, e.g.,

43 structure, cost, health and safety, sustainability and environmental issues. resulting in

44 an extremely complex process. Hence, proactive, holistic and smart lifecycle

45 approaches are required to comprehend the complexity of bridge structural conditions.

46 Embracing advanced technologies has continuously improved the intelligence level of

47 bridge maintenance. The wide application of IoTs along with traditional bridge surveys,

48 e.g., non-destructive technologies [11] and sensors [12], make it possible to obtain

49 more comprehensive detection and monitoring data. Unstructured and structured

50 datasets are quickly accumulated to form multisource, heterogeneous and

51 autonomous maintenance ‘Big Data’, but only a small part of those are utilised

52 successfully for bridge maintenance decision-making. Cloud computing [13], building

53 information modelling (BIM) [14], artificial intelligence [15,16], etc., have greatly

54 improved the data processing power. Virtual reality (VR) [17], digital twins (DT) [5],

55 semantic web technology (SWT) [18], etc., provide a more intelligent data visualisation

56 and reasoning approach for superlarge-scale bridge maintenance. All of these

2
57 innovations are leading the industry towards a more productive, more effectively

58 managed digital age, where real-time data and project reporting will be available for

59 maintenance projects. Recent developments are showing a gradual shift from data to

60 knowledge-driven decision-making for bridge maintenance, and those accumulated

61 large amounts of data can be turned into a vast base for knowledge mining for

62 knowledge-driven approaches.

63 Regarding smart bridge maintenance, some studies focus on specific technical

64 applications, e.g., Abu Dabous et al. [19] comprehensively reviewed the commonly

65 used noncontact testing technology for condition monitoring of concrete bridges.

66 Agnisarman et al. [20] reviewed the application of automated visual inspection

67 technology to inspect infrastructure, such as bridges. Fujino et al. [21] and Sun et al.

68 [22] discussed the prospects and driving forces of big data technology in bridge

69 monitoring. Zhou et al. [2] compiled the development of China’s bridge maintenance

70 information system. These articles indicated that using information technologies to

71 improve the quality and efficiency of management is the unanimous choice of bridge

72 engineers. Other studies review details of analytical methods, e.g., Banerjee et al. [23]

73 conducted a systematic and comprehensive review of the literature on the resilience

74 assessment of bridges and bridge networks under single and multiple hazardous

75 conditions. Kabir et al. [24] reviewed the application of multicriteria decision-making

76 technology in the field of infrastructure management. Frangopol et al. [25] briefly

77 reviewed the research results related to the design, maintenance, and lifecycle

78 management of infrastructure (involving bridges). Although these reviews cover the

79 application of advanced technologies or analytical methods in bridge maintenance,

80 thus showing benefits from data-driven approaches, they have not been pushed further

81 to examine the potential of knowledge-based approaches. The relevant

82 understandings about knowledge-driven approaches are still missing.

83 To address these gaps, this paper provides a critical review of smart bridge

84 maintenance by collecting and analysing a large number of papers from the Web of

3
85 Science (WoS) core collection database, using literature visualisation analysis and

86 critical review to study and conclude key areas for knowledge-driven smart bridge

87 maintenance. The wider review visually analyses bridge maintenance using the

88 CiteSpace software (6.1.R2), which covers four perspectives: literature quantity

89 analysis, journal co-citation analysis, document co-citation analysis, keywords

90 clustering and burst analysis; the later focused review summarises knowledge-driven

91 smart bridge maintenance aiming at three areas: bridge maintenance tasks and issues,

92 advanced technologies supporting smart maintenance, and holistic decision-making

93 approaches. Based on the above review and analysis, a novel knowledge-oriented

94 framework is finally proposed, with the aim of facilitating new developments towards

95 smart bridge maintenance.

96 2 Review and Analysis Methodology

97 To explore the state-of-the-art development of intelligent technologies in bridge

98 maintenance, a literature analysis is performed on the WoS core collection database.

99 WoS is an important database resource for obtaining global academic knowledge

100 supported by powerful combined retrieval functions [26,27]. The search is based on

101 using the ‘AND’ and ‘OR’ operators search benchmark, and the search code in the

102 database is as follows:

103 TS = (xxx* AND/OR xxx*)

104 Knowing that ‘TS’ represents the topic of the articles, ‘xxx’ and ‘*’ are standard for

105 the search term and the fuzzy search, respectively. Further details of the operators’

106 search benchmark are as follows:

107

4
108
109 Fig. 1. Steps to search for papers in the WoS core collection database.

110 This search in this paper is composed of four steps (concluded in Fig. 1): in step

111 1, papers published between January 1, 2000, and December 31, 2021, are retrieved

112 from the database on the topic of bridge maintenance. Papers that are not related to

113 bridging engineering are excluded, such as the cytology, immunology and oncology

114 categories, which include the term “intercellular bridge”. In addition, only articles and

115 review articles written in English were selected for the document types because of their

116 high quality and cutting-edge research [28]. After preliminary filtering, 2732 papers are

117 selected, and their contents reveal that BIM, IoT, cloud technology and other

5
118 technologies are often mentioned in smart bridge maintenance driven by data or

119 knowledge approaches. Consequently, in step 2, the following search benchmarks are

120 jointly used in the WoS core collection database to refine the search results: ‘BIM OR

121 “Building Information Model*”’, ‘IoT OR “Internet of Things”’, ‘“Big Data”’, ‘cloud’, and

122 ‘Semantic’.

123
124 Fig. 2. Network of co-occurring keywords for AI.

6
125
126 Fig. 3. Network of co-occurring keywords for the intelligent bridge.

127 Furthermore, to avoid missing important papers, step 3 searches for intelligent

128 technology-related keywords in the bridge engineering field by using ‘AI OR “Artificial

129 Intelligence”’ and ‘“Intelligent Bridge”’ as topics. As shown in Figs. 2 and 3, the

130 CiteSpace software is used to rank and visualise the occurrence frequency of all the

131 keywords. In the CiteSpace network, each node shown with a coloured circle (or cross,

132 triangle, square) represents an object. Some objects are linked by lines. The thickness

133 of the link is used to indicate the partnership strength. The colour is used to correspond

134 to different years. In the keyword co-occurrence network, each node represents a

135 keyword, and the node’s size (or font size) reflects the frequency of the keyword

136 occurring in the dataset. The higher the frequency is, the larger the node size.

137 According to the results in Figs. 2 and 3, 10 keywords with high frequency and related

138 to data- or knowledge-driven approaches are selected for research. A total of 323

139 papers were obtained. In step 4, 2732 papers are visually analysed to infer the

140 research hotspots and trends of intelligent technology application in bridge

141 maintenance, and a critical review analysis of 323 papers summarise the development

142 status of smart bridge maintenance.

7
143 3 Literature Visualisation Analysis

144 The visualisation analysis relies on four types of bibliometric techniques [29,30]

145 applied using CiteSpace, including literature quantity analysis, journal co-citation

146 analysis, document co-citation analysis, keywords clustering and burst analysis.

147 CiteSpace maps the knowledge domain by systematically creating various accessible

148 graphics, which can discover the semantic knowledge hidden in a large amount of

149 information and track the development frontier of technology [31,32]. CiteSpace

150 software provides multiple options for input thresholds, e.g., time slicing, data selection

151 criteria, and pruning strategies. Sensible input thresholds can make the generated

152 network layout clearer and more reasonable.

153 3.1 Literature quantity analysis

154 As shown in Fig. 4, the paper publication times are identified according to the

155 information in the bibliographic records. The number of papers published has grown

156 steadily with slight fluctuations. In 2000, the number was only 27. In 2021, the number

157 reached 414. In 2008 and 2014, there were slight fluctuations. The decrease is

158 probably due to the limited budgets and lack of data for decision-making, which are

159 long-existing challenges for bridge maintenance.

450
414
400

350 321

300
260
250 221228
Count

200
164
150
150
114121112
93
100 77
66
55 64 63 55
39 42
50 27 21 25

0
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022
Time
160
161 Fig. 4. Statistical graph of the number of papers over time.

162 Fig. 5 shows the co-occurring network of countries. In this network, the time slice

8
163 length is 2. The criteria of data selection are g-index (k=25), LRF=3.0, L/N=10, LBY=8,

164 and e=2.0. There are 92 nodes and 398 links. Each node represents a country, and

165 the node’s size reflects the number of published articles in that country. The more

166 papers a country publishes, the larger the node size. The United States has issued the

167 most papers, followed by China. This is not surprising because the peak period of

168 bridge construction in developed countries, such as the U.S. and countries in Europe,

169 was concentrated in the 1950s and 1970s. The large-scale ‘ageing’ of bridge structures

170 appeared earlier than in China. Therefore, European countries and the U.S. have

171 conducted relatively more research on bridge maintenance technology. There are

172 many problems during bridge operation due to insufficient maintenance in China. Many

173 bridges even collapsed due to improper management, e.g., Qijiang District Rainbow

174 Bridge (1999.1), Liaoning Panjin Tianzhuangtai Bridge (2004.6), Sichuan Panzhihua

175 Jinsha River Bridge (2012.12), Guangdong Heyuan Chengnan Ramp Bridge (2015.6)

176 and Yilan Bridge (2019.10). Therefore, domestic scholars continue research to solve

177 problems in the field of bridge maintenance in China.

178
179 Fig. 5. Countries’ co-occurring network.

180 3.2 Journal co-citation analysis

181 Journal co-citation analysis reflects the correlation between various journals.

9
182 Through this type of analysis, the intellectual root sources for published works in a field

183 are obtained. The number of co-citations of various journals is shown in Fig. 6, where

184 the time slice length is 2. The selection criteria are Top 50, LRF=3.0, L/N=10, LBY=8,

185 and e=2.0. To remove excessive links, network pruning is used through the Pathfinder

186 strategy, which was recommended by Chen and Morris [33]. In this network, there are

187 201 nodes and 604 links. Each node represents a journal, and the node’s size

188 represents the number of times the journal has been co-cited. The more times the

189 journal is co-cited, the larger the node size. Among them, “Engineering Structures”,

190 “Journal of Bridge Engineering”, “Journal of Structural Engineering”, and “Structure and

191 Infrastructure Engineering” have the most co-citations.

192
193 Fig. 6. Journal co-citation network.

194 Moreover, if a node connects two or more large groups of nodes with the node

195 itself in between, it has high betweenness centrality represented by a purple ring in

196 CiteSpace. Table 1 lists journals with betweenness centrality values greater than or

197 equal to 1.0. The journal “Engineering Structures” has the highest centrality, with a

198 value of 0.27. The high centrality represents a large amount of importance for these

199 journals. These analysis results provide a basis for follow-up in-depth research, and

200 follow-up researchers can examine the direction of smart bridge maintenance in depth

10
201 by collecting papers from these top journals.
202 Table 1. Cited journals sorted by centrality.
Cited Journals Centrality Count
Engineering Structures 0.27 910
Structural Engineering International 0.20 224
Journal of Infrastructure Systems 0.17 433
Computer-Aided Civil and Infrastructure
0.17 364
Engineering
Structural Safety 0.12 409
Computers & Structures 0.12 307
Journal of Computing in Civil Engineering 0.12 304
Reliability Engineering & System Safety 0.10 379
Automation in Construction 0.10 283

203 3.3 Document co-citation analysis

204 Document co-citation analysis demonstrates the quantity and authority of

205 references and their authors cited by publications [34]. Leading researchers for a

206 knowledge domain can be identified. Fig. 7 shows the co-citations network of various

207 documents, where the time slice length is 2. The selection criteria are Top 50, LRF=3.0,

208 L/N=10, LBY=8, and e=2.0. Network pruning is a Pathfinder strategy. In this network,

209 there are 511 nodes and 744 links. Each node represents a document with the first

210 author's name and the publication year, and the node’s size represents the number of

211 times the document has been co-cited.

11
212
213 Fig. 7. Document co-citation network.

214 The top 10 documents are summarised in Table 2. Frangopol [35-39] received the

215 most attention, with 238 citations in total. This was followed by Melchers [40], Okasha

216 [41], Biondini [42], Sabatino [43] and Kim [44], with approximately 30 citations each.

217 According to the WOS citation metrics, Frangopol has 650 publications in the WOS

218 database, with a total of 15,543 citations and 7,030 citing articles. Melchers has 307

219 publications with a total of 7,548 citations and 4,729 citing articles. In addition,

220 documents with high betweenness centralities are also worth attention. These

221 documents with betweenness centrality values greater than or equal to 0.15 are listed

222 in Table 3, including authors, Bocchini [45,46], Furuta [47], Van Noortwijk [48], and Liu

223 [49]. According to the WOS citation metrics, the number and citations of published

224 articles are high. Therefore, all documents listed in tables can be regarded as the major

225 intellectual turning points, and their authors are leading researchers in the field of

226 bridge maintenance.

227 Table 2. Cited documents sorted by count.

CiteSpace Metrics WOS Citation Metrics

Cited References
times citing
count centrality publications
cited articles

12
 Frangopol DM, 2011, Structure and
75 0.05 650 15543 7030
Infrastructure Engineering [35]
Frangopol DM, 2017, Structure and
44 0.01 \ \ \
Infrastructure Engineering [36]
Frangopol DM, 2007, Structure and
43 0.02 \ \ \
Infrastructure Engineering [37]
Frangopol DM, 2001, Journal of
38 0.08 \ \ \
Computing in Civil Engineering [38]
Frangopol DM, 2016, Structure and
38 0.06 \ \ \
Infrastructure Engineering [39]
Melchers RE, 2018, Structure and
34 0.14 307 7548 4729
Infrastructure Engineering [40]
Okasha NM, 2009, Structural Safety
33 0.02 26 601 433
[41]
Biondini F, 2016, Journal of
31 0.01 112 1596 1054
Structural Engineering [42]
Sabatino S, 2015, Engineering
30 0.11 16 316 256
Structures [43]
Kim S, 2013, Journal of Structural
28 0.08 38 1123 684
Engineering [44]

228
229 Table 3. Cited documents sorted by centrality.

CiteSpace Metrics WOS Citation Metrics

Cited References
times citing
centrality count publications
cited articles
Bocchini P, 2011, Structural
0.26 5 51 1284 1134
Safety [45]
Furuta H, 2006, Structure and
0.21 11 75 966 739
Infrastructure Engineering [47]
Van Noortwijk JM, 2004,
Probabilistic Engineering 0.17 11 44 2029 1506
Mechanics 48]
Bocchini P, 2011, Reliability
0.15 16 \ \ \
Engineering & System Safety [46]
Liu M, 2006, Journal of Bridge
0.15 9 42 548 492
Engineering [49]

230

231 3.4 Keywords clustering and burst analysis

232 The current research trend of intelligent technology is explored through keywords

13
233 clustering maps and burst word analysis. Cluster analysis is used to detect and analyse

234 the emergence of research trends over time and identify the focus of research trends

235 at a specific time in its knowledge base [31,32]. Clustering can reveal the

236 interconnection between different research trends. Burst words represent a substantial

237 increase in the number of occurrences of the keyword in a short period of time, which

238 indicates that such articles have attracted substantial attention in the corresponding

239 year [31,32]. First, a keyword cooccurring network is generated. As shown Fig. 8,

240 keywords co-occurring network has 516 nodes and 1,442 links. The time slice length

241 is 2. The selection criteria are g-index (k=25), LRF=3.0, L/N=10, LBY=8, and e=2.0.

242 Network pruning is a Pathfinder strategy. There are 10 keywords with frequencies over

243 100, including bridge maintenance (frequency = 277), model (frequency = 208), system

244 (frequency = 203), structural health monitoring (frequency = 165), bridge management

245 (frequency = 163), optimisation (frequency = 154), etc.

246
247 Fig. 8. Keywords co-occurring network.

248 Second, a total of 11 important clusters are identified and shown in Fig. 9 based

249 on the keywords by the log likelihood ratio (LLR) algorithm. The LLR algorithm can

250 select the best cluster labels in terms of uniqueness and coverage [34]. Clusters are

14
251 sorted by size, i.e., the number of members the cluster contains. The cluster #0

252 “optimisation” is the largest, with 71 members, while the cluster #11 “seismic effects”

253 is the smallest, with only 6 members. Table 4 lists all of the clusters and their

254 information, including “cluster size”, “silhouette value”, “mean year”, and “LLR label”.

255 The silhouette metric measures the average homogeneity of a cluster [50]. The greater

256 the silhouette score represents, the more consistency of the cluster members. The

257 silhouette values of clusters range from 0.667 to 0.946, which indicates that the

258 members of each cluster are sufficiently consistent. The mean year of publication of a

259 cluster refers to whether it consists of recent papers or older papers. Except for clusters

260 #0 and #7, all other clusters are formed by recent papers. Based on the cluster map,

261 clusters #0-11 form the application framework of the new-generation information

262 technology with intelligent algorithms and BIM as the core to support bridge

263 maintenance and management.

264
265 Fig. 9. Cluster map of keywords.
266 Table 4. Clusters sorted by size.
Cl Mean
Size Silhouette Label (LLR)
uster ID (Year)
optimization; maintenance; uncertainty; life-
#0 71 0.898 2007
cycle cost; genetic algorithm

15
 structural health monitoring; damage
#1 69 0.816 2013 detection; cable-stayed bridge; operational
modal analysis; system identification
corrosion; concrete structures; reinforced
#2 69 0.724 2010
concrete; service life; finite element analysis
building information modeling (bim); life
#3 55 0.694 2014 cycles; risk management; bridge management
system; structural health monitoring
bridge inspection; bridge health monitoring;
#4 46 0.751 2016
digital twin; big data; optimization
asset management; bridge management
#5 45 0.667 2013 system; railway bridge; bridge management;
transition probability
machine learning; deep learning; artificial
#6 34 0.825 2015
intelligence; computer vision; machine vision
bridge deck; bridge tests; fiber reinforced
#7 33 0.847 2008 polymers; composite materials; signal
processing
system reliability; semi-integral bridges;
#8 22 0.853 2010 integral bridges; performance prediction; finite
element
condition assessment; masonry arch bridges;
#9 21 0.881 2017 long-span bridges; point cloud; terrestrial laser
scanning
reliability analysis; concrete cracking; bond
#10 13 0.941 2013 strength; reinforcement corrosion;
maintenance planning
seismic effects; probabilistic modelling;
#11 6 0.946 2015 hysteresis; generalised extreme value
distribution; existing concrete bridges
267

268 Finally, burst detection is carried out based on the algorithm developed by

269 Kleinberg [51]. The top 41 keywords with the strongest citation burst are sorted by

270 strength in Fig. 10. Lifecycle cost (2004-2013) received the strongest attention, with a

271 burst strength of 11.6, followed by bridge maintenance (burst strength = 11.13, 2004–

272 2009) and machine learning (burst strength = 9.76, 2018–2021). Some keywords have

273 always been the focus of attention in the field of bridge maintenance, such as bridge

274 deck (burst strength = 9.68, 2002-2013), concrete structure (burst strength = 9.14,

275 2000–2013), oriented multiobjective optimisation (burst strength = 8.01, 2006–2017)

276 and lifecycle (burst strength = 5.92, 2002–2015) with long duration and high strength.

16
277
278 Fig. 10. Top 41 keywords with the strongest citation burst.

279 Fig. 11 shows the classification of some burst keywords extracted from Fig. 10

280 that are extremely relevant to bridge maintenance. These burst keywords are sorted

281 by the beginning time. From the perspective of maintenance projects, management

282 has drawn attention earlier via bridge management system (burst strength = 3.3, 2006-

283 2011), bridge management (burst strength = 3.15, 2008-2013), and maintenance

17
284 management (burst strength = 3.25, 2010-2013). Inspection has recently started to

285 gain attention via maintenance & inspection (burst strength = 5.93, 2016-2019). From

286 the perspective of methods or technologies, the application of new-generation

287 information technology, e.g., machine learning (burst strength = 9.76), deep learning

288 (burst strength = 4.99), building information modelling (burst strength = 4.17), and

289 digital twin (burst strength = 3.88), in the bridge maintenance phase has exploded

290 recently. Moreover, in general, keywords cluster analysis can help map documents and

291 classify them. However, when the topic is relatively new or the number of certain

292 keywords is not enough to form a category, it is easy to ignore. Therefore, a critical

293 review analysis is adopted in the following section to further analyse the collected

294 literature.

295
296 Fig. 11. Burst keywords classification.

297 4 Smart bridge maintenance – from data to knowledge-driven

298 After visually analysing the collected papers, those papers are further reviewed to

299 identify the gaps and shape the vision development for smart bridge maintenance.

300 Critical review analysis process is shown in Fig. 12. First, the tasks of bridge

301 maintenance are summarised. Then, the current issues that limit development are

302 identified. Finally, aiming at these issues, advanced technologies or approaches and

303 their development trends are analysed.

18
304
305 Fig. 12. Critical review analysis process.

306 4.1 Bridge maintenance tasks and issues

307 Bridge maintenance is very complex. According to all of these collected articles,

308 bridge maintenance systems normally include four parts: detection, evaluation, MR&R

309 (maintenance, repair, and rehabilitation), and management (Fig. 13). Bridge detection

310 is the cornerstone of checking the hazards of bridges [52]. It is basic to evaluate the

311 safety and maintenance of bridges. The development of detection technology has

312 shifted from visual inspection in the early stage to full-coverage detection, which

313 combines visual inspection, equipment inspection and monitoring technology [53].

314 Research shows that non-destructive technologies (NDTs) [54-57] are popular

315 development directions. In the next stage, further exploration is carried out by

316 combining new-generation information technologies, such as unmanned aerial vehicle

317 systems [58-60] and robots [61-63].

19
318
319 Fig. 13. Bridge maintenance system.

320 Bridge evaluation is used to assess the condition of the structure by

321 comprehensively describing the defects of each component. In addition, it can provide

322 decision support for bridge MR&R [64]. Bridge evaluation is divided into general

323 evaluation and adaptability evaluation. General evaluation refers to the comprehensive

324 assessment of each component’s technical condition to determine the bridge’s level of

325 technical condition. The data mainly come from the periodic survey. Adaptability

326 assessment refers to the evaluation of the actual bearing capacity, traffic capacity and

327 flood resistibility of bridges by combining the test and structural stress analysis. The

328 data are mainly from periodic surveys and special monitors. In addition, some studies

329 defined bridge evaluation as the observation and evaluation of the state of the built

330 structure, including damage identification [65,66]. The purpose of damage

331 identification is to find possible local damage, which is used in emergencies, such as

332 ship collisions, strong winds, and earthquakes. At present, an evaluation standard

20
333 system matching the existing detection technology level has been formed for small and

334 medium-sized bridges.

335 Bridge MR&R refers to the regular maintenance and repair operations for the

336 normal use of bridges. More precisely, tasks are carried out to prevent and repair

337 catastrophic damage to bridges and improve bridge quality and service levels [10].

338 Bridge maintenance activities are typically divided into two categories [67,68]: time-

339 based maintenance (TBM) or regular maintenance. It is protective maintenance

340 behaviour to delay or postpone the degradation of the structure; the other is condition-

341 based maintenance (CBM), which is reinforcement maintenance behaviour to enhance

342 structural performance. Due to the conventional hazards of bridge structures, a

343 relatively mature MR&R technique has been formed. These methods have been widely

344 used in the repair and reinforcement of small- and medium-sized bridges [69-71].

345 Bridge maintenance involves both technology and data/information management.

346 Storing the data/information can better serve the follow-up maintenance behaviours.

347 The bridge management system (BMS) [72] and structural health monitoring system

348 (SHMS) [73] are the two most developed systems in bridge maintenance. SHM has

349 been widely used in the maintenance and management of long-span bridges. The BMS

350 coverage can be extended to all bridges, including small- and medium-span bridges.

351 The combination of the two systems can provide full coverage of the bridge network.

352 The complexity of bridge maintenance is reflected in the fact that the system has

353 massive data and rich knowledge to work with; it involves various information from

354 multiple sources, and a large number of stakeholders and organisations collaborate

355 throughout its entire lifecycle. When massive data, information, equipment and people

356 are intertwined, it is concluded that there are several critical issues that need to be

357 addressed, including (1) equipment constraints and subjective surveys; (2) data and

358 information silos; (3) superficial data mining; and (4) lack of holistic decisions.

359 Equipment constraints and subjective surveys – With the existing equipment

360 constraints, it is still not easy to have the most cost-effective technical solutions or

21
361 equipment for a large amount of data acquisition. The data collected often lack good

362 quality, and the data acquisition accuracy is low. Structural defects are identified and

363 classified manually by engineers and inspectors who need to control the entire process

364 from hazard discovery, testing, recording and entry of results, and the entire process

365 is highly error prone and unreliable.

366 Data and information silos –The whole lifecycle bridge maintenance involves

367 many different hardware and software systems to work together; the data

368 interoperability issues are still critical, as there are no mature solutions to help to

369 overcome data and information silos. For example, BMS is used to manage bridge

370 survey data, and the generated information regarding bridge structural condition is

371 stored in the inspection reports. SHMS is used for monitoring the external

372 environment and structural response, and it can deal with large-capacity data

373 measurement, transmission and storage. However, these two systems work mostly

374 independently; hence, the survey cannot be timely and effectively shared through

375 different working stages and systems. The intuitionistics of manual inspection and real-

376 time health monitoring are not sufficiently integrated.

377 Superficial data mining – The current maintenance is more focused on collecting

378 rather than utilising data. Over the years, bridge systems have collected a large amount

379 of survey data, but only a small part of those are utilised successfully for bridge

380 maintenance decision-making. Various types of data have a low degree of correlation

381 and lack connectivity analysis, which means that the potential scientific value of the

382 obtained data has not been fully explored. The diversification of data formats (e.g.,

383 structured data, unstructured data) also increases the difficulty of information sharing

384 and integration.

385 Lack of holistic decision-making – Generally, the knowledge and information in

386 the whole process of bridge maintenance are dispersed among different teams. People

387 with different skills and professional backgrounds perform various tasks, and different

388 engineers tend to focus on their deliverables and operate in silos. As a result,

22
389 information and knowledge are not easily shared between different departments. In the

390 absence of effective computer-aided tools, it is difficult for a single person or team to

391 master multidomain knowledge. Therefore, decision-makers tend to rely on subjective

392 experience to make critical solutions, which are often not holistic or comprehensive.

393 4.2 Data Acquisition Technologies

394 For critical issues in bridge maintenance, embracing advanced technologies and

395 methods has become essential. First, advanced techniques can collect rich data to

396 accurately reflect bridge conditions and serve as a basis for maintenance decision-

397 making. Second, analytical methods are developed to deeply mine the raw data, derive

398 meaningful data, and make maintenance decisions. In practice, maintenance data are

399 mainly captured by sensors or NDTs. Some data that are difficult to collect with sensors

400 and NDTs can be obtained from second-hand sources, e.g., some environment and

401 social data that need to be collected from databases of relevant agencies [1]. Table 5

402 summarises mainstream data types for bridge maintenance. Data collected by

403 sensors are divided into three groups: environment data, traffic data and structure

404 response data. Structural data, acceleration, strain, and stress are the most common

405 types because they are the basis for most structure analyses [74]. Major data types

406 collected by NDTs include bridge profiles, point cloud data, photo images, acoustics,

407 radar, infrared images, electrical data and chemical data. Bridge profiles are the most

408 common meaningful data, followed by structure response data. Moreover, due to the

409 strict requirements on equipment, environment, and operators to generate point clouds

410 [75], more studies choose to create point clouds from photos [76,77].
411 Table 5. Mainstream data types for bridge maintenance.
Data acquisition
Data type Instances
manner
Temperature
Wind speed
Sensors Environment data Humidity
Topography and geology
Hydrology

23
 Traffic volume
Traffic data Vehicle load
Vehicle speed
Strain and stress
Displacement and deflection
Structure responses Cracks and deformation
Acceleration
Vibration and frequency
Properties of components, e.g.
Bridge profile geometry, number, material properties,
and connection of components

Bridge structure features: 1) both local

Point clouds features, e.g. girder cracks, and global
NDTs features, e.g. holistic bridge geometry.
Photo images 2)only reflect surface conditions.
Acoustics and ultrasonic
Local features: reflect both surface and
Radar
subsurface conditions.
Infrared images
Electrical and chemical Structural corrosion

412
413 For data collection sensors, a SHM system is the dominant method. Various

414 sensors and bridge structures are fused as a whole system through IoTs, which may

415 gather the various sensor data of bridges regularly or in real time to provide a scientific

416 basis for decision-making in bridge maintenance [78,79]. However, SHM systems are

417 only installed on the superstructures of critical bridges in practice, while other parts are

418 ignored. Therefore, many studies have optimised sensor placement problems based

419 on intelligent algorithms [53,80] to consider the quality and cost-effectiveness of data

420 acquisition. The current optimal placement methods of sensors include the effective

421 independent method, MinMAC (Minimise Modal Assurance Criterion), modal matrix

422 summation and integration method, origin residual method, modal matrix QR

423 decomposition method, SVD (Singular Value Decomposition), GRM (Guyan Reduction

424 Method), etc. Moreover, the sensor network can be wired or wireless. It is recognised

425 that wireless sensor networks (WSNs) are easier to maintain, more stable, and more

426 cost-effective in the long term [12,81]. However, the reliability of WSNs is affected by

24
427 several technical challenges, e.g., the lack of a power supply and unstable sensor

428 communication. In practice, the wired system is still implemented more widely,

429 although its data transferring relies on traditional cables rather than on wireless

430 communication techniques.

431 For data collection devices of NDTs, the application of unmanned aerial vehicles

432 (UAVs) or drones, mobile detection terminals, and wall-climbing robots has greatly

433 improved the safety, efficiency, and intelligence of data acquisition. For example, a

434 drone can conveniently access hard-to-reach areas to capture data efficiently using

435 sensors (e.g., infrared cameras) that it carries. However, these devices require

436 additional calibration, path planning, and control to avoid being blocked by barriers [74].

437 Another research hotspot is the intelligent detection equipment loaded with image

438 recognition technology, and then the computer vision technology is used to process

439 data for timely analysis to determine the crack profile, width, length, and propagation

440 direction [5]. Hence, the demand for data processing in real-time is met. In addition,

441 facing the strategic needs of emergency rescue in major natural disasters and access

442 to transportation facilities, research on rapid diagnostic technology of bridge

443 performance after disasters has attracted the attention of scholars [83].

444 Existing studies on data acquisition technologies mainly focus on the innovation

445 of modern inspection devices. Modern inspection devices can still be inefficient, as

446 they require sending staff onsite to collect data [84]. Inspectors need to develop the

447 skill to operate modern inspection devices. The advent of intelligent virtual assistants

448 (IVA) technology helps data collection from the human factors perspective. IVA is an

449 AI-powered agent that integrates machine learning, AR, VR, data science and other

450 technologies to perform tasks or services based on user commands or questions [85].

451 Li et al. [86] presented a VR-based training and assessment system to assist bridge

452 inspectors in controlling drones. This study demonstrated that IVA has the ability to

453 identify needs of individuals in detail and help them develop the skill in bridge

454 inspection. Table 6 lists the advantages and disadvantages of data acquisition

25
455 techniques. Furthermore, the development trend and the key performance indicators

456 (KPIs) targeted by technologies in data acquisition are summarised in Fig. 14. With

457 the improvement of hardware and software technologies, the quality and cost-

458 effectiveness of data acquisition is improved. The volume and integrity of maintenance

459 data have increased at the same time.

460 Table 6. Advantages and disadvantages of data acquisition technologies.
Technologies Advantages Disadvantages
All-around data;
Placement problems;
Technologies High quality;
Technical challenges;
of sensors Getting data in real-time;
Only for critical bridges;
Getting data in the long-term;
High accuracy;
High intelligence; Strict pre-commissioning operations
Technologies
Unmanned; of devices;
of NDTs
Reducing data processing time; Staff needing to be onsite;
Accessing to most blind spots;

461
462 Fig. 14. Trend and KPIs in data acquisition.

463 4.3 Data and Information Mining Technologies

464 In this section, the similarity between characteristics of bridge maintenance data

465 and big data is analysed (Fig. 15). Maintenance data have low-value density and time

466 variability and are in line with the 4V characteristics of big data, which are large in

467 volume, diverse in variety, frequently changing in velocity, and of great value but low in

468 value density [87.88]. Bridge maintenance data have the characteristics of big data.

26
469 The collection of a large amount of data can be turned into a vast base for knowledge

470 mining to go for knowledge-driven approaches. Thus, it is important to adopt big data

471 analysis methods to deeply leverage the raw data. What is converted from raw data is

472 meaningful data, which is also called information.

473
474 Fig. 15. Similarity analysis between characteristics of bridge maintenance data and big data.

475 To derive meaningful data from raw data, all kinds of methods are adopted to carry

476 out data preprocessing, data fusion, feature extraction, pattern recognition and other

477 processes step-by-step. Various data types are processed with various technologies,

478 e.g., electrical signals are converted to digital signals [89], traffic load data are

479 converted to structure strain and stress [90], and displacement and strain data are

480 fused together to gain more comprehensive results [91]. The analysis technology of

481 unstructured data is relatively complicated. A considerable amount of data under the

482 condition of the bridge and maintenance actions is buried in the textual bridge

483 inspection reports and not utilised [92]. Information extraction (IE) methods can

484 automatically recognise and extract information from unstructured textual bridge

485 inspection reports and represent them in a structured format. IE methods used in

486 papers can be classified into two primary categories [93-95]: rule-based methods and

487 machine learning ML-based methods. Rule-based methods use manually coded rules

488 for text processing. ML-based methods use ML algorithms for training text processing

27
489 models based on the text features of a given training text. However, compared to other

490 IE efforts, e.g., IE from social media text, automated IE from bridge reports is more

491 challenging because bridge inspection reports written by different organisations from

492 various locations are highly variable in terms of text characteristics and patterns. To

493 capture the variability in text patterns, IE methods require the development of a

494 comprehensive set of rules specifically for bridging the reports domain. This process

495 is time-consuming and requires a great amount of human effort. Therefore, a few

496 ontology-based IE algorithms have been studied [96,97]. For example, semantic

497 modelling and semantic natural language processing (NLP) techniques were used to

498 facilitate automated textual regulatory document analysis (e.g., code analysis).

499 Then, the processed data can be used to analyse structural conditions, including

500 current condition evaluation, failure probability computation, and life expectancy

501 prediction. For current condition evaluation, condition indexes [98] can be calculated

502 directly based on sensors or survey data to indicate if damages happen and the if

503 structure is out of service, e.g., discrete indexes are estimated by mapping detected

504 damages and abnormal responses to discrete values to evaluate the current condition

505 [11,99]. for the mechanism for failure probability computation includes two aspects:

506 deterioration severity computation and load computation. The former estimates the

507 probability that deterioration (e.g., loss of stiffness) exceeds the limit [100,101]. The

508 input data mainly come from the survey data, including the bridge profile, damage data

509 (e.g., the size of crack), and environmental data. The latter estimates the probability

510 that the load exceeds the design capacity [102,103]. The input data mainly come from

511 the sensor data, including the bridge profile, traffic data, structure responses (e.g.,

512 vibration and displacement) and environment data. Then, life expectancy can be

513 predicted as the time that the failure probability and a condition index decline below a

514 threshold [104-106].

515 Whether assessing the load rating or reliability of deterioration during the service

516 life of structures, time variables are always the first issue to be considered in many

28
517 studies, e.g., the prediction model deteriorating in time due to corrosion and live load

518 increase [107], the lifetime performance indicators for the deteriorating structures [108].

519 The main analytical methods are “model-based” and “data-driven” [22]. The model-

520 based method is essentially a process of bridge structure finite element modelling,

521 model modification, and system parameter inversion. It has high requirements for the

522 accuracy of the theoretical model and the quality of the data. The data-driven method

523 identifies the changing pattern of the structural state by studying the changing trends

524 and probability distributions of the data itself. It is widely used in structural health

525 monitoring. However, only a small part of the maintenance data is used for the analysis

526 process, and its performance improvement is often limited when used on large

527 datasets [22]. In fact, there are many studies on the combined use of the above two

528 methods, e.g., developing stochastic deterioration models for bridge elements [109-

529 112]. Stochastic models capture the uncertainty and randomness of the facility

530 deterioration process as one or more random variables. Stochastic approaches are

531 more in line with the degraded state of the bridge in the real environment, e.g., physics-

532 based stochastic models, Markov chains or Weibull distribution models [113,116]. With

533 the development of artificial intelligence (AI) technology, many research efforts have

534 developed different AI models to better predict and understand bridge deterioration.

535 For example, an artificial neural network (ANN) model is used to develop an application

536 model for estimating the future condition of bridges [117]. Artificial neural networks

537 (ANNs) and k-nearest neighbours (KNNs) are used to build two computational machine

538 learning models to predict deck conditions [118].

539 At the same time, the digital twin (DT) concept proposed by Michael Grieves [119]

540 has been gradually introduced into the bridge maintenance field with the development

541 of intelligent technology. The DT concept has shown pivotal potential in security

542 prewarning [120]. DT can make full use of data (such as physical models, sensor data

543 (real-time data), operating history (real data), and related derived data generated

544 through mining to integrate multidisciplinary, multiphysical, multiscale, and

29
545 multiprobability simulation processes. Comparing the application of the BIM and DT

546 models in bridge maintenance work [6], the DT model pays more attention to how to

547 capture and store the historical data of the bridge and, based on that, to predict the

548 future behaviour of the bridge. Data, such as accumulated damage history and repair

549 history can be directly exported, which provides important support for project

550 maintenance teams and decision-making agents to respond appropriately in time when

551 the bridge fluctuates suddenly [121].

552 Table 7 lists the advantages and disadvantages of analytical methods. These

553 methods can be collectively referred to as data and information mining technologies,

554 which are the combination and improvement of methods represented by various terms.

555 Specifically, these terms encompass various algorithms, such as classification,

556 clustering, association analysis, and regression. However, stressing the use of a

557 certain method alone is not enough to determine the success of big data analysis from

558 knowledge mining to turn to knowledge-driven approaches. The final analysis result is

559 often the intersection of the effects that can be achieved by each link in the process.
560 Table 7. Advantages and disadvantages of analytical methods
Analytical Methods Advantages Disadvantages

IE methods for text Automatically reorganization Time-consuming;

files and extraction information; An amount of human effort;
The model-based
Easy for engineers; High requirements for models
method for
Well-established methodology; and the data;
evaluation
The data-driven Mining hidden patterns in data;
method for Widely used in structural health Only using a small part of data;
evaluation monitoring;
Stochastic High accuracy;
Difficult to simulate random
approaches for In line with the degradation state
variables;
evaluation of the real bridge;
Making full use of data;
Digital Twin Integration multi-disciplinary,
technology for multi-physical, multi-scale, and Unclear methodology in bridge
evaluation multi-probability simulation maintenance;
processes;

30
561 4.4 Holistic decision-making approaches

562 Maintenance decision-making involves multifactor (e.g., structure safety, costing

563 issues, sustainability) and multidomain knowledge (e.g., structure engineering,

564 material perspectives, environmental perspectives). It is difficult to describe the

565 relationship between variations of factors and the decision-making objective with

566 precise mathematical and mechanical methods. Knowledge-based and artificial

567 intelligence-supported reasoning, which is characterised as a knowledge-based

568 approach, has received increasing attention. Prioritisation indices (PIs) are a need-

569 based bridge maintenance approach that allows short-term maintenance decisions to

570 be made [122]. A PI of Valenzuela et al. [123] considers the structural condition,

571 hydraulic vulnerability and seismic risk, in addition to the importance of the bridge

572 within the road network and the productive system. Echaveguren et al. [124] proposed

573 a systematic method for maintenance decisions and their associated costs by using a

574 PI based on bridge conditions, strategic importance and vulnerability. Additionally,

575 decision-making can be treated as a special case of reinforcement learning (RL) by

576 using a family of efficient sampling algorithms, such as the bootstrapping TD (temporal

577 difference) method, the Monte-Carlo tree search (MCT) method, the deep neural

578 network (DNN) method, and the convolutional neural network (CNN) method [125].

579 Recently, an important trend has been using semantic web technology (SWT) to

580 address issues related to knowledge representation and decision-making [126-128].

581 For example, the bridge hazard knowledge base was created to realise hazard

582 classification management. A specific bridge ontology was developed to solve the

583 intelligent retrieval of massive data and knowledge reasoning that is difficult to achieve

584 with traditional approaches [18].

585 Bridge maintenance supported by knowledge-driven approaches also needs to

586 define the data needed for different decision-making scenarios and how to use

587 technologies to complete the retrieval and service in a smart way [1]. Currently, most

588 data needs are defined in bridge inspection manuals. For instance, primary inspection

31
589 only requires structure photos, general evaluations and a description of damages, and

590 detailed inspection can require accurate measurement of components’ structural

591 performance (e.g., concrete strength) and surface and subsurface damages. A few

592 studies also attempt to define data needs for operation and maintenance applications,

593 e.g., defining the data needs for bridge life estimation [129] and investigating

594 information requirements of stakeholders (e.g., the owner and maintenance teams)

595 [130]. However, comprehensive data requirements have not yet been defined.

596 Although not defined, online databases and cloud servers reduce hardware

597 dependency and provide the opportunity for accessing unified and up-to-date models,

598 as well as their associated data that could be easily accessed through mobile devices.

599 Hadoop is a software framework that supports distributed applications. It is used to

600 build a distributed platform and use other program groups to perform specific functions,

601 such as storage and calculation, which is the “Hadoop ecosystem”, including

602 MapReduce, Spark, cloud computing, etc [22,131]. Cloud computing is another

603 research hotspot. Its architecture can respond to the needs of heterogeneous big data

604 storage in maintenance projects and efficient information sharing and transmission

605 across participants, disciplines, and project stages. Bridge managers assess data

606 during the bridge operation phase from the cloud platform in real-time and understand

607 the health conditions of bridges [132,133]. At the same time, facing the real-time

608 computing requirements of maintenance data, the powerful computing power of cloud

609 computing makes it possible to execute complex algorithms online [13].

610 Database technology holds considerable potential in the bridge maintenance field.

611 The use of a database allows valuable information to be captured, stored, sorted, and

612 extracted according to a predetermined set of selection criteria [134]. Relational

613 database (RDB) systems and NoSQL (Not-Only SQL) database systems are often

614 employed as the primary data storage for bridge maintenance applications [135,136].

615 One of the most cutting-edge database technologies is to build an integrated

616 information platform for bridge construction, management and maintenance based on

32
617 building information modelling (BIM). In bridge maintenance, BIM research has been

618 an additional research hotspot for a long time. Its main objectives are twofold: (1)

619 Enabling an integrated bridge database [137-139]. A wide array of information about

620 the bridge, including the 3D geometry, project management information, such as time

621 schedules and costs or operation and maintenance metrics, are stored in a central,

622 object-oriented database. (2) Developing electronic data exchange standards to

623 facilitate information sharing and collaboration [140-143], such as how the incoming

624 and outgoing information is handled and how project participants build, use and

625 manage this information. The application of BIM further promotes the openness,

626 sharing and multiparty collaboration of data. However, its centralised paradigm is

627 degraded by the risk of data manipulation [144].

628 Blockchain technology and InterPlanetary file system (IPFS) are emerging

629 solutions to prevent the problems caused by centralisation. Blockchain is a type of

630 distributed ledger technology (DLT) that uses a decentralised architecture based on

631 distributed computing, crypto-chain block structures to store data, node consensus

632 algorithms to verify data and smart contracts to program data [145]. A recent research

633 trend has presented the feasibility of integrating blockchain with BIM, involving

634 methods of blockchain-BIM integration [146-148] and methods of BIM data storage in

635 the blockchain [149,150]. The IPFS, which is a peer-to-peer network, is regarded as

636 an appropriate technical complement to blockchain for storing large files [151]. Tao et

637 al. [152] presented a framework for secure BIM design collaboration in which an IPFS

638 network is responsible for storing large design files (e.g., BIM models), while a

639 blockchain network is leveraged to keep and exchange design information (e.g., design

640 changes). Similar to other emerging technologies in their first years, research on

641 blockchain and IPFS related to bridge engineering is still new and fragmented, but in

642 the future, it may have the potential to play a critical role [153,154]. Finally, the

643 development of technologies, such as Web3D, VR and mixed reality (MR) has greatly

644 promoted the process of information visualisation. Three-dimensional models are

33
645 displayed to clients using these techniques. More importantly, the maintenance data

646 (e.g., monitoring data, inspection maintenance data, construction management data)

647 are visually displayed and interact with a three-dimensional model as the carrier [155].

648 The development trend and KPIs in smart bridge maintenance are summarised in

649 Fig. 16. Bridge management workers are changing their mindset. They consider the

650 heterogeneity of massive data, attach importance to the implicit relationship between

651 data, and promote the improvement of data analysis results. In addition, they pay more

652 attention to the real-time and powerful analysis and unified storage of dynamic

653 incomplete big data/information, as well as needs for multi-source information sharing

654 and transmission across participants, disciplines, and project phases. The critical

655 review hence reveals the trends of moving from data to knowledge-driven smart bridge

656 maintenance, and a proactive, holistic and smart lifecycle approach is needed to

657 comprehend the complexity of bridge structural conditions.

658
659 Fig. 16. Trend and KPIs in smart bridge maintenance.

660 5. A framework facilitating knowledge-driven smart bridge maintenance

661 Knowledge-driven bridge maintenance supported by big data mining ideally needs

662 to address several challenges. Table 8 lists a broader range of challenges identified

663 by many researchers to further classify them into two different categories: (1) technical

664 challenges in the data-driven manner and (2) challenges in the transition towards

665 knowledge-driven approaches.

666 Table 8. Challenges identified by previous researchers.

34
 Challenges in the data-driven manner

 Lack of higher performance hardware to handle large volumes of data;

 Lack of data integration;
 Limitation and isolation of distinct types of data sources;
 Lack of a standard neutral exchange format and schema;
 Poor interoperability of heterogeneous software and platforms;
 Poor interoperability and information sharing among software and technology;
 Poorly developed theory and method for structural state evaluation;
 Unclear method to create accuracy analytical model;
 Poorly developed MR&R technologies;

Challenges in the transition towards knowledge-driven approaches

 Lack of the definitions of data flow requirements;

 The differences and lack of fully adopted work flow;
 Lack of standard procedures;
 Lack of standard data needs;
 Inadequacy of representation of semantics and geometrics for data models;
 Poor collaboration between academia and industry;

667

668 (1) Technical challenges in the data-driven manner

669 Many technology-related challenges prevent data-driven approaches from

670 meeting the growing demands of smart bridge maintenance. A major challenge is the

671 lack of unified data formats for bridge maintenance covering the entire lifecycle data

672 exchange and across different sectors. As the data formats cannot be unified, the big

673 data collected cannot be effectively exchanged and shared in different sectors. Thus,

674 only a small portion of those have been utilised successfully for maintenance decision-

675 making. Currently, there are various examples of integrating different types of datasets

676 and data formats, e.g., expanding the industry foundation classes (IFC) architecture.

677 However, IFC provides a rich, redundant yet ambiguous schema for interoperability of

678 heterogeneous software and platforms, leading to the lack of semantic clarity in

679 mapping entities and relationships. Moreover, the standards for the component

680 classification in the bridge operation and maintenance phase are different from those

681 in the design and construction phases, which means that engineers must do a great

682 amount of work to fully expand the common data standards throughout the lifecycle of

683 bridges. Thus, no unified data formats have been fully extended to encompass the

35
684 major types of bridge maintenance projects.

685 The technology-related challenges may be mitigated by improvements to the

686 technology over time, but data-driven approaches are still not sufficient to solve tasks

687 of bridge maintenance decision-making. Bridge maintenance is a complex task that

688 requires the cooperation of different stakeholders. The coordination of work among

689 different teams and organisations in this type of task is important, while is a

690 complicated process. Hence, it requires more holistic and smart lifecycle approaches.

691 (2) Challenges in the transition towards knowledge-driven approaches

692 An ideal smart maintenance system can determine data/information requirements

693 and identify by whom and when the data/information should be provided throughout

694 the project lifecycle according to different decision-making scenarios. During this

695 process, data exchange, as one of the important requirements, should be exhaustive

696 in representation of semantics, as well as geometrics. Currently, there are several

697 studies integrating blockchain and IPFS technologies with BIM software to address

698 challenges, such as interoperability and information sharing among software and

699 technology and the definitions of data flow requirements. However, the focus is only

700 on the model level. These BIM systems lack efficient semantic query and reasoning

701 capabilities. There is still a lack of such a holistic and comprehensive semantic-level

702 knowledge system that can provide enough semantic interoperability and

703 representation of the knowledge. This may be because numerous concepts and their

704 logical relationships defined in maintenance standards require engineers to perform

705 manual extraction rather than directly be recognised by computer programs to form a

706 complete knowledge system. It requires manual labour and a high-quality collaboration

707 of experts among different teams and organisations. In addition, it is difficult to reuse

708 or expand existing knowledge bases. These knowledge bases that are built for the

709 same purpose may have different terms and structures. The problem of collaboration

710 between knowledge bases established for different purposes is also difficult to solve.

711 The above challenges hinder the transition towards knowledge-driven approaches.

36
712 Fig. 17 shows a proposed framework, which is knowledge driven and targets the

713 development of suitable knowledge networking mechanisms to drive numerous tools.

714 Specifically, this roadmap uses semantic web technology, BIM, and IoTs to integrate

715 maintenance data with embedded big data methods support to enable smart reasoning

716 and holistic maintenance decision-making. The framework includes three key

717 components: 1) A dynamic semantic knowledge base. In a knowledge-driven manner,

718 a dynamic semantic knowledge base is used for intelligent semantic recognition, data

719 and information integration, numerical-based and logical-based reasoning, and holistic

720 decision-making. 2) A database. In a data-driven manner, a database is used for real-

721 time data/information mining with high-performance computing power. 3) Data

722 acquisition system. Big data are collected in a large volume and comprehensively

723 throughout the whole bridge lifecycle. The collaboration of these three crucial

724 components allows the whole framework to work seamlessly and effectively.

37
 Maintenance solutions Maintenance demands
Uers

In a knowledge-driven manner, a dynamic semantic knowledge base is used for intelligent semantic recognition, data and information
integration, numerical-based and logical-based reasoning, holistic decision-making.

Numerical-based Analysis Logical-based Reasoning Process

Holistic Decision-making Process
Process Finite Element Rules Semantic
Decision Criteria
Checking analysis
Query Interface Taxonomy Triples
Analytical analysis

Update

Semantic Models
OWL to IFC
BIM Information Integration Platform with a unified format …
Web Ontology
Industry Foundation Language (OWL) Model 2
Digital models
Classes (IFC) architecture IFC to OWL API Model 1

Input the Required Useful Data/Information Output Data/Information Requirements

In a data-driven manner, a database is used for real-time data/information mining with high-performance computing power.

Big data analysis methods

Big data processing technologies
Machine Learning
Big data storage technologies Hadoop
Information Extraction
RDB MapReduce
Natural Language Processing
NoSQL Spark
Artificial Neural Network
… Cloud computing
Digital Twin
...
…

Data Uploaded to the Database

Raw Data Downloaded from NDTs Raw Data Downloaded from Sensors

Data capture by NDTs Data capture by Sensors Second-hand

Sources

Bridge profile: Properties of Environment data: Temperature,

Wind, Humidity, Topography and Maintenance
components, e.g. geometry, Point clouds, Photo images:
geology, Hydrology. costs
number, material properties, and Bridge structure features: 1)both
connection of components. local features, e.g. girder cracks,
and global features, e.g. holistic Traffic data: Traffic volume, Vehicle
bridge geometry. 2)only reflect load ,Vehicle speed. Social data
Acoustics and ultrasonic, surface conditions.
Radar，Infrared images: local Structure responses: Strain and
features which reflect both stress, Displacement and deflection,
surface and subsurface Environment
Electrical and chemical: Cracks and deformation, Acceleration,
conditions. data
Structural corrosion. Vibration and frequency.

725
726 Fig. 17. A knowledge framework to implement smart bridge maintenance.

727 The workflow is provided as follows: First, bridge maintenance personnel input

728 their needs. Semantic models of the knowledge base match the corresponding

729 maintenance scenarios that define the required data/information and their details.

730 These details are passed into the BIM platform through the translation between the

731 semantic web standard language (Web Ontology Language) and the industry

732 foundation classes architecture. Then, the BIM information integration platform with a

38
733 unified data format drives the database to call the data collected by the corresponding

734 source in a targeted manner. The process can be called and released in real-time

735 (shown in blue lines). Furthermore, big data storage and processing technologies and

736 big data analysis methods are jointly used to obtain the required useful

737 data/information from raw data in real-time online analysis mode. They are unified and

738 coordinated by the BIM platform. Some of them are used to build finite elements or

739 mathematical models for numerical-based analysis. Some are transformed into

740 semantic models for logic-based reasoning. The results of numerical-based analysis

741 can be embedded in logic-based reasoning to support holistic decision-making. The

742 correctness of the results is determined according to certain criteria. The results that

743 meet the requirements are fed back to engineers at the query interface to assist them

744 in making maintenance decisions, which are further updated into the knowledge base

745 as facts (shown in red lines).

746 The framework is an open, computable, and evolvable knowledge network based

747 on maintaining big data. Openness means that the sources of the data are diverse. Big

748 data comes from massive, heterogeneous and autonomous sources. Computability

749 means that the knowledge network can use various methods to explore complex and

750 evolving relationships between maintenance data, and it can perform reasoning

751 calculations on knowledge itself. Evolvability means that the network can continuously

752 infer the latest knowledge and update itself. At the same time, knowledge in other

753 networks can be transformed into a standard form and absorbed into the network.
754

755 6 Conclusion

756 This paper presents a critical review and comprehensive literature analysis to

757 investigate state-of-the-art methods used in smart bridge maintenance, which reveals

758 the need for a knowledge-driven approach supported by large survey/monitoring data

759 mining. First, 2,732 papers collected from the WoS core collection database are

760 visually analysed using the CiteSpace software, including four perspectives: literature

39
761 quantity analysis, journal co-citation analysis, document co-citation analysis, and

762 keywords clustering and burst analysis.. Second, the result of visualisation analysis

763 helps to pinpoint 323 papers for further critical review, focusing on three areas: bridge

764 maintenance tasks and issues, advanced technologies supporting smart maintenance,

765 and holistic decision-making approaches. The analysis informs that bridge engineers

766 need to change their mindset from traditional experience oriented to holistically

767 consider the heterogeneity of maintenance big data, to understand the implicit

768 relationship and knowledge among different data and information streams. Based on

769 the concluded technical challenges in the data-driven manner and challenges in the

770 transition towards knowledge-driven approaches, this paper proposes a novel

771 framework and methodology in the end with an aim to leverage the underused large

772 amounts of bridge maintenance big data by using knowledge-driven approaches,

773 including three key components: smart raw data acquisition, data and information

774 unification through BIM, and dynamic ontological knowledge processing, to facilitate

775 future developments towards smart bridge maintenance.

776 Acknowledgement

777 This research was supported by Fundamental Research Funds for the Central

778 Universities of China (grant number 3132019349), China Scholarship Council (CSC

779 No. 202006570025), and BIM for Smart Engineering Centre in Cardiff University, UK.

780 The author would like to thank them for their supports.

781 Declaration of competing interest

782 The authors declare that they have no known competing financial interests or personal

783 relationships that could have appeared to influence the work reported in this paper.

784 References
785 [1] C. Wu, P. Wu, J. Wang, R. Jiang, M. Chen, X. Wang, Critical review of data-driven decision-making
786 in bridge operation and maintenance, Struct. Infrastruct. E. (2020) 1-24,
787 https://doi.org/10.1080/15732479.2020.1833946.
788 [2] X. Zhou, X. Zhang, Thoughts on the Development of Bridge Technology in China, Engineering 5 (6)

40
789 (2019) 1120-1130, https://doi.org/10.1016/j.eng.2019.10.001.
790 [3] Y. Chen, X. Xue, Advances in the structural health monitoring of bridges using piezoelectric
791 transducers, Sensors 18 (12) (2018) 4312, https://doi.org/10.3390/s18124312.
792 [4] Z. Liu, T. Guo, X. Yu, X. Huang, J. Correia, Corrosion fatigue and electrochemical behaviour of steel
793 wires used in bridge cables, Fatigue Fract. Eng. M. (2020), https://doi.org/10.1111/ffe.13331.
794 [5] C. Shim, N. Dang, S. Lon, C. Jeon, Development of a bridge maintenance system for prestressed
795 concrete bridges using 3D digital twin model, Struct. Infrastruct. E. 15 (10) (2019) 1319-1332,
796 https://doi.org/10.1080/15732479.2019.1620789.
797 [6] W. Locke, J. Sybrandt, L. Redmond, I. Safro, S. Atamturktur, Using drive-by health monitoring to
798 detect bridge damage considering environmental and operational effects, J. Sound Vib. 468 (2020),
799 https://doi.org/10.1016/j.jsv.2019.115088.
800 [7] Important traffic news from the Ministry of Transport of the People's Republic of China, [Online].
801 Available https://www.mot.gov.cn/jiaotongyaowen/202111/t20211111_3625593.html
802 [8] LGA responds to RAC Foundation press release on substandard bridges, [Online]. Available
803 https://www.local.gov.uk/about/news/lga-responds-rac-foundation-press-release-substandard-
804 bridges
805 [9] Local bridge collapses up 70%, [Online]. Available https://www.localgov.co.uk/Local-bridge-
806 collapses-up-70/53924
807 [10] T. Hong, M. Hastak, Evaluation and determination of optimal MR&R strategies in concrete bridge
808 decks, Automat. Constr. 16 (2) (2007) 165-175, https://doi.org/10.1016/j.autcon.2006.03.002.
809 [11] S. Abu Dabous, S. Yaghi, S. Alkass, O. Moselhi, Concrete bridge deck condition assessment using
810 IR Thermography and Ground Penetrating Radar technologies, Automat. Constr. 81 (2017) 340-
811 354, https://doi.org/10.1016/j.autcon.2017.04.006.
812 [12] C. Yun, S. Cho, H. Park, J. Min, J. Park, Smart wireless sensing and assessment for civil
813 infrastructure, Struct. Infrastruct. E. 10 (4) (2014) 534-550,
814 https://doi.org/10.1080/15732479.2013.769011.
815 [13] X. Tong, H. Yang, L. Wang, Y. Miao, The development and field evaluation of an IoT system of Low-
816 Power vibration for bridge health monitoring, Sensors 19 (5) (2019) 1222,
817 https://doi.org/10.3390/s19051222.
818 [14] A. Costin, A. Adibfar, H. Hu, S.S. Chen, Building Information Modeling (BIM) for transportation
819 infrastructure - Literature review, applications, challenges, and recommendations, Automat. Constr.
820 94 (2018) 257-281, https://doi.org/10.1016/j.autcon.2018.07.001.
821 [15] C.V. Dung, H. Sekiya, S. Hirano, T. Okatani, C. Miki, A vision-based method for crack detection in
822 gusset plate welded joints of steel bridges using deep convolutional neural networks, Automat.
823 Constr. 102 (2019) 217-229, https://doi.org/10.1016/j.autcon.2019.02.013.
824 [16] H. Kim, E. Ahn, M. Shin, S. Sim, Crack and noncrack classification from concrete surface images
825 using machine learning, Structural Health Monitoring 18 (3) (2018) 725-738,
826 https://doi.org/10.1177/1475921718768747.
827 [17] S. Krohn, J. Tromp, E.M. Quinque, J. Belger, F. Klotzsche, S. Rekers, P. Chojecki, J. de Mooij, M.
828 Akbal, C. Mccall, A. Villringer, M. Gaebler, C. Finke, A. Thone-Otto, Multidimensional evaluation of
829 virtual reality paradigms in clinical neuropsychology: Application of the VR-Check framework, J.
830 Med. Internet Res. 22 (4) (2020), https://doi.org/10.2196/16724.
831 [18] G. Ren, R. Ding, H. Li, Building an ontological knowledgebase for bridge maintenance, Adv. Eng.

41
832 Softw. 130 (2019) 24-40, https://doi.org/10.1016/j.advengsoft.2019.02.001.
833 [19] S. Abu Dabous, S. Feroz, Condition monitoring of bridges with non-contact testing technologies,
834 Automat. Constr. 116 (2020) 103224, https://doi.org/10.1016/j.autcon.2020.103224.
835 [20] S. Agnisarman, S. Lopes, K. Chalil Madathil, K. Piratla, A. Gramopadhye, A survey of automation-
836 enabled human-in-the-loop systems for infrastructure visual inspection, Automat. Constr. 97 (2019)
837 52-76, https://doi.org/10.1016/j.autcon.2018.10.019.
838 [21] Y. Fujino, D.M. Siringoringo, Y. Ikeda, T. Nagayama, T. Mizutani, Research and implementations of
839 structural monitoring for bridges and buildings in japan, Engineering 5 (6) (2019) 1093-1119,
840 https://doi.org/10.1016/j.eng.2019.09.006.
841 [22] L. Sun, Z. Shang, Y. Xia, Development and Prospect of Bridge Structural Health Monitoring in the
842 Context of Big Data. China Journal of Highway and Transport 32 (11) (2019) 1-20.
843 https://doi.org/10.19721/j.cnki.1001-7372.2019.11.001.
844 [23] S. Banerjee, B.S. Vishwanath, D.K. Devendiran, Multihazard resilience of highway bridges and
845 bridge networks: A review, Struct. Infrastruct. E. 15 (12) (2019) 1694-1714,
846 https://doi.org/10.1080/15732479.2019.1648526.
847 [24] G. Kabir, R. Sadiq, S. Tesfamariam, A review of multi-criteria decision-making methods for
848 infrastructure management, Struct. Infrastruct. E. 10 (9) (2014) 1176-1210,
849 https://doi.org/10.1080/15732479.2013.795978.
850 [25] Dan M. Frangopol, D. Saydam, S. Kim, Maintenance, management, life-cycle design and
851 performance of structures and infrastructures: a brief review, Struct. Infrastruct. E. 8(1) (2012) 1-
852 25, https://doi.org/10.1080/15732479.2011.628962.
853 [26] A. Bradley, H. Li, R. Lark, S. Dunn, BIM for infrastructure: An overall review and constructor
854 perspective, Automat. Constr. 71 (2016) 139-152, https://doi.org/10.1016/j.autcon.2016.08.019.
855 [27] R. Vanderstraeten, F. Vandermoere, Inequalities in the growth of Web of Science, Scientometrics
856 126 (10) (2021) 8635-8651, https://doi.org/10.1007/s11192-021-04143-2.
857 [28] Z. Irani, M.M. Kamal, Intelligent systems research in the construction industry, Expert Syst. Appl.
858 41 (4) (2014) 934-950, https://doi.org/10.1016/j.eswa.2013.06.061.
859 [29] J. Song, H. Zhang, W. Dong, A review of emerging trends in global ppp research: analysis and
860 visualization, Scientometrics 107 (3) (2016) 1111-1147, https://doi.org/10.1007/s11192-016-1918-1.
861 [30] J. Han, H. Kang, M. Kim, G.H. Kwon, Mapping the intellectual structure of research on surgery with
862 mixed reality: bibliometric network analysis (2000–2019), J. Biomed. Inform. 109 (2020) 103516,
863 https://doi.org/10.1016/j.jbi.2020.103516.
864 [31] C. Chen, Citespace ii: detecting and visualizing emerging trends and transient patterns in scientific
865 literature, Journal of the American Society for Information Science and Technology 57 (3) (2006)
866 359-377, https://doi.org/10.1002/asi.20317.
867 [32] C. Chen, F. Ibekwe-Sanjuan, J. Hou, The structure and dynamics of cocitation clusters: a multiple-
868 perspective cocitation analysis, Journal of the American Society for Information Science and
869 Technology 61 (7) (2010) 1386-1409, https://doi.org/10.1002/asi.21309.
870 [33] C. Chen, S. Morris, Visualizing evolving networks: Minimum spanning trees versus pathfinder
871 networks, IEEE Symposium on Information Visualization 2003, IEEE, Seattle, WA, (2003) 67–74,
872 https://doi.org/10.1109/INFVIS.2003.1249010
873 [34] X. Zhao, A scientometric review of global bim research: analysis and visualization, Automat. Constr.
874 80 (2017) 37-47, https://doi.org/10.1016/j.autcon.2017.04.002.

42
875 [35] D.M. Frangopol, Life-cycle performance, management, and optimisation of structural systems
876 under uncertainty: accomplishments and challenges 1, Struct. Infrastruct. E. 7 (6) (2011) 389-413,
877 https://doi.org/10.1080/15732471003594427.
878 [36] D.M. Frangopol, Y. Dong, S. Sabatino, Bridge life-cycle performance and cost: analysis, prediction,
879 optimisation and decision-making, Struct. Infrastruct. E. 13 (10) (2017) 1239-1257,
880 https://doi.org/10.1080/15732479.2016.1267772.
881 [37] D.M. Frangopol, M. Liu, Maintenance and management of civil infrastructure based on condition,
882 safety, optimization, and life-cycle cost, Struct. Infrastruct. E. 3 (1) (2007) 29-41,
883 https://doi.org/10.1080/15732470500253164.
884 [38] D.M. Frangopol, J.S. Kong, E.S. Gharaibeh, Reliability-based life-cycle management of highway
885 bridges, J. Comput. Civil Eng. 15 (1) (2001) 27-34, https://doi.org/10.1061/(ASCE)0887-
886 3801(2001)15:1(27).
887 [39] D.M. Frangopol, M. Soliman, Life-cycle of structural systems: recent achievements and future
888 directions, Struct. Infrastruct. E. 12 (1) (2016) 1-20, https://doi.org/10.1080/15732479.2014.999794.
889 [40] R.E. Melchers, Progress in developing realistic corrosion models, Struct. Infrastruct. E. 14 (7) (2018)
890 843-853, https://doi.org/10.1080/15732479.2018.1436570.
891 [41] N.M. Okasha, D.M. Frangopol, Lifetime-oriented multi-objective optimization of structural
892 maintenance considering system reliability, redundancy and life-cycle cost using ga, Struct. Saf. 31
893 (6) (2009) 460-474, https://doi.org/10.1016/j.strusafe.2009.06.005.
894 [42] F. Biondini, D.M. Frangopol, Life-cycle performance of deteriorating structural systems under
895 uncertainty: review, J. Struct. Eng. 142 (9) (2016), https://doi.org/10.1061/(ASCE)ST.1943-
896 541X.0001544.
897 [43] S. Sabatino, D.M. Frangopol, Y. Dong, Sustainability-informed maintenance optimization of
898 highway bridges considering multi-attribute utility and risk attitude, Eng. Struct. 102 (2015) 310-321,
899 https://doi.org/10.1016/j.engstruct.2015.07.030.
900 [44] S. Kim, D.M. Frangopol, M. Soliman, Generalized probabilistic framework for optimum inspection
901 and maintenance planning, J. Struct. Eng. 139 (3) (2013) 435-447,
902 https://doi.org/10.1061/(ASCE)ST.1943-541X.0000676.
903 [45] P. Bocchini, D.M. Frangopol, Generalized bridge network performance analysis with correlation and
904 time-variant reliability, Struct. Saf. 33 (2) (2011) 155-164,
905 https://doi.org/10.1016/j.strusafe.2011.02.002.
906 [46] P. Bocchini, D.M. Frangopol, A probabilistic computational framework for bridge network optimal
907 maintenance scheduling, Reliab. Eng. Syst. Safe. 96 (2) (2011) 332-349,
908 https://doi.org/10.1016/j.ress.2010.09.001.
909 [47] H. Furuta, T. Kameda, K. Nakahara, Y. Takahashi, D.M. Frangopol, Optimal bridge maintenance
910 planning using improved multi-objective genetic algorithm, Struct. Infrastruct. E. 2 (1) (2006) 33-41,
911 https://doi.org/10.1080/15732470500031040.
912 [48] J.M. van Noortwijk, D.M. Frangopol, Two probabilistic life-cycle maintenance models for
913 deteriorating civil infrastructures, Probabilist. Eng. Mech. 19 (4) (2004) 345-359,
914 https://doi.org/10.1016/j.probengmech.2004.03.002.
915 [49] M Liu, D.M. Frangopol, Probability-based bridge network performance evaluation, J. Bridge Eng.
916 11(5) (2006) 633-641, https://doi.org/10.1061/(ASCE)1084-0702(2006)11:5(633).
917 [50] P.J. Rousseeuw, Silhouettes: a graphical aid to the interpretation and validation of cluster analysis,

43
918 J. Comput. Appl. Math. 20 (1987) 53–65, https://doi.org/10.1016/0377-0427(87)90125-7.
919 [51] J. Kleinberg, Bursty and Hierarchical Structure in Streams. Data Mining and Knowledge Discovery
920 7 (2003) 373–397, https://doi.org/10.1023/A:1024940629314.
921 [52] B.M. Phares, G.A. Washer, D.D. Rolander, B.A. Graybeal, M. Moore, Routine highway bridge
922 inspection condition documentation accuracy and reliability, J. Bridge Eng. 9 (4) (2004) 403-413,
923 https://doi.org/10.1061/(ASCE)1084-0702(2004)9:4(403).
924 [53] D. Zymelka, K. Togashi, T. Kobayashi, Concentric array of printed strain sensors for structural health
925 monitoring, Sensors-Basel 20 (7) (2020) 1997, https://doi.org/10.3390/s20071997.
926 [54] H. Azari, A. Ghorbanpoor, S. Shams, Development of robotic nondestructive testing of steel
927 corrosion of prestressed concrete bridge girders using magnetic flux leakage system,
928 Transportation Research Record: Journal of the Transportation Research Board 2674 (8) (2020)
929 466-476, https://doi.org/10.1177/0361198120925471.
930 [55] S. Hsieh, H. Wang, C. Hu, C. Wang, Ultrasonic velocity tomography for inspecting the condition of
931 a bridge pylon, Insight 62 (4) (2020) 192-198, https://doi.org/10.1784/insi.2020.62.4.192.
932 [56] S. Tang, C. Ramseyer, P. Samant, L. Xiang, X-ray-induced acoustic computed tomography of
933 concrete infrastructure, Appl. Phys. Lett. 112 (2018), https://doi.org/10.1063/1.5009936.
934 [57] G. Washer, N. Bolleni, R. Fenwick, Thermographic imaging of subsurface deterioration in concrete
935 bridges, Transportation Research Record: Journal of the Transportation Research Board 2201 (1)
936 (2010) 27-33, https://doi.org/10.3141/2201-04.
937 [58] D. Erdenebat, D. Waldmann, Application of the DAD method for damage localisation on an existing
938 bridge structure using close -range UAV photogrammetry, Eng. Struct. 218 (2020),
939 https://doi.org/10.1016/j.engstruct.2020.110727.
940 [59] P. Liu, A.Y. Chen, Y. Huang, J. Han, J. Lai, S. Kang, T. Wu, M. Wen, M. Tsai, A review of rotorcraft
941 Unmanned Aerial Vehicle (UAV) developments and applications in civil engineering, Smart Struct.
942 Syst. 13 (6) (2014) 1065-1094, https://doi.org/10.12989/sss.2014.13.6.1065.
943 [60] T. Omar, M.L. Nehdi, Remote sensing of concrete bridge decks using unmanned aerial vehicle
944 infrared thermography, Automat. Constr. 83 (2017) 360-371,
945 https://doi.org/10.1016/j.autcon.2017.06.024.
946 [61] F. Xu, Q. Jiang, Dynamic obstacle-surmounting analysis of a bilateral-wheeled cable-climbing robot
947 for cable-stayed bridges, Industrial Robot: the international journal of robotics research and
948 application 46 (3) (2019) 431-443, https://doi.org/10.1108/IR-07-2018-0152.
949 [62] H. Yun, S. Kim, L. Wu, J. Lee, Development of inspection robots for bridge cables, The Scientific
950 World Journal 2013 (2013) 1-17, https://doi.org/10.1155/2013/967508.
951 [63] S. Phillips, S. Narasimhan, Automating data collection for robotic bridge inspections, J. Bridge Eng.
952 24 (2019), https://doi.org/10.1061/(ASCE)BE.1943-5592.0001442.
953 [64] Y.N. Yang, M.M. Kumaraswamy, H.J. Pam, G. Mahesh, Integrated qualitative and quantitative
954 methodology to assess validity and credibility of models for bridge maintenance management
955 system development, J. Manage. Eng. 27 (3) (2011) 149-158,
956 https://doi.org/10.1061/(ASCE)ME.1943-5479.0000051.
957 [65] Y. An, E. Chatzi, S. Sim, S. Laflamme, B. Blachowski, J. Ou, Recent progress and future trends on
958 damage identification methods for bridge structures, Struct. Control Hlth. 26 (2019),
959 https://doi.org/10.1002/stc.2416.
960 [66] Y.B. Yang, J.P. Yang, State-of-the-Art review on modal identification and damage detection of

44
 961 bridges by moving test vehicles, Int. J. Struct. Stab. Dy. 18 (2018),
962 https://doi.org/10.1142/S0219455418500256.
963 [67] S. Li, S. Wei, Y. Bao, H. Li, Condition assessment of cables by pattern recognition of vehicle-
964 induced cable tension ratio, Eng. Struct. 155 (2018) 1-15,
965 https://doi.org/10.1016/j.engstruct.2017.09.063.
966 [68] J.M. van Noortwijk, A survey of the application of gamma processes in maintenance, Reliab. Eng.
967 Syst. Safe. 94 (1) (2009) 2-21, https://doi.org/10.1016/j.ress.2007.03.019.
968 [69] W. Lin, N. Taniguchi, T. Yoda, Novel method for retrofitting superstructures and piers in aged steel
969 railway bridges, J. Bridge Eng. 22 (2017), https://doi.org/10.1061/(ASCE)BE.1943-5592.0001120.
970 [70] A. Parghi, M.S. Alam, Seismic behavior of deficient reinforced concrete bridge piers confined with
971 FRP - a fractional factorial analysis, Eng. Struct. 126 (2016) 531-546,
972 https://doi.org/10.1016/j.engstruct.2016.08.011.
973 [71] W. Ren, L.H. Sneed, Y. Gai, X. Kang, Test results and nonlinear analysis of RC t-beams
974 strengthened by bonded steel plates, International Journal of Concrete Structures and Materials 9
975 (2) (2015) 133-143, https://doi.org/10.1007/s40069-015-0098-3.
976 [72] D. Isailović, V. Stojanovic, M. Trapp, R. Richter, R. Hajdin, J. Döllner, Bridge damage: Detection,
977 IFC-based semantic enrichment and visualization, Automat. Constr. 112 (2020) 103088,
978 https://doi.org/10.1016/j.autcon.2020.103088.
979 [73] A. Li, Y. Ding, H. Wang, T. Guo, Analysis and assessment of bridge health monitoring mass data—
980 progress in research/development of “Structural Health Monitoring”, Science China Technological
981 Sciences 55 (8) (2012) 2212-2224, https://doi.org/10.1007/s11431-012-4818-5.
982 [74] A. Miyamoto, A. Yabe, Development of practical health monitoring system for short- and medium-
983 span bridges based on vibration responses of city bus, Journal of Civil Structural Health Monitoring
984 2 (1) (2012) 47-63, https://doi.org/10.1007/s13349-012-0017-0.
985 [75] C. Koch, S.G. Paal, A. Rashidi, Z. Zhu, M. König, I. Brilakis, Achievements and challenges in
986 machine Vision-Based inspection of large concrete structures, Adv. Struct. Eng. 17 (3) (2014) 303-
987 318, https://doi.org/10.1260/1369-4332.17.3.303.
988 [76] G. Morgenthal, N. Hallermann, J. Kersten, J. Taraben, P. Debus, M. Helmrich, V. Rodehorst,
989 Framework for automated UAS-based structural condition assessment of bridges, Automat. Constr.
990 97 (2019) 77-95, https://doi.org/10.1016/j.autcon.2018.10.006.
991 [77] A. Khaloo, D. Lattanzi, K. Cunningham, R. Dell'Andrea, M. Riley, Unmanned aerial vehicle
992 inspection of the Placer River Trail Bridge through image-based 3D modelling, Struct. Infrastruct.
993 E. 14 (1) (2018) 124-136, https://doi.org/10.1080/15732479.2017.1330891.
994 [78] R. Sandoval, A. Garcia-Sanchez, F. Garcia-Sanchez, J. Garcia-Haro, Evaluating the More Suitable
995 ISM Frequency Band for IoT-Based Smart Grids: A Quantitative Study of 915 MHz vs. 2400 MHz,
996 Sensors 17 (12) (2017) 76, https://doi.org/10.3390/s17010076.
997 [79] L.Y. Sui, Z.H. Chen, W. Li, L.X. Wang, Study on Monitoring and Safety Early Warning Technology
998 of Bridge Health Based on Internet of Things Technology, Applied Mechanics and Materials 556-
999 562 (2014) 5994-5998, https://doi.org/10.4028/www.scientific.net/AMM.556-562.5994.
1000 [80] M. Aguero, A. Ozdagli, F. Moreu, Measuring Reference-Free total displacements of piles and
1001 columns using Low-Cost, Battery-Powered, efficient wireless intelligent sensors (LEWIS2),
1002 Sensors 19 (7) (2019) 1549, https://doi.org/10.3390/s19071549.
1003 [81] G. Zhou, M. Xie, T. Yi, H. Li, Optimal wireless sensor network configuration for structural monitoring

45
1004 using automatic-learning firefly algorithm, Adv. Struct. Eng. 22 (4) (2019) 907-918,
1005 https://doi.org/10.1177/1369433218797074.
1006 [82] Y. Xu, Y. Turkan, BrIM and UAS for bridge inspections and management, Engineering, Construction
1007 and Architectural Management 27 (3) (2019) 785-807, https://doi.org/10.1108/ECAM-12-2018-
1008 0556.
1009 [83] E. Merschman, M. Doustmohammadi, A.M. Salman, M. Anderson, Postdisaster decision framework
1010 for bridge repair prioritization to improve road network resilience, Transportation Research Record:
1011 Journal of the Transportation Research Board 2674 (3) (2020) 81-92,
1012 https://doi.org/10.1177/0361198120908870.
1013 [84] B.C. Croop, P. Thompson, T. Ahlborn, C.N. Brooks, J. Puckett, M. Fyrster, T.P. Murphy, M. Lopez,
1014 D.M. Banach, The Use of Element Level Data & Bridge Management Software in the Network
1015 Analysis of Big Bridges, No. 0R14-022. 2017. [Online]. Available https://trid.trb.org/view/1506624
1016 [85] P. Sprengholz, C. Betsch, Ok google: using virtual assistants for data collection in psychological
1017 and behavioral research, Behavior Research Methods 54 (3) (2022) 1227-1239,
1018 https://doi.org/10.3758/s13428-021-01629-y.
1019 [86] Y. Li, M.M. Karim, R. Qin, A virtual-reality-based training and assessment system for bridge
1020 inspectors with an assistant drone, IEEE T. Hum.-Mach. Syst. 52 (4) (2022) 591-601,
1021 https://doi.org/10.1109/THMS.2022.3155373.
1022 [87] H. Hemingway, F.W. Asselbergs, J. Danesh, R. Dobson, N. Maniadakis, A. Maggioni, G. van Thiel,
1023 M. Cronin, G. Brobert, P. Vardas, S.D. Anker, D.E. Grobbee, S. Denaxas, Big data from electronic
1024 health records for early and late translational cardiovascular research: Challenges and potential,
1025 Eur. Heart J. 39 (16) (2018) 1481-1495, https://doi.org/10.1093/eurheartj/ehx487.
1026 [88] E. Menditto, D.G.A. Bolufer, C. Cahir, A. Marengoni, S. Riegler, G. Fico, E. Costa, A. Monaco, S.
1027 Pecorelli, L. Pani, A. Prados-Torres, Scaling up health knowledge at European level requires
1028 sharing integrated data: An approach for collection of database specification, Clinicoecon
1029 Outcomes Res 8 (2016) 253-265, https://doi.org/10.2147/CEOR.S97548.
1030 [89] D. Agdas, J.A. Rice, J.R. Martinez, I.R. Lasa, Comparison of visual inspection and Structural-Health
1031 monitoring as bridge condition assessment methods, J. Perform. Constr. Fac. 30 (2016),
1032 https://doi.org/10.1061/(ASCE)CF.1943-5509.0000802.
1033 [90] R. Ma, S. Xu, D. Wang, A. Chen, Vehicle models for fatigue loading on steel box-girder bridges
1034 based on weigh-in-motion data, Struct. Infrastruct. E. 14 (6) (2018) 701-713,
1035 https://doi.org/10.1080/15732479.2017.1359308.
1036 [91] R. Soman, M. Kyriakides, T. Onoufriou, W. Ostachowicz, Numerical evaluation of multi-metric data
1037 fusion based structural health monitoring of long span bridge structures, Struct. Infrastruct. E. 14
1038 (6) (2018) 673-684, https://doi.org/10.1080/15732479.2017.1350984.
1039 [92] K. Liu, N. El-Gohary, Ontology-based semi-supervised conditional random fields for automated
1040 information extraction from bridge inspection reports, Automat. Constr. 81 (2017) 313-327,
1041 https://doi.org/10.1016/j.autcon.2017.02.003.
1042 [93] P. Tierney, A qualitative analysis framework using natural language processing and graph theory,
1043 International review of research in open and distance learning 13 (5) (2012) 173-189,
1044 https://doi.org/10.19173/irrodl.v13i5.1240.
1045 [94] H. Xu, S.P. Stenner, S. Doan, K.B. Johnson, L.R. Waitman, J.C. Denny, MedEx: A medication
1046 information extraction system for clinical narratives, J. Am. Med. Inform. Assn. 17 (1) (2010) 19-24,

46
1047 https://doi.org/10.1197/jamia.M3378.
1048 [95] Q. Li, H. Zhai, L. Deleger, T. Lingren, M. Kaiser, L. Stoutenborough, I. Solti, A sequence labeling
1049 approach to link medications and their attributes in clinical notes and clinical trial announcements
1050 for information extraction, J. Am. Med. Inform. Assn. 20 (5) (2013) 915-921,
1051 https://doi.org/10.1136/amiajnl-2012-001487.
1052 [96] J. Zhang, N.M. El-Gohary, Semantic NLP-Based information extraction from construction regulatory
1053 documents for automated compliance checking, J. Comput. Civil Eng. 30 (2) (2016),
1054 https://doi.org/10.1061/(ASCE)CP.1943-5487.0000346.
1055 [97] E. Soysal, I. Cicekli, N. Baykal, Design and evaluation of an ontology based information extraction
1056 system for radiological reports, Comput. Biol. Med. 40 (11-12) (2010) 900-911,
1057 https://doi.org/10.1016/j.compbiomed.2010.10.002.
1058 [98] Z. Sun, H. Sun, Jiangyin Bridge: An Example of Integrating Structural Health Monitoring with Bridge
1059 Maintenance, Struct. Eng. Int. 28 (3) (2018) 353-356,
1060 https://doi.org/10.1080/10168664.2018.1462671.
1061 [99] B. McGuire, R. Atadero, C. Clevenger, M. Ozbek, Bridge Information Modeling for Inspection and
1062 Evaluation, J. Bridge Eng. 21 (4) (2016), https://doi.org/10.1061/(ASCE)BE.1943-5592.0000850.
1063 [100] N. Hoang, K. Liao, X. Tran, Estimation of scour depth at bridges with complex pier foundations
1064 using support vector regression integrated with feature selection, Journal of Civil Structural Health
1065 Monitoring 8 (2018) 431-442, https://doi.org/10.1007/s13349-018-0287-2.
1066 [101] Z. Kala, Global sensitivity analysis of reliability of structural bridge system, Eng. Struct. 194 (1)
1067 (2019) 36-45, https://doi.org/10.1016/j.engstruct.2019.05.045.
1068 [102] Y.F. Liu, X.P. Fan, Dynamic reliability prediction for the steel box girder based on multivariate
1069 Bayesian dynamic Gaussian copula model and SHM extreme stress data, Structural Control and
1070 Health Monitoring 27 (6) (2020), https://doi.org/10.1002/stc.2531.
1071 [103] J.S. Kong, D.M. Frangopol, Life-Cycle Reliability-Based Maintenance Cost Optimization of
1072 Deteriorating Structures with Emphasis on Bridges, Journal of structural engineering 129 (6) (2003)
1073 818-828, https://doi.org/10.1061/(ASCE)0733-9445(2003)129:6(818).
1074 [104] S. Kim, D.M. Frangopol, Efficient multi-objective optimisation of probabilistic service life
1075 management, Struct. Infrastruct. E. 13 (1) (2017) 147-159,
1076 https://doi.org/10.1080/15732479.2016.1198405.
1077 [105] S. Sabatino, D.M. Frangopol, Y. Dong, Life cycle utility-informed maintenance planning based
1078 on lifetime functions: optimum balancing of cost, failure consequences and performance benefit,
1079 Struct. Infrastruct. E. 12 (7) (2016) 830-847, https://doi.org/10.1080/15732479.2015.1064968.
1080 [106] S. Kim, D.M. Frangopol, Multi-objective probabilistic optimum monitoring planning considering
1081 fatigue damage detection, maintenance, reliability, service life and cost, Structural and
1082 Multidisciplinary Optimization 57 (2018) 39-54, https://doi.org/10.1007/s00158-017-1849-3.
1083 [107] D. Saydam, D.M. Frangopol, Time-dependent performance indicators of damaged bridge
1084 superstructures, Eng. Struct. 33 (9) (2011) 2458-2471,
1085 https://doi.org/10.1016/j.engstruct.2011.04.019.
1086 [108] S. Yang, D.M. Frangopol, L.C. Neves, Optimum maintenance strategy for deteriorating bridge
1087 structures based on lifetime functions, Eng. Struct. 28 (2) (2006) 196-206,
1088 https://doi.org/10.1016/j.engstruct.2005.06.024.
1089 [109] G. Jia, P. Gardoni, Stochastic life-cycle analysis: Renewal-theory life-cycle analysis with state-

47
1090 dependent deterioration stochastic models, Struct. Infrastruct. E. 15 (8) (2019) 1001-1014,
1091 https://doi.org/10.1080/15732479.2019.1590424.
1092 [110] A. Manafpour, I. Guler, A. Radlinska, F. Rajabipour, G. Warn, Stochastic analysis and Time-
1093 Based modeling of concrete bridge deck deterioration, J. Bridge Eng. 23 (2018),
1094 https://doi.org/10.1061/(ASCE)BE.1943-5592.0001285.
1095 [111] R. Kumar, D.B.H. Cline, P. Gardoni, A stochastic framework to model deterioration in
1096 engineering systems, Struct. Saf. 53 (2015) 36-43, https://doi.org/10.1016/j.strusafe.2014.12.001.
1097 [112] G. Jia, P. Gardoni, State-dependent stochastic models: A general stochastic framework for
1098 modeling deteriorating engineering systems considering multiple deterioration processes and their
1099 interactions, Struct. Saf. 72 (2018) 99-110, https://doi.org/10.1016/j.strusafe.2018.01.001.
1100 [113] G. Jia, P. Gardoni, Simulation-based approach for estimation of stochastic performances of
1101 deteriorating engineering systems, Probabilist. Eng. Mech. 52 (2018) 28-39,
1102 https://doi.org/10.1016/j.probengmech.2018.03.001.
1103 [114] S. Ranjith, S. Setunge, R. Gravina, S. Venkatesan, Deterioration prediction of timber bridge
1104 elements using the markov chain, J. Perform. Constr. Fac. 27 (3) (2013) 319-325,
1105 https://doi.org/10.1061/(ASCE)CF.1943-5509.0000311.
1106 [115] P.D. Thompson, E.P. Small, M. Johnson, A.R. Marshall, The pontis bridge management system,
1107 Struct. Eng. Int. 8 (4) (1998) 303-308, https://doi.org/10.2749/101686698780488758.
1108 [116] Z. Tao, R.B. Corotis, J.H. Ellis, Reliability-based bridge design and life cycle management with
1109 Markov decision processes, Struct. Saf. 16 (1) (1994) 111-132,
1110 https://doi.org/https://doi.org/10.1016/0167-4730(94)00030-T.
1111 [117] Y. Huang, Artificial neural network model of bridge deterioration, J. Perform. Constr. Fac. 24
1112 (6) (2010) 597-602, https://doi.org/10.1061/(ASCE)CF.1943-5509.0000124.
1113 [118] R. Assaad, I.H. El-Adaway, Bridge infrastructure asset management system: Comparative
1114 computational machine learning approach for evaluating and predicting deck deterioration
1115 conditions, J. Infrastruct. Syst. 26 (2020), https://doi.org/10.1061/(ASCE)IS.1943-555X.0000572.
1116 [119] Z. Hu, X. Fang, J. Zhang, A digital twin-based framework of manufacturing workshop for marine
1117 diesel engine, The International Journal of Advanced Manufacturing Technology 117 (2021) 3323-
1118 3342, https://doi.org/10.1007/s00170-021-07891-w.
1119 [120] D. Dan, Y. Ying, L. Ge, Digital twin system of bridges group based on machine vision fusion
1120 monitoring of bridge traffic load, IEEE T. Intell. Transp. (2021) 1-16,
1121 https://doi.org/10.1109/TITS.2021.3130025.
1122 [121] J. Kang, K. Chung, E.J. Hong, Multimedia knowledge‐based bridge health monitoring using
1123 digital twin, Multimed. Tools Appl. 80 (2021) 34609-34624, https://doi.org/10.1007/s11042-021-
1124 10649-x.
1125 [122] T. Echaveguren, P. Dechent, M. Giuliano, J. Sepulveda, Proposal of a condition index for
1126 maintenance of runway beams, Proceedings of the Institution of Civil Engineers - Structures and
1127 Buildings 167 (6) (2014) 369-379, https://doi.org/10.1680/stbu.11.00078.
1128 [123] S. Valenzuela, H. de Solminihac, T. Echaveguren, Proposal of an integrated index for
1129 prioritization of bridge maintenance, J. Bridge Eng. 15 (3) (2010) 337-343,
1130 https://doi.org/10.1061/(ASCE)BE.1943-5592.0000068.
1131 [124] T. Echaveguren, P. Dechent, D.C. Universidad, Allocation of bridge maintenance costs based
1132 on prioritization indexes, Revista de la construcción (Universidad Católica de Chile) 18 (3) (2019)

48
1133 568-578, https://doi.org/10.7764/RDLC.18.3.568.
1134 [125] S. Wei, Y. Bao, H. Li, Optimal policy for structure maintenance: A deep reinforcement learning
1135 framework, Struct. Saf. 83 (2020) 101906, https://doi.org/10.1016/j.strusafe.2019.101906.
1136 [126] S. Hou, H. Li, Y. Rezgui, Ontology-based approach for structural design considering low
1137 embodied energy and carbon, Energ. Buildings 102 (2015) 75-90,
1138 https://doi.org/10.1016/j.enbuild.2015.04.051.
1139 [127] J. Zhang, H. Li, Y. Zhao, G. Ren, An ontology-based approach supporting holistic structural
1140 design with the consideration of safety, environmental impact and cost, Adv. Eng. Softw. 115 (2018)
1141 26-39, https://doi.org/10.1016/j.advengsoft.2017.08.010.
1142 [128] K. Meng, C. Cui, H. Li, An ontology framework for pile integrity evaluation based on analytical
1143 methodology, IEEE Access 8 (2020) 72158-72168, https://doi.org/10.1109/ACCESS.2020.2986229.
1144 [129] R. Helmerich, B. Kühn, A. Nussbaumer, Assessment of existing steel structures. A guideline
1145 for estimation of the remaining fatigue life, Struct. Infrastruct. E. 3 (3) (2007) 245-255,
1146 https://doi.org/10.1080/15732470500365562.
1147 [130] C. Shim, H. Kang, N. S. Dang, D. Lee, Development of BIM-based bridge maintenance system
1148 for cable-stayed bridges, Smart Structures and Systems, 20(6) (2017) 697–708,
1149 https://doi.org/10.12989/sss.2017.20.6.697.
1150 [131] C. Yang, Q. Huang, Z. Li, K. Liu, F. Hu, Big Data and cloud computing: Innovation opportunities
1151 and challenges, Int. J. Digit. Earth 10 (1) (2017) 13-53,
1152 https://doi.org/10.1080/17538947.2016.1239771.
1153 [132] Y.F. Duan, Y.L. Xu, Q.G. Fei, K.Y. Wong, K.W.Y. Chan, Y.Q. Ni, C.L. Ng, Advanced finite
1154 element model of tsing ma bridge for structural health monitoring, Int. J. Struct. Stab. Dy. 11 (02)
1155 (2011) 313-344, https://doi.org/10.1142/S0219455411004117.
1156 [133] H. Yang, M. Guo, L. Wang, Y. Hou, Q. Zhao, D. Cao, B. Zhou, D. Wang, Investigation on the
1157 factors influencing the performance of piezoelectric energy harvester, Road Mater. Pavement 18
1158 (2017) 180-189, https://doi.org/10.1080/14680629.2017.1329873.
1159 [134] P. Forsythe, Proactive construction safety systems and the human factor, Proceedings of the
1160 Institution of Civil Engineers - Management, Procurement and Law 167 (5) (2014) 242-252,
1161 https://doi.org/10.1680/mpal.13.00055.
1162 [135] K. Grolinger, W.A. Higashino, A. Tiwari, M.A. Capretz, Data management in cloud
1163 environments: NoSQL and NewSQL data stores, Journal of cloud computing : advances, systems
1164 and applications 2 (1) (2013) 22-24, https://doi.org/10.1186/2192-113x-2-22.
1165 [136] S. Lee, Y. Jeong, A system integration framework through development of ISO 10303-based
1166 product model for steel bridges, Automat. Constr. 15 (2) (2006) 212-228,
1167 https://doi.org/10.1016/j.autcon.2005.05.004.
1168 [137] J. Bien, Modelling of structure geometry in Bridge Management Systems, Arch. Civ. Mech.
1169 Eng. 11 (3) (2011) 519-532, https://doi.org/10.1016/S1644-9665(12)60099-5.
1170 [138] P. Huthwohl, I. Brilakis, A. Borrmann, R. Sacks, Integrating RC Bridge Defect Information into
1171 BIM Models, J. Comput. Civil Eng. 32 (3) (2018), https://doi.org/10.1061/(ASCE)CP.1943-
1172 5487.0000744.
1173 [139] R. Sacks, A. Kedar, A. Borrmann, L. Ma, I. Brilakis, P. Hüthwohl, S. Daum, U. Kattel, R. Yosef,
1174 T. Liebich, B.E. Barutcu, S. Muhic, SeeBridge as next generation bridge inspection: Overview,
1175 Information Delivery Manual and Model View Definition, Automat. Constr. 90 (2018) 134-145,

49
1176 https://doi.org/10.1016/j.autcon.2018.02.033.
1177 [140] K. Afsari, C.M. Eastman, D. Castro-Lacouture, JavaScript Object Notation (JSON) data
1178 serialization for IFC schema in web-based BIM data exchange, Automat. Constr. 77 (2017) 24-51,
1179 https://doi.org/10.1016/j.autcon.2017.01.011.
1180 [141] C.M. Eastman, Y.S. Jeong, R. Sacks, I. Kaner, Exchange model and exchange object concepts
1181 for implementation of national BIM standards, J. Comput. Civil Eng. 24 (1) (2010) 25-34,
1182 https://doi.org/10.1061/(ASCE)0887-3801(2010)24:1(25).
1183 [142] Y. Ji, A. Borrmann, J. Beetz, M. Obergrießer, Exchange of parametric bridge models using a
1184 neutral data format, J. Comput. Civil Eng. 27 (6) (2013) 593-606,
1185 https://doi.org/10.1061/(ASCE)CP.1943-5487.0000286.
1186 [143] Y. Lee, C.M. Eastman, W. Solihin, Logic for ensuring the data exchange integrity of building
1187 information models, Automat. Constr. 93 (2018) 388-401,
1188 https://doi.org/10.1016/j.autcon.2018.06.002.
1189 [144] X. Tao, Y. Liu, P.K. Wong, K. Chen, M. Das, J.C.P. Cheng, Confidentiality-minded framework
1190 for blockchain-based bim design collaboration, Automat. Constr. 136 (2022) 104172,
1191 https://doi.org/10.1016/j.autcon.2022.104172.
1192 [145] J. Xu, H. Liu, Q. Han, Blockchain technology and smart contract for civil structural health
1193 monitoring system, Comput.-Aided Civ. Inf. 36 (10) (2021) 1288-1305,
1194 https://doi.org/10.1111/mice.12666.
1195 [146] M. Das, X. Tao, J.C.P. Cheng, Bim security: a critical review and recommendations using
1196 encryption strategy and blockchain, Automat. Constr. 126 (2021),
1197 https://doi.org/10.1016/j.autcon.2021.103682.
1198 [147] N.O. Nawari, S. Ravindran, Blockchain and the built environment: potentials and limitations,
1199 Journal of Building Engineering 25 (2019), https://doi.org/10.1016/j.jobe.2019.100832.
1200 [148] A.S. Erri Pradeep, T.W. Yiu, Y. Zou, R. Amor, Blockchain-aided information exchange records
1201 for design liability control and improved security, Automat. Constr. 126 (2021),
1202 https://doi.org/10.1016/j.autcon.2021.103667.
1203 [149] R. Zheng, J. Jiang, X. Hao, W. Ren, F. Xiong, Y. Ren, Bcbim: a blockchain-based big data
1204 model for bim modification audit and provenance in mobile cloud, Math. Probl. Eng. 2019 (2019) 1-
1205 13, https://doi.org/10.1155/2019/5349538.
1206 [150] F. Xue, W. Lu, A semantic differential transaction approach to minimizing information
1207 redundancy for bim and blockchain integration, Automat. Constr. 118 (2020),
1208 https://doi.org/10.1016/j.autcon.2020.103270.
1209 [151] N. Nizamuddin, K. Salah, M. Ajmal Azad, J. Arshad, M.H. Rehman, Decentralized document
1210 version control using ethereum blockchain and IPFS, Computers & Electrical Engineering, 76 (2019)
1211 183-197, https://doi.org/10.1016/j.compeleceng.2019.03.014.
1212 [152] X. Tao, M. Das, Y. Liu, J.C.P. Cheng, Distributed common data environment using blockchain
1213 and interplanetary file system for secure bim-based collaborative design, Automat. Constr. 130
1214 (2021), https://doi.org/10.1016/j.autcon.2021.103851.
1215 [153] M. Cheng, G. Liu, Y. Xu, M. Chi, When blockchain meets the aec industry: present status,
1216 benefits, challenges, and future research opportunities, Buildings 11 (8) (2021) 340,
1217 https://doi.org/10.3390/buildings11080340.
1218 [154] V. Plevris, N.D. Lagaros, A. Zeytinci, Blockchain in civil engineering, architecture and

50
1219 construction industry: state of the art, evolution, challenges and opportunities, Frontiers in Built
1220 Environment 8 (2022), https://doi.org/10.3389/fbuil.2022.840303.
1221 [155] T. Guzsvinecz, V. Szucs, C. Sik-Lanyi, Suitability of the kinect sensor and leap motion
1222 controller—a literature review, Sensors 19 (5) (2019) 1072, https://doi.org/10.3390/s19051072.

1223

51