Journal of Modern Power Systems and Clean Energy

A Novel ANFIS-based Power controller for grid-connected PV system with Z-source

Inverter
--Manuscript Draft--

Manuscript Number: MPCE-D-17-00028

Full Title: A Novel ANFIS-based Power controller for grid-connected PV system with Z-source
Inverter

Article Type: Original Research

Funding Information: King Fahd University of Petroleum and Professor Mohammad Abido
Minerals
(Power Research Group project #
RG1420-1&2.)

Abstract: Grid-connected PV power-conversion systems have become an integral part of

modern power systems. A novel Adaptive Neuro-fuzzy Inference System (ANFIS)-
based power controller of Z-source PV conversion system is presented in this work. A
PV module model developed in Simulink was used in this work along with maximum
power point tracking controller (MPPT). PV modules connected to Z-source inverter
have been employed. Pulse width modulation (PWM) techniques for Z-source inverter
have been developed and used. An improved PI based feedback controller is also
implemented. Implementation and results for ANFIS-based feedback controller are
shown and its comparison with PI-based feedback controller is also done. The two
controllers were compared for their robustness, reliability and fast response both under
constant and changing environmental conditions. An experimental setup was also built
for ANFIS-based feedback controller in Real Time Digital Simulator (RTDS)
interconnected with dSPACE1103 controller. The experimental results show the
reliability and effectiveness of the proposed ANFIS-based controller. These results are
extremely important as it opens up the use of these controllers for complex power
conversion systems of all type with effective feedback control under various
environmental conditions.

Corresponding Author: Muhammad Usman Mukhtiar

King Fahd University of Petroleum and Minerals
SAUDI ARABIA

Corresponding Author Secondary

Information:

Corresponding Author's Institution: King Fahd University of Petroleum and Minerals

Corresponding Author's Secondary

Institution:

First Author: Muhammad Usman Mukhtiar

First Author Secondary Information:

Order of Authors: Muhammad Usman Mukhtiar

Mohammad Abido

Zakariya Hamouz

Abdul Mateen Khan

Bilal Jehanzeb

Fazal Illahi

Order of Authors Secondary Information:

Author Comments: Dear Sir,

This Paper is an original article and is part of MS Thesis in Electrical Engineering at

KFUPM, KSA. This has not been published in any other journal and i expect that this
paper has potential to be published in your prestigious Journal.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
 Regards,

Usman

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Title Page Click here to download Title Page Title Page to energy
journal.pdf

1 1

A Novel ANFIS-based Power controller for

2
3
4
5

grid-connected PV system with Z-source

6
7
8

Inverter
9
10
11
12
13
14

Muhammad U. Mukhtiar 1,a, M. A. Abido 1,b, Zakariya Hamouz 1.c, M. A. Mateen 1,d,
15
16
17
18
B. Jehanzeb 1,e , Fazal Illahi 1,f
19
20
21
22
This paper was submitted on 10-01-2017 to Journal of Power Systems and Clean Energy.. This was supported in part by the
23
research projects going in King Fahd University of Petroleum and Minerals (KFUPM), KSA.
24
25 1. King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia
26 a. Muhammad Usman Mukhtiar is with King Fahd University of Petroleum and Minerals (KFUPM), PO 31261, KSA
27 (email: Usman.Mukhtiar@hotmail.co.uk).
28 b. Dr. Mohammad Ali Abido is University Distinguished Professor at KFUPM, PO 1225, KSA (email:
mabido@kfupm.edu.sa).
c. Dr. Zakariya Hamouz is Professor at KFUPM, KSA (email: Zhamouz@kfupm.edu.sa).
29
d. M. Abdul Mateen Khan was research assistant at KFUPM (email:meermateen2000@gmail.com).
30
e. Bilal Jehanzeb was graduate student at KFUPM(bilaljehanzeb@live.com).
31
f. Fazal Illahi was a graduate student at KFUPM(fazalilahi@uohb.edu.sa).
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Blinded Manuscript Click here to download Blinded Manuscript clean energy
1
journal.pdf

1
2
3
4
5
6 Abstract—Grid-connected PV power-conversion systems have become an integral part of modern power
systems. A novel Adaptive Neuro-fuzzy Inference System (ANFIS)-based power controller of Z-source PV
7

conversion system is presented in this work. A PV module model developed in Simulink was used in this work
8
9
10 along with maximum power point tracking controller (MPPT). PV modules connected to Z-source inverter
have been employed. Pulse width modulation (PWM) techniques for Z-source inverter have been developed
11

and used. An improved PI based feedback controller is also implemented. Implementation and results for
12

ANFIS-based feedback controller are shown and its comparison with PI-based feedback controller is also
13
14
15 done. The two controllers were compared for their robustness, reliability and fast response both under
constant and changing environmental conditions. An experimental setup was also built for ANFIS-based
16

feedback controller in Real Time Digital Simulator (RTDS) interconnected with dSPACE1103 controller. The
17

experimental results show the reliability and effectiveness of the proposed ANFIS-based controller. These
18
19
results are extremely important as it opens up the use of these controllers for complex power conversion
20

systems of all type with effective feedback control under various environmental conditions.
21
22

Index Terms—ANFIS Controller, Maximum Power Point Tracking (MPPT), PI Controller,

23
24
Photovoltaics, Pulse Width Modulation (PWM), Real Time Digital Simulator (RTDS), Z-source
25

inverter.
26
27
28
29
30
31

I. INTRODUCTION
32
33
34
35
36
37
Energy crisis and environmental concerns have developed the interest of researchers in renewable energy

sources. Consumption of electricity is likely to increase from 51 GW in 2011 to 120 GW in 2030 [1].
38
39

Renewable energy contribution is expected to be 54 GW by 2032 with PV adding 16GW to the mix
40
41
42

[1].With annual population growth of 3.2%, economic growth of 6.8% and industrial production growth of
43
44

around 6%, the demand for optimum new energy mix is increasing [1]. Increased industrialization over the
45
46
47
last two decades has increased focus on reliability and stability of the supply [2]-[3]. With a large number
48
49

of consumers connected to transmission systems and the number still increasing it’s, therefore, very
50
51
52
important to study ways and means to enhance power system performance in the many countries.
53
54
55
56
57

Reduction in fossil fuels and climatic changes has resulted in search for alternative energy resources of
58
59
60
61 which PV has emerged as the biggest contributor to global energy markets [1], [2]. PV panel production
62
63
64
65
 2
1
2
rate is increasing and has reached 52 GW in 2012 and global installed PV capacity has reached more than
3

100 GW [1]. PV is dependent upon environmental conditions like irradiation and temperature, so the output
4
5

of PV panel is changing, intermittent and not constant. Therefore, an effective, robust and reliable control
6
7
8

system are required to operate PV connected systems effectively and efficiently.

9
10
11
12
13
The control system for a power conversion system must have a fast and reliable dynamic response to
14
15

minimize the effects on users and grids. More recently, various studies of voltage-fed Z-source Inverters
16
17
18
(ZSIs)/ quasi-Z-Source Inverters (qZSI) have been done and their derived topologies are studied to improve
19
20

the voltage gain of the basic ZSI/qZSI [5]. Here, voltage-fed ZSI/qZSI are taken into account in view of
21
22
23
24 their widespread investigations.
25
26
27

Single-loop feedback control with PI controller has been investigated for Z-source using input dc voltage in
28
29
30
[6], using capacitor voltage of Z-source network and using dc-link voltage in [7]. In [8], a sampling circuit
31
32

is used to sample pulsating dc-link voltage to be used as a feedback variable which makes the hardware
33
34
35
very complex. Feedback control is usually based on Z-source capacitor voltage which is used as control
36
37

variable. A relationship between Z-source capacitor voltage and dc link peak voltage has been used. In [9],
38
39
40
41
a proportional control has been implemented for inner current loop to regulate dc-link peak voltage and fast
42

response. Non-linear control for Z-source with a dc source has been implemented using model predictive
43
44

control (MPC) in [10]. In [11], sliding mode control has been employed for simple Z-source network to
45
46
47
regulate dc –link voltage.
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 3
1
C2
L1
2
3

L2 S5
4
5 S1 S3
6

D1
7
Low-Pass
LOAD
Cin Filter
8
C1
S2
9
10 S4 S6
11
12
13

GRID
14
15
16
17

Vabc , Iabc
18
19
20
21 Fig.1 Grid connected PV system with Z-source inverter
22

Fuzzy control for simple Z-source network with a dc source has been used to regulate dc voltage in
23
24
25
26 [12]. Neural network is another intelligent control that has been used to regulate dc-link voltage for Z-
27

source connected with dc source in [13]. Neural network controller needs training by using training data.
28
29

All of these non-linear techniques have been used for dc-source connected to Z-source. However, for
30
31
32
complex system connected to Z-source, these controllers have not been used. In these non-linear control
33
34

methods, quantities like input dc voltage, capacitor voltage of Z-source network, load current, and dc-link
35
36
37
voltage are sampled and used as controller input variables to modulate control signals for switching devices
38
39

[17]-[25].
40
41
42
43
44

Performance, speed and reliability of these non-linear control methods are better as compared to
45
46

conventional PI controllers [17]-[22]. However, in complex application like grid connected mode of wind
47
48
49
energy and PV systems, the use of these controllers is limited and needs improvement and robust design
50
51

[25]-[32]. Model Predictive Control (MPC) was used to adjust output waveform, response and robustness
52
53
54
with minimum cost function but it limits the switching states which makes it unsuitable under complex
55
56

applications [21]-[24].
57
58
59
60
61
62
63
64
65
 4
1
2
3
In this work, a novel ANFIS-based feedback controller has been developed and simulated in Simulink. Real

time simulation for grid-connected PV system with Z-source inverter have been carried out to improve the
4
5

effectiveness and robustness of feedback control. A model of PV module has been developed and Z-source
6
7
8

inverter has been integrated with PV-module. The ANFIS controller has been employed in dSPACE1103
9
10

and grid-connected PV system has been developed in Real Time Digital Simulator (RTDS).
11
12
13
14

II. MODELLING OF PV SYSTEMS

15
16
17

A. PV Module
18
19
20

PV modules can be modelled using an equivalent electrical models which can also be connected in any
21
22
23
electrical circuit and thus can easily be used for the overall transient and dynamic studies of a power system
24
25

especially when connected to the grid. Different types of PV models have been given in literature such as
26
27
28
29
30
double diode model, RSH model, RS and ideal diode model [4]. Fig.2 shows the five parameters model (R SH

model) with a shunt resistor and series resistor to give results closer to that of practical module.
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48 Fig.2 PV Module model
49
50

The efficiency, performance and effectiveness of these models depend on the values of the parameters
51
52
53
54 which in turn depend on the algorithm used for identification and calculation. The five parameters are Rsh,
55

Rs, a, Id and IL. Rsh represents shunt resistance, Rs stands for series resistance, a is the modified ideality
56
57

factor, Id is the diode current and IL is the PV current.

58
59
60
The equations governing the five parameter model are given below:
61
62
63
64
65
 5
1
2
Applying Kirchhoff’s current law:
3

= − − (1)
4
5
6
7 The current through the diode is given by:
8
( )
= {exp − 1} (2)
9
10
11

And current through the series resistance is given by:

12
13
14
= (3)
15
16
17
Putting these values in equation (1) to obtain the following equation:
18
19
( )
= − exp −1 −
20
21 (4)
22

Where “a “ is the modified diode ideality factor and is given by:

23
24
25

= (5)
26
27
28

The IV-characteristic at some key points is obtained as below:

29
30
31
32 At short circuit condition, V =0, and
33

= − exp −1 −
34
I= . (6)
, , , ,
, , , ,
35
36 ,
37

At open circuit, I =0 and V = ,

38
,
39
40

= {exp − 1} +
41
(7)
, ,
, ,
42
43 ,
44
45
46
At maximum power point, I= , and V= ,
47
= − exp −1 − (8)
, , , , , ,
, , ,
48
49 ,
50
Diode ideality factor (a), reference diode current (Io), reference load current (IL), series resistance and
51
52

shunt resistance are the five parameters that were optimized through Differential Evolution technique were
53
54
55
56
57
used to build the PV model. Differential Evolution (DE) technique is an optimization technique to calculate

the unknown values of variables that satisfies an objective function to minimize the error between actual
58
59
60
61
62
63
64
65
 6
1
2
output and calculated one by the DE. The developed PV model was used in Simulink simulations and real-
3

time simulation in Real Time Digital Simulator.

4
5
6

B. Z-source inverter
7
8
9

Z-source Inverter has been used by researchers in place of conventional current source and voltage source
10
11
12
13 inverter to improve the efficiency of the PV systems. Conventional current source inverter or voltage
14

source inverter can only boost or buck the voltage and also electromagnetic interference and other
15
16

restrictions lowers their efficiency while Z-source inverter is buck-boost inverter which eliminates various
17
18
19
problems with conventional inverters and also the need for boost converter is removed by the use of Z-
20
21

source inverter. Z-source inverter and its improved performance with PV systems is explored in this work.
22
23
24
To overcome these limitations, Z-source inverter is suggested which is given in Fig.1. It consists of two
25
26

capacitors and two inductors connected [5] as shown in Fig.1. The switches used can have series diodes or
27
28
29
30
anti parallel diodes
31

If the switching period is T and the shoot through time is To, is the voltage across capacitor C1, is
32
33
34
35
the voltage across capacitor C2, and are the voltage across inductor L1 and L2, is output voltage
36

of PV module, is the input voltage of inverter.

37
38
39
= , + = , =0 (9)
40
41

During any non-zero state as shown in Figure 2.15, the equations are as follow:
42
43
44
− =− , = − = + (10)
45
46

The average voltage across the DC link is given by:

47
48
49
( )
= = = . (11)
50
51
52

Where B is the boost factor obtained from zero state and is given by:
53
54
55
= = ≥1 (12)
56
57
58

The output ac voltage from the inverter is given by:

59
60
61
62
63
64
65
 7
1
= . . (13)
2
3
4
5
6
Where M is the modulation index. The output voltage can be stepped up or stepped down by choosing a

factor Bb such that

7
8

= . = (0 ~ ∞)
9
10 (14)
11

Quasi-Z-source inverter was used in this research work as shown in figure.1 to transfer power to load in
12
13

grid connected mode.

14
15
16
17
18
19
20
21
22

III. Proposed Maximum Power Point Tracking Control

23
24
25

The purpose of Maximum power tracking controller is to apply proper resistance after sampling the output
26
27

from PV to track the maximum power point, thus extracting maximum power from the PV under the given
28
29
30
conditions of irradiation and temperature. For constant condition there is only operating point at which the
31
32

corresponding voltage and current would result in maximum power.

33
34
35
The IV characteristic is generally linear but behaves like inverse exponential function at the knee.
36
37

Maximum power will be extracted from the device when the slope of the curve is equal and opposite to I/V
38
39
40
41
42
ratio. This point is also called maximum power point. A resistance value of V/I will extract the maximum

power from the device. If the resistance is different from this value, maximum power will not be extracted
43
44

and maximum power point tracking controller is required to search for this point [6].
45
46
47

Improved Incremental conductance method is used in this work to calculate the required duty cycle. This
48
49

duty cycle is used by shoot-through reference generator to generate upper and lower shoot-through
50
51
52
references for PWM as shown in Fig.3.The shoot-through reference is calculated using the formulae for
53
54

simple boost control. A PI based controller and an ANFIS-based feedback control have been employed for
55
56
57
58
59
system given in Fig.1.
60
61
62
63
64
65
 8

A. PI based αβ-Controller
1
2
3
4
PI-based alpha-beta controllers based on alpha-beta conversions without phase angle consideration by PLL,
5
6

thus allowing simplifying the feedback control using stationary reference frame domain. Alpha-beta based
7
8
9
10
PI controllers are implemented as shown in Fig.4.
11
12
13
14
15 Vn*
-1
16
17
18
Vpv
19
Duty
20
cycle Shoot-
Vp*
21
through
MPPT
22
Ipv reference
23
generator
24
25
26
27
28
29 Fig .3. Shoot-through reference generator
30
31
32
33
The voltages and currents are sampled from the grid side and converted to alpha-beta domain using

transformations and reference Iα and Iβ are calculated using (24) and (25). Three PI controllers are used to
34
35

calculate reference real power P*, reference Vα* and Vβ*.

36
37
38
39
40

The gains for first PI controller that tracks the DC voltage reference for Z-source capacitor C1 comes out to
41
42
43
be Kp = 800 and KI = 1800. The gains for second and third PI controllers were same and come out to be
44
45

Kp = 50 and KI = 10 for both. The block diagram of Alpha-beta based PI controllers is shown in Fig.4. The
46
47
48
three reference modulating signals and two shoot-through references are calculated and these five signals
49
50

are compared with triangular carrier signal to generate required pulses for the IGBTs.
51
52
53
54
55

If reactive power reference Q is zero, reference Iα and Iβ are given by:

56
57
58
59
60
61
62
63
64
65
 9
1
∗
= (15)
2
3
4
5
6
7
∗
= (16)
8
9
10

B. ANFIS based Controller

11
12
13
14
15 Adaptive Neuro-fuzzy inference systems are used to train systems according to input-output data. The
16

ANFIS system uses either Takagi-Sugeno models or Mamdani models to build up trained system. Takagi-
17
18

Sugeno system allows membership functions for output as well while Mamdani models allow only linear
19
20
21
outputs. In ANFIS models, inputs are assigned membership functions and the inputs will attain values from
22
23

these membership functions and Takagi-Sugeno models are used to fuzzify the inputs and then
24
25
26
defuzzifications are done using one of the four available methods to obtain outputs. For training of ANFIS,
27
28

there are two methods; assign the rules manually or using training data to train ANFIS which assign the
29
30
31
32
33
rules itself.
34
35

For the dc reference controller, rules are developed according to Table.1 where different states are assigned
36
37
38

to the two inputs. The two inputs for the first ANFIS controller is difference between reference value of
39
40

Vc1 and its measured value and the change in error. Five states are assigned to each input namely negative
41
42
43
small (NS), Negative large (NL), positive small (PS), positive large (PL) and zero (Z).
44
45
46
47
48
49
50

Table.1 Rules for ANFIS DC Regulating Controller

51
52
53

e NL NS Z PS PL
54
55
56
57
58
59 NL NL NL NS Z PS
60
61
62
63
64
65
 10
1
2 NS NL NS NS NL PS
3
4
5 Z NS NS Z PS PS
6
7
8 PS Z NL PS PS PL
9
10
11 PL PS PS PS PL PL
12
13
14

The Table.1 shows that as the error (e) and derror ( ) increases, the output of the ANFIS control increases
15
16
17
in positive direction. With decrease in both, the output decreases. Similarly, the increase or decrease in
18
19

positive or negative direction is determined by relative change in error in positive and negative direction.
20
21
22
23
24
For current controller 1 and 2, training was done using cyclic and linear data. For training ANFIS, 7

Gaussian membership functions were assigned to each input of current controller. Similarly, 7 membership
25
26
27
28 functions were assigned to each input of current controller 2 and training was done using back propagation
29

algorithm. Training data was obtained from grid-connected PV systems. After training ANFIS controllers
30
31

were obtained.
32
33
34
The complete control diagram based on ANFIS controllers is shown in Fig.5. The feedback control is based
35
36

on ANFIS to track required dc reference. The Reference modulating signals are also calculated using
37
38
39
ANFIS controllers in order to control power flow to load in grid-connected. The shoot-through reference
40
41

for modulating signals is calculated by SBC PWM technique.

42
43
44
45
46
47
IV. SIMULATION RESULTS

The System Specifications are given in Table 2. The maximum voltage of PV is 170 V and maximum
48
49
50
current is 6.05 A which corresponds to 1000 watt at maximum power point. Three phase LC low pass filter
51
52

was designed. The capacitance value for LC filter was 200 uF per phase and inductance value was 12 mH
53
54
55
per phase. PV module parameters are obtained using differential evolution with reference diode ideality
56
57

factor, a equal to 1.77549, reference diode current, Io equal to 1.648437 A, reference load current, IL equal
58
59
60
61
to 3.35 A, series resistance equal to 0.004773 ohm and shunt resistance equal to 6.59505 ohm.
62
63
64
65
 11
1
2
3
PV module parameters are obtained using differential evolution with reference diode ideality factor, a equal

to 1.77549, reference diode current, Io equal to 1.648437 A, reference load current, I L equal to 3.35 A,
4
5

series resistance equal to 0.004773 ohm and shunt resistance equal to 6.59505 ohm.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Fig .4. PI-based αβ Controller
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
 12
1

Comparison of PI and ANFIS control strategies

2
3
4

Case I : Constant Irradiance

5
6
7
8
9
10
The simulations results shows the values of output current and voltages, V pv and Ipv tracking, Vc1 tracking,

output power and reference signals generated by controllers for fixed irradiance of 1000 W/m 2 are shown in
11
12

Fig.7(a)-7(c). The output voltage is synchronized with the grid voltage and maximum power point is
13
14
15

efficiently tracked.
16
17
18
19
20
The irradiation for PV module was constant at 1000 w/m 2. Fig.7e and Fig.7f show output current and
21
22

voltage. The voltage settles to nominal value after some time at rms value of 190 V and output current
23
24
25
settles at the rms value of 5 A. The output waveforms are sinusoidal at 60Hz synchronized with the grid.
26
27

The inverter voltage switches between 400V and -400V. The output voltage and current has some initial
28
29
30
31 transients after which it comes to a steady state value. This transient is during the voltage rise across
32

capacitor of Z-source inverter but once the voltage across the capacitor reaches the reference voltage of
33
34

400, the output voltage and current also reaches the steady state value of 190 V and 10 A rms. ANFIS
35
36
37
causes the capacitor voltage to reach reference voltage quickly as compared to αβ-controller thus reaching
38
39

the steady state values quickly.

40
41
42
43
44
45 Table 2. System Specifications
46
Quantity Value
47
48
PV Maximum Voltage 170 V
49
50

PV Maximum Current 6.05 A

51
52

PV Maximum Power 1000 watt

53
54
55
56 Z-source Inductors 160 uH
57
58 Frequency 60 Hz
59
Z-source Capacitors 1000 uF
60
61
62
63
64
65
 13
1
2 Low Pass Filter Capacitor 200 uF
3
Low Pass Filter Inductor 12 mH
4
5
Carrier frequency 7000Hz
6
7

Three Phase Load 100 ohm per phase

8
9

Grid Voltage 190 V

10
11
12

The MPPT tracking is shown in Fig.7. The maximum power of 1000 watt is tracked by the controller as
13
14
15
shown in Fig.7d. The maximum voltage of PV is 170 V and is tracked quickly by αβ-controller based
16
17

MPPT as shown in Fig.7a. The maximum current is 6.05 from PV module is tracked by the controller as
18
19
20
21
shown in Fig.7c. The dc voltage is tracked to its reference value to keep the output voltages and currents
22

constant. After the tracking of maximum power point, reference voltage, maximum PV voltage and current,
23
24

these quantities are kept constant which shows the reliable behavior of the controller.
25
26
27
.
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Fig.5. ANFIS Based Power Controller
58
59
60
61
62
63
64
65
 14
1
2
The results for ANFIS controller for the same system under same conditions are shown in comparison to
3

αβ-controller. The maximum power point tracking comparison by ANFIS controller has been shown in
4
5

Fig.7 in. comparison to αβ-controller. The maximum voltage and current tracking comparison are shown in
6
7
8

Fig.7a and Fig.7c. ANFIS based feedback controller for the same system gives improved results. For
9
10

constant irradiance and temperature case, the ANFIS Controller is fast and accurate in tracking the dc
11
12
13
reference.
14
15

Case II : Changing Irradiance

16
17
18

Also, ANFIS controllers are fast in tracking the maximum power point with accuracy. The comparison of
19
20

the two types of controllers are on the basis of dc-reference tracking, output power delivery to load and
21
22
23

grid, maximum PV voltage and current tracking fully explains the reliability and a fast and effective control
24
25

under
26
27
28
The input PV maximum voltage is 215 volts and it is boost up to 320 volts. The ANFIS controllers were
29
30

implemented in dSPACE 1103. This controller takes output current, out voltage, Z-source capacitor
31
32
33
voltage, Vpv, Ipv as control variables and generate five control signals for PWM modulation, i.e., three
34
35

modulation signals and 2 shoot-through envelops.

36
37
38
Zsourcegrid_ANFIS/Irradiance (W//m2) 2 : Ramp-up/down Irradiance
39
40 1100
Ir
41
1050
42
43
1000
44
45
950
46
47
900
48
49 850
50
51 800
52
750
53
54
55 700
56
57 650
0 0.5 1 1.5 2 2.5
58 Time (sec)
59
Fig.6 Changing irradiations
60
61
62
63
64
65
 15
1
2
The results show the robustness and reliability of ANFIS controllers in real-time environment. The output
3

voltage level are 190 volts rms and output current for load is 10 A rms. Carrier frequency was set at 6500
4
5

Hz. Fig.8(e) shows the interconnection between small time-step and power system components through
6
7
8

Norton equivalent diagram. The Z-source inverter is in built in small time-step components while the grid,
9
10

LC filter and PV module are in power system components. Input and Output cards send and receive signals
11
12
13
to and from dSPACE1103.
14
15

Nine signals including three phase output voltage and three output current, PV module voltage and current
16
17
18
and Z-source capacitor dc voltage is sent to dSPACE 1103. All the results are shown in Fig 9(a)-9(f). Fig
19
20

9(a) shows the maximum PV voltage being tracked efficiently by the ANFIS based controller. The PV
21
22
23
24 voltage is 215 V and dc capacitor voltage after boosting PV voltage is 320 V as shown in Fig 9(b). Fig 9(c)
25

shows that the output voltage is 190 V and is synchronized with the grid voltage at 60 Hz. Fig 9(d) shows
26
27

the output current of 10A delivered to the grid-connected load.

28
29
30
31
32
33 Vpv Comparison for ANFIS and PI Controllers
Vc1 Comparison
180
500
34
160 450

35 140
400

36 120 ANFIS
350

300
37 100 PI
Voltage(V)

Voltage(V)

250 PI

38 80
200 ANFIS

39 60
150

40
40 100

20 50
41
0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5
42 Time(Seconds) Time(Seconds)

Fig.7a.Comparison of Vpv Fig.7b.Comparison of Vc1

43
44
45
Ipv Comparison for ANFIS and PI controllers
46 8.5
3000
Output Power Comparison

8
47
7.5 2500
48 7 ANFIS

PI 2000
49 6.5

50 6
1500
Current(A)

P(watt)

5.5
51 1000
5

52 4.5
500

53 4
ANFIS
PI

0
54 3.5

3
55 0 0.2 0.4 0.6 0.8 1
Time(Seconds)
1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1
Time(Seconds)
1.2 1.4 1.6 1.8 2

56
57 Fig.7c. Comparison of Ipv Fig.7d.Comparison of Output Power
58
59
60
61
62
63
64
65
 16
1
Output Voltage
2 250
15
output current

3 200

10
4 150

5 100 5

6 50
0
voltage(V)

current(A)
7 0

-5
-50
8
-100
9 -10

-150
10 -15
-200
11
-250 -20
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6
12 Time (seconds) Time (seconds)

13
14
15
16 Fig.7f. Output Voltage for ANFIS Controller Fig.7e. Output Current for ANFIS Controller
17
18
19

Fig 9(e) shows the pulses generated by the implemented PWM technique with the MPPT control. The
20
21
22
Shoot-through and non-shoot-through states are shown and these pulses are provided to the IGBTS for
23
24

switching. Control Signal generated by ANFIS based control coming from dSPACE1103 controller are
25
26
27
shown in Fig 9(f). The maximum duty cycle ratio calculated by the controller is shown in Fig 8(f). The
28
29

ANFIS based controller is fast in tracing the reference points and its performances in the real time
30
31
32
33
34
environment is reliable and effective.

The five reference signals calculated by the controller are sent to the RTDS through interconnected cables
35
36
37
38 and these five signals are compared with a triangular carrier signal to generate pulses to track the reference
39

dc voltage and the maximum power and synchronize them with grid signals. The simulation of controller in
40
41

Simulink\Matlab environment and real time environment shows the effectiveness of the controller. The
42
43
44
design of ANFIS controller is simpler yet it is robust in its performance. The time step of simulation for
45
46

dSPACE1103 controller is 50 microseconds.

47
48
49
50

V. CONCLUSION
51
52
53

New ANFIS-based power controllers were proposed in this research work. Three ANFIS controllers were
54
55

used; one for dc reference tracking and two for reference voltage signal generators to control PWM and
56
57
58
generate switching signals for IGBTs. The proposed ANFIS controller was implemented in ANFIS
59
60

controller was also implemented in real time environment of RTDS and dSPACE 1103 and its performance
61
62
63
64
65
 17
1
2
3
was evaluated in real time environment. This proposed controller is very efficient, reliable, and fast is

suitable for use in complex power conversion systems and in multi-level cascaded inverter systems.
4
5
6
7 180
Vpv Comparison for ANFIS and PI controllers
Ipv Comparson for ANFIS and PI Controllers
7
8 160
Anfis
9 140
6.5 PI
Anfis
10 120
PI 6

11 100
Voltage(V)

Current(A)
12 80 5.5

13 60
5

14 40

15 20 4.5

16 0
0 0.5 1 1.5 2 2.5
Time(seconds) 4
0 0.5 1 1.5 2 2.5
17 Time(seconds)

Fig.8a. Comparison of Vpv under changing Conditions Fig.8b. Comparison of I pv under changing conditions
18
19
20
Vc1 Comparison for Anfis and PI Controllers
21 500
1400
Pout Comparison for ANFIS and PI Controllers

22 450
1200

400
23 1000

350
24 800

300
25 600
Voltage(V)

Anfis
P(WATT)

250
Anfis
26 200 PI
400

27 150
200

28 100
0 PI

29 50 -200

30 0
0 0.5 1 1.5 2 2.5
-400
0 0.5 1 1.5 2 2.5
Time(Seconds) Time (seconds)

31
Fig.8c.Comparison of Vc1 under changing conditions Fig.8d. Comparison of Pout under changing conditions
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49

Fig.8e. PV system in RSCAD Fig.8f. Shoot-through reference from dSPACE11

50
51
52
53

VI. ACKNOWLEDGEMENTS
54
55
56

The authors would like to acknowledge the support of Deanship of Scientific Research (DSR) at the King
57
58
59
60
Fahd University of Petroleum and Minerals, through the Power Research Group project # RG1420-1&2.
61
62
63
64
65
 18
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fig.9a. PV voltage Fig.9b. Vc1 voltage across Z-source capacitor
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Fig.9c Output Voltage from RSCAD Fig.9d. Output Current from RSCAD
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Fig.9e. Pulses for IGBTs Fig.9f. Control signals Generated by dSPACE1103
50
51
52
53
54

VII. REFERENCES
55
56
57

[1] “International Energy Outlook 2013 - Energy Information Administration.”

58
59
60
[Online]. Available: http://www.eia.gov/forecasts/ieo/. [Accessed: 20-May-2014].
61
62
63
64
65
 19
1
2
[2] “ICER RES and DG Report_FINAL - ICER RES and DG Report_FINAL.pdf.” [Online].
3

Available:http://www.naruc.org/international/Documents/ICER RES and DG Report_FINAL.pdf.

4
5

[Accessed: 20-May-2014].
6
7
8

[3] O. Ellabban, J. Van Mierlo, and P. Lataire, “Experimental study of the shoot-through boost control
9
10

methods for the Z-source inverter,”Eur. Power Electron. Drives Assoc. J., vol. 21, no. 2, pp. 18–29,
11
12
13
June 2011.
14
15

[4] R. Strzelecki, W. Bury, M. Adamowicz, and N. Strzelecka, “New alternative passive networks to
16
17
18
improve the range output voltage regulation of the PWM inverters,” in Proc. 2009 IEEE Applied Power
19
20

Electronics Conf. Exposition (APEC), vols. 1–4, pp. 857–863.

21
22
23
24 [5] H. Rostami and D. A. Khaburi, “Voltage gain comparison of different control methods of the Z-source
25

inverter,” in Proc. Int. Conf. Electrical Electronics Engineering (ELECO 2009), Nov. 5–8, pp. I-268–I-
26
27

272.
28
29
30
[6] Y. Liu, B. Ge, H. Abu-Rub, and F. Z. Peng, “Overview of space vector modulations for three-phase Z-
31
32

source/quasi-Z-source inverters,” IEEE Trans. Power Electron., vol. 29, no. 4, pp. 2098–2108, Apr.
33
34
35
2014
36
37

[7] "Global Market Outlook For Photovoltaics Until 2016,"(EPIA) Industry Association
38
39
40
41
[Online].Available:http://www.epia.org/fileadmin/user_upload/Publications/Global Market-Outlook-
42

2016.pdf. [Accessed: 20-May-2014].

43
44

[8] M. Zagrouba, A. Sellami, M. Bouaïcha, and M. Ksouri, “Identification of PV solar cells and modules
45
46
47
parameters using the genetic algorithms: Application to maximum power extraction,” Sol. Energy, vol.
48
49

84, no. 5, pp. 860–866, May 2010.

50
51
52
[9] R. Chenni, M. Makhlouf, T. Kerbache, and A. Bouzid, “A detailed modeling method for photovoltaic
53
54

cells,” Energy, vol. 32, no. 9, pp. 1724–1730, Sep. 2007.

55
56
57
58
59
60
61
62
63
64
65
 20
1
2
[10] Mei Shan Ngan; Chee Wei Tan, "A study of maximum power point tracking algorithms for stand-
3

alone Photovoltaic Systems," Applied Power Electronics Colloquium (IAPEC), 2011 IEEE , vol., no.,
4
5

pp.22,27, 18-19 April 2011

6
7
8

[11] Y. Li, J. Anderson, F. Z. Peng, and D. Liu, “Quasi-Z-Source Inverter for Photovoltaic Power
9
10

Generation Systems,” 2009 Twenty-Fourth Annu. IEEE Appl. Power Electron. Conf. Expo., pp. 918–
11
12
13
924, Feb. 2009.
14
15

[12] J. B. Liu, J. G. Hu, and L. Y. Xu, “Dynamic modeling and analysis of Z-source converter—Derivation
16
17
18
of AC small signal model and designoriented analysis,” IEEE Trans. Power Electron., vol. 22, no. 5, pp.
19
20

1786–1796,Sept. 2007.
21
22
23
24 [13] Abu-Rub, H.; Iqbal, A; Moin Ahmed, S.; Peng, F.Z.; Yuan Li; Ge Baoming, "Quasi-Z-Source
25

Inverter-Based Photovoltaic Generation System With Maximum Power Tracking Control Using
26
27

ANFIS,"Sustainable Energy, IEEE Transactions on , vol.4, no.1, pp.11,20, Jan. 2013

28
29
30
[14] M. Mosa, H. Abu-Rub, and J. Rodríguez, “High performance predictive control applied to three phase
31
32

grid connected quasi-Z-source inverter,” in Proc. 39th Annu. Conf. IEEE Industrial Electronics Society
33
34
35
(IECON 2013), Nov.10–13, pp. 5812–5817.
36
37

[15] H. Rostamiand D. A. Khaburi, “Neural networks controlling for both the DC boost and AC output
38
39
40
41
voltage of Z-source inverter,” in Proc. 1st Power Electronics Drive System Technology Conf., Feb.17–
42

18, 2010, pp. 135–140.

43
44

[16] W. Mo, P. C. Loh, and F. Blaabjerg, “Model predictive control for Z-source power converter,” in
45
46
47
Proc. 8th Int. Conf. Power Electronics, 30 May–3 June, 2011, pp. 3022–3028.
48
49

[17] H. Abu-Rub, A. Iqbal, Sk. Moin Ahmed, F. Z.Peng, Y. Li, and G. Baoming, “Quasi-Z-source
50
51
52
inverter-based photovoltaic generation system with maximum power tracking control using
53
54

ANFIS,”IEEE Trans. Sustain. Energy, vol. 4, no. 1, pp. 11–20, Jan. 2013.
55
56
57
58
59
60
61
62
63
64
65
 21
1
2
[18] Y. Xue, B. Ge, and F. Z. Peng, “Reliability, efficiency, and cost comparisons of MW scale
3

photovoltaic inverters,” in Proc. 2012 IEEE Energy Conversion Congress and Exposition
4
5

(ECCE),Sept.15–20, pp. 1627–1634.

6
7
8

[19] Y. Shuitao, F. Z. Peng, L. Qin, R. Inoshita, and Q.Zhaoming, “Current-fed quasi-Z-source inverter
9
10

with voltage buck-boost and regeneration capability,” IEEE Trans. Ind. Applicat., vol. 47, no. 2, pp.
11
12
13
882–892, 2011.
14
15

[20] Q. Lei, D. Cao, and F. Z. Peng, “Novel loss and harmonic minimized vector modulation for a current-
16
17
18
fed quasi-Z-source inverter in HEV motor drive application,” IEEE Trans. Power Electron., vol. 29, no.
19
20

3, pp. 1344–1357, Mar. 2014.

21
22
23
24 [21] L. Poh Chiang, D. M. Vilathgamuwa, C. J. Gajanayake, W. Li Tya n, and A. C. Ping, “Z-source
25

current-type inverters: digital modulation and logic implementation,”IEEE Trans. Power Electron., vol.
26
27

22, no. 1, pp. 169–177, 2007.

28
29
30
[22] D. Caoand F. Z. Peng, “A family of Z-source and quasi-Z-source DC-DC converters,” in Proc.2009
31
32

IEEE Applied Power Electronics Conf. Exposition (APEC), vols. 1–4, pp. 1097–1101.
33
34
35
[23] R.Strzeleckiand D. Vinnikov, “Models of the qZ-Converters,” Przeglad Elektrotechniczny, vol. 86,
36
37

no. 6, pp. 80–84, June 2010.

38
39
40
41
[24] D.Vinnikovand I. Roasto, “Quasi-Z-sourcebased isolated DC/DC converters for distributed power
42

generation,” IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 192–201, Jan. 2011.
43
44

[25] D. Vinnikov, I. Roasto, R. Strzelecki, and M. Adamowicz, “Step-up DC/DC converters with cascaded
45
46
47
quasi-Z-source network,” IEEE Trans. Ind. Electron., vol. 59, no. 10, pp. 3727–3736, Oct. 2012.
48
49

[26] D. Vinnikov, J. Zakis, O. Husev, and R. Strzelecki, “New high-gain step-up DC/DC converter with
50
51
52
high-frequency isolation,” in Proc. 27th Annu. IEEE Applied Power Electronics Conf. Exposition
53
54

(APEC), Mar. 2012, pp. 1204–1209.

55
56
57
58
59
60
61
62
63
64
65
 22
1
2
[27] O. Husev, A. Blinov, D. Vinnikov, and A. Chub, “Steady-state analysis of qZS-derived push-pull
3

DC/DC converter with wide input voltage regulation range,” in Proc. 8th Int. Conf. Compatibility and
4
5

Power Electronics (CPE), June5–7, 2013, pp. 320–325.

6
7
8

[28] E. Romero-Cadaval, G. Spagnuolo, L. Garcia Franquelo, C. A. Ramos-Paja, T. Suntio, and W.

9
10

M.Xiao, “Grid-connected photovoltaic generation plants: Components and operation,” IEEE Ind.
11
12
13
Electron. Mag., vol. 7, no. 3, pp. 6–20,Sept. 2013.
14
15

[29] M. Begovic, “Renewable energy technologies: The future of photovoltaics,” in Proc. 2012 IEEE PES
16
17
18
Innovative Smart Grid Technologies (ISGT), Jan.16–20, pp. 1–1.
19
20

[30] R. Singh, G. F. Alapatt, and A. Lakhtakia, “Making solar cells a reality in every home: opportunities
21
22
23
24 and challenges for photovoltaic device design,” IEEE J. Electron Devices Soc., vol. 1, no. 6, pp. 129–
25

144, June 2013.

26
27

[31] J. M. Carrasco, L. G. Franquelo, J. T.Bialasiewicz, E. Galvan, R. C. P. Guisado, M. A. M. Prats, J.

28
29
30
I. Leon, and N. Moreno-Alfonso, “Power-electronic systems for the grid integration of renewable
31
32

energy sources: A survey,” IEEE Trans. Ind. Electron., vol. 53, no. 4, pp. 1002–1016, June 2006.
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65