DEDAN KIMATHI UNIVERSITY OF TECHNOLOGY

SCHOOL OF ENGINEERING

DEPARTMENT OF ELECTRICAL AND ELECTRONICS ENGINEERING.

BACHELOR OF SCIENCE IN ELECTRICAL AND ELECTRONICS ENGINEERING

PROJECT TITLE:

OPTIMIZATION OF TRANSFORMER TAPS TO REDUCE VOLTAGE

VARIATION.

NAMES: REG NO:

WAFULA JOEL KATE E021-1229/2011

JIMMY KISAKA E021-1215/2011

PROJECT SUPERVISOR: MR. JUMA WABWIRE

PROJECT PROPOSAL SUBMITTED IN PARTIAL FULFILLMENT OF THE

REQUIREMENT FOR THE AWARD OF THE DEGREE OF BACHELOR OF SCIENCE

IN ELECTRICAL AND ELECTRONICS ENGINEERING OF DEDAN KIMATHI

UNIVERSITY OF TECHNOLOGY 2015.

DECLARATION OF ORIGINALITY:

1) We declare that this final year project report is our original work and has not been

submitted elsewhere for examination, award of a degree or publication. Where other

people’s work or our own work has been used, this has properly been acknowledged and

referenced in accordance with the University requirements.

2) We understand what plagiarism is and we are aware of the university policy in this regard

to this.

3) We understand that any false claim in respect of this work shall result in disciplinary

action, in accordance with University anti-plagiarism policy.

4) We have not sought or used the services of any professional agencies to produce this

work.

KATE JOEL WAFULA

Signature:……………………………………………………Date:………………………………..

KISAKA JIMMY

Signature:……………………………………………………Date:………………………………
CERTIFICATION:

This report has been submitted to the Department of Electrical and Electronics Engineering of

Dedan Kimathi University of Technology with my approval as supervisor:

Signature:………………………………………Date:……………………………………………..

MR.JUMA WABWIRE.
DEDICATION:

We would like to dedicate this to our entire families for their faith in us and our colleagues for

their solid support.

ACKNOWLEDGEMENT:

We would like to acknowledge the Electrical and Electronics Engineering Department for

entrusting us with this project. We thank our supervisor, Mr. Juma Wabwire for supervising us

throughout this endeavor. His insightful guidance cannot go unmentioned. We also give our

gratitude to Mr.Opondo for guiding us through some areas of difficulty.

We would also like to thank our families for their hard work and dedication in ensuring that we

have the chance to pursue this degree. We would also like to thank our friends and fellow

classmates who believed in us and encouraged us to always push on.

ABSTRACT:

Voltage control is an essential part of the electric energy transmission and distribution system to

maintain proper voltage limit at the consumer’s terminal. Besides the generating units that

provide the basic voltage control, there are many additional voltage-controlling agents, but the

most popular one, among all those agents for controlling voltage levels at the distribution and

transmission system, is the on-load tap changer transformer. On load tap changing (OLTC)

transformer has become a vital link in modern power systems. It acts to maintain the load bus

voltage within its permissible limits despite any load changes. This paper presents the

implementation of the on-load transformer tap changer using two approaches on MATLAB

environment; the Fuzzy Logic system (FLC) and the Artificial Neural Network (ANN) approach.

The main objective is to reduce the voltage variations at the secondary side or the load side of the

transformer by determining the best tap location that will maintain the voltage in range close to

the nominal voltage. However the implementation has been done using both the Fuzzy logic

approach and the Artificial Neural Network approach separately and their results compared.
Table of Contents

LIST OF FIGURES:..........................................................................................................................................i
LIST OF TABLES:...........................................................................................................................................ii
ABBREVIATIONS..........................................................................................................................................iii
CHAPTER ONE:........................................................................................................................................1
INTRODUCTION.......................................................................................................................................1
Damaged Lines....................................................................................................................................2
Iron losses............................................................................................................................................2
Load changes.......................................................................................................................................2
BACKGROUND.....................................................................................................................................3
Artificial Neural Network....................................................................................................................5
Fuzzy Logic Controller (FLC) Approach............................................................................................5
OBJECTIVES.........................................................................................................................................6
MAIN OBJECTIVE............................................................................................................................6
SPECIFIC OBJECTIVES...............................................................................................................................6
RELEVANCE OF THE PROJECT.........................................................................................................7
THE SCOPE OF THE PROJECT............................................................................................................8
CHAPTER TWO.........................................................................................................................................9
LITERATURE REVIEW:.......................................................................................................................9
THE ANT COLONY OPTIMIZATION (ACO)..................................................................................9
THE PARTICLE SWARM OPTIMIZATION (PSO).......................................................................10
METHODOLOGY................................................................................................................................13
INTRODUCTION.............................................................................................................................13
THE FUZZY LOGIC TECHNIQUE.................................................................................................13
THE ARTIFICIAL NEURAL NETWORK TECHNIQUE...............................................................21
Figure 3.1.3.1:.......................................................................................................................................23
CHAPTER FOUR.....................................................................................................................................25
EXPECTED RESULTS............................................................................................................................25
LIST OF FIGURES: Pg.

Figure 1.21: On-load tap changer……………………………………………………………10.

Figure 1.22: Fuzzy logic system……………………………………………………………..12.

Figure 3.0: Fuzzy logic approach flow chart……………………………………………….25.

Figure 3.1.3.0: Artificial Neural Network flow chart……………………………………….26.

Figure 3.1.3.1: Expected ANN network…………………………………………………….27.

LIST OF TABLES: Pg

Table 3.0: Rules of tap location……………………………………………………………...22

ABBREVIATIONS:

PI: Proportional integrator.

OLTC: On-load Tap Changer.

ACO: Ant Colony Optimization.

ANN: Artificial Neural Network.

FLS: Fuzzy Logic system.

FLC: Fuzzy Logic Controller.

PSO: Particle Swarm Optimization.

CHAPTER ONE:

INTRODUCTION.

As the population of the human beings grows the gap between power supply and power demand

widens. The imbalance or variation in power availability sent the researchers in to the field to

study and come up with many ways of settling this age long problem. For many years the gap

between the power supply and power demand has been reducing due to many available sources

of energy kept in place such as the renewable sources of energy and the wide electrical power

network. However there is still a problem of imbalance and variation in this available power

especially from the electrical machines such as the generators and transformers. Considering

such electrical machines, as the transformers, which are rated and designed by the manufacturers

to output a given known level of voltage, current or power. However, this is not the real case in

the field because these machines will always output a voltage, current or power slightly higher or

lower than the rated value. Variations are part and parcel of the transmission process. These

variations are caused by environmental factors such as: damaged lines, iron losses, load changes

among many other factors.

Damaged Lines

Electrical cables and power lines do experience damage to their structure due to falling trees, and

short circuit along the power line. Strong winds may be the cause of the short circuit as a result

of swinging. This would cause a considerable variation in electrical parameters of the line.
Iron losses

These are the losses that are experienced in the core of the transformer due to the electrical and

magnetic properties of the iron. The losses are due to hysteresis effect and eddy currents in the

core. Therefore a considerable power loss is experienced, and consequently leading to the

voltage variation at the secondary side.

Load changes

Various load types are placed at the secondary side of the transformer. These loads vary in

requirement of both real and reactive power, together with the voltage magnitude and phase

angle in the case of three phase.

Due to the variation of voltage caused due to the factor highlighted above the load may

experience an unstable power supply and voltage levels. These variations are not a good as far as

proper functioning of some equipment is concerned. Since it is crystal clear that some

institutions such as hospitals, other health related institutions or delicate systems should not be

allowed to suffer from this problem of power variation. Optimization of transformer taps is one

of the most vital ways of reducing voltage or power variations. This can be achieved by

involving the use of MATLAB software in relation to Artificial Neural Network and Fuzzy

Logic method to come up with the best optimized transformer taps in order to reduce voltage

variations especially at the transformer output. This optimization may also be achieved by using

other methods such as Ant Colony Optimization or Proportional Integrator (PI) controller

method.
BACKGROUND.

Optimization basically refers to the determination of the best working conditions of any

equipment or tool. So when we talk about transformer tap optimization, we are basically

referring to the most suitable location of the transformer taps on the secondary windings to

output voltage that is of minimized variations. Transformer tap changing is one of the methods

that are used to control voltage. This operation works on the principal of regulating the voltage at

the secondary based on change of the number of turns on the primary or the secondary side of the

transformer. When the number of the turns on the primary side of the transformer is increased,

the e.m.f per turn increases hence an increase in the output voltage at the secondary. The taps of

these transformers are arranged in 2.5 % steps in order to obtain the rated value of the secondary

voltage in case the primary voltage is about 0.2.5, 5, and 7.5 % away from the nominal voltage

rating of the primary winding. As a result the secondary side voltage is regulated in relation to

tap position. The diagram below shows a model an on-load tap changer:

Figure 1.21, on-load Tap Changer (OLTC)

There are two types of tap changing, the on-load and the off-load tap changer. The on-load tap

changer is able to regulate the voltage levels of the power system while the transformer is

delivering power to the load, while an off-load tap changer changes the taps only when the

transformer is disconnected. Though it is the less in cost since the amount of insulation required

is minimal.

Optimization of transformer taps is a project that basically focuses on the best allocation of the

position of the transformer taps in order to ensure voltage stability and reducing power loss.

Optimization of tap changer in transformers has been done through several approaches such as;

the Ant Colony Optimization (ACO), the PI controller method among others. However this

project focuses entirely on the Fuzzy Logic approach and the Artificial Neural Network, both

being tools in Matlab software.

Artificial Neural Network

Artificial neural networks (ANNs) have the potential advantage over conventional techniques in

significantly improving the performance of the other techniques. The ANNs have been

recognized as useful tools for the identification and control of the system with different

nonlinearities. This is so by virtue of the fact that ANNs have the capability of non-linear

mapping, parallel processing and learning; these attributes make them ideally suited for power

system control application. In the power systems ANNs have been successfully used for short

term load forecasting, power system control, fault detection, harmonic propagation, and so on.

The needs for faster and automotives, which can be used online and in real time application is

raised. The artificial neural networks (ANN) are an ideal choice, given the ability to cover the
nonlinearity and its fast response time. Neural network considered to be the most promising area

in artificial intelligence as it is based on human experiences and on link of the input and output

sets, learning or training concepts and a pattern recognition function. The neural network adopts

various learning mechanisms and self-organization. They have been successfully applied to

problems in the fields of pattern recognition, image processing, data compression, forecasting,

and optimization.

Fuzzy Logic Controller (FLC) Approach

Theory of fuzzy sets and fuzzy logic are widely used as a methodological basis of solving

problems in the case of uncertain and incomplete original data, and the lack of reliable

mathematical description of system behavior fuzzy logic system (FLS) can be defined as the

nonlinear mapping of an input data set to a scalar output data. A FLS consists of four main parts:

fuzzifier, rules, inference engine, and defuzzifier.

Figure 1.22,A Fuzzy Logic System

OBJECTIVES

MAIN OBJECTIVE

The main objective of this project is to optimize the transformer taps in order to reduce the

voltage variation at the output.

SPECIFIC OBJECTIVES.
The analysis of this project cannot go without paying attention to the goals meant to be achieved

by this project. These objectives include:

i. Implementation of on-load transformer taps changer using fuzzy logic and

artificial neural network (ANN).

ii. Comparing results obtained using FLC and ANN.

iii. To match the load voltage required by the load to the voltage at the secondary

side of the transformer.

iv. To enhance voltage stability in a given power system.

RELEVANCE OF THE PROJECT

Electrical and Electronic equipments are designed to operate on specified voltage, power rating

and frequency. Deviation from the specified values poses serious threats to the entire power

system.

Too wide variation of voltage may cause erratic operation or even malfunctioning of consumer’s

appliances. The main cause responsible for voltage variation is the variation in load on the

supply system.

Domestic, industrial or offices have delicate appliances that need to operate under less or no

voltage variations. Apart from many ways of ensuring voltage stability in a transmission system,

optimizing the tap changers in transformers is one of the ways of restoring voltage stability on

the consumer side (transformer secondary side). Voltage variations beyond which a given system

is rated to operate may lead to overheating and deterioration of system. Consumers are satisfied

with quality and reliable electric power when the above mentioned factors are eliminated. This

work considers the application of tap-changing transformer to minimize distribution system

losses. The use of tap-changing transformer has the advantage of being able to regulate the

voltage at a bus. With this approach, the appropriate tap settings required to compensate for the

voltage drops in the distribution system are determined and hence, the voltage variations are

minimized.

In this project we shall implement a simple fuzzy logic controller (FLC) and artificial neural

network (ANN) to select the best (optimum) tap location to reduce voltage variation.
THE SCOPE OF THE PROJECT

This project is aimed at providing a permanent solution to voltage variation at the consumer or

transformer output. Solution provision will entail the optimization of the transformer taps using

the Fuzzy Logic approach and the Artificial Neural Network approach on the MATLAB

software or environment. This project covers aspects such as the features, applications and

operation of the tap changers in transformers and the transformer taps in relation to voltage

variation.

The implementation of the Fuzzy logic approach and Artificial Neural Network approach on

MATLAB software will involve the checks the transformer parameters: voltage and voltage error

in relation to transformer tap position. However both the Fuzzy logic and the Artificial Neural

Network methods are implemented separately and then the results are compared. The results

obtained after implementation of the FLC are used in training the Artificial Neural Network.

This optimization procedure shall be carried out through the fuzzy logic and Artificial Neural

Network in Matlab software.

CHAPTER TWO

LITERATURE REVIEW:

Apart from the Fuzzy logic optimization method, there are a variety of methods that have been

made use of to come up with the tap optimization of a transformer. Among these methods we

have; the ant colony optimization (ACO) and the proportional integrator (PI) logic. These

methods are described as follows.

THE ANT COLONY OPTIMIZATION (ACO)

This optimization technique was developed by the observation of the behavior of ants. How the

ants traverse from their nest to look for food and return back to the nest using the shortest route

possible. In the ant colony optimization (ACO), a colony of artificial ants cooperates in finding

good solutions to difficult optimization problems. Cooperation is key in the design component of

ACO algorithms: The choice is to allocate the computational resources to a set of relatively

simple agents (artificial ants) that communicate indirectly by stigmergy. Good solutions are an

emergent property of the agent’s cooperative interaction. This method was invented by Marco

Dorigo in 1992. It uses iterative mechanism which is meta-heuristic scholastic combinatorial

computational based with keen reference to a real ant colony behavior. The ants’ population is

used as the agents who come up with candidate solutions. The entire process is guided

probabilistically through the heuristic imitation of real ants’ behavior which is linked to the
characteristics of a given engineering problem that is to be solved.An ant k, when located at node

i, uses the pheromone trail (τ ij ¿)to compute the probability of choosing j as the next node i.e.
❑

{ }
α
τ ij (k )
if j∈ N i
(k)
Pij =
∑τ k
α
ij
j ∈N i
(k)
0 if j ∈ N i
¿

The advantages of this optimization technique entail;

i. It requires a lesser amount of time as compared to other types of optimizations.

ii. It improves the voltage regulation better than any other technique.

The disadvantages of the ACO method are as follows;

i. The need to carry out numerous iterations may not necessarily bring out the best

optimized tap location compared to fuzzy logic controller.

THE PARTICLE SWARM OPTIMIZATION (PSO)

The PSO has been proposed by Eberhart and Kennedy [24] to simulate the motion of bird

swarms. The particle swarm process is stochastic in nature; it uses a velocity vector to update the

current position of each particle in the swarm. The velocity vector is updated based on the

memory gained by each particle, conceptually resembling an autobiographical memory, as well

as the knowledge gained by the swarm as a whole. Thus, the position of each particle in the

swarm is updated based on the social behavior of the swarm which adapts to its environment by
returning to promising regions of the space previously discovered and searching for better

positions over time. Numerically, the position of the ith particle, Xi, at iteration (t + 1) is updated

as follows:

t +1 t t +1
X i = X i +V i

Where by the updated velocity vector is given as:

V ti +1=ῳ V ti +C 1 r 1 ( Pti −X ti ) +C 2 r 2 ( Gtbest −X ti )

The advantages of the Particle Swarm Optimization (PSO) include:

i. PSO is based on the intelligence. It can be applied into both scientific research and

engineering use.

ii. PSO have no overlapping and mutation calculation. The search can be carried out by the

speed of the particle. During the development of several generations, only the most

optimist particle can transmit information onto the other particles, and the speed of the

researching is very fast.

iii. The calculation in PSO is very simple. Compared with the other developing calculations,

it occupies the bigger optimization ability and it can be completed easily.

iv. PSO adopts the real number code, and it is decided directly by the solution. The number

of the dimension is equal to the constant of the solution.

The disadvantages of the Particle Swarm Optimization approach include:

i. The method easily suffers from the partial optimism, which causes the less exact at the

regulation of its speed and the direction.

 ii. The method cannot work out the problems of scattering and optimization.

iii. The method cannot work out the problems of non-coordinate system, such as the solution

to the energy field and the moving rules of the particles in the energy field.
 CHAPTER THREE

METHODOLOGY

INTRODUCTION

This project was achieved using two different approaches, the Fuzzy logic approach and the

Artificial Neural Network on the MATLAB software.

THE FUZZY LOGIC TECHNIQUE

This fuzzy logic technique basically works on a given set of input parameters known as crisp

inputs. These inputs are then converted into a format that can be understood by the fuzzy logic

controller i.e. they are fuzzified. After this process the result is led to the inference base where

they are applied unto certain set rules. These rules are in the format of IF THEN statements.

Where antecedents (those that precede the (IF) infer a given specific consequences (after THEN).

Deffuzification is the process where the fuzzy parameters are then converted back into normal

parameters that initiate a given command following the process.

Fuzzy logic is a rule-based method for soft computing that can be used to model the uncertainty

of crisp-valued objective data, as well as for the modeling of subjective linguistic variables. A

Fuzzy Logic System (FLS) is unique in its ability to concurrently consider multiple input design

constraints of different types, such as objective and subjective, and of different units, such as

length, voltage or current. The Fuzzy logic approach is grouped into three main components: the

member function, the fuzzification and the defuzzification. A key component of Fuzzy logic

modeling is the fuzzification of crisp input design constraints. Fuzzification itself is the act of

mapping a crisp input constraint to a membership function (MF), also called a fuzzy utility
curve. The Member Functions are used by the Fuzzy Logic System (FLS) to model uncertainty

of the input data. Taken the current state of optimization techniques, it is believed that further

improvement in optimization may be possible by introducing the modeling capabilities of Fuzzy

Logic.

A Fuzzy Logic Controller (FLC) performs certain nonlinear Mapping of its inputs into its

outputs. The designing procedure of a FLC consists of a sequence of the following steps:

1. Defining the linguistic variables and terms (input and output signals.

2. Construction of the membership function.

3. Construction of the rule base and collecting the set of control rules governing the FLC.

4. Fuzzification; converting crisp input data into fuzzy values using membership functions

or defining the fuzzy sets describing the linguistic terms.

5. Inference; evaluating rules in the rule base and combining the results in each rule base.

6. Selecting a method for fuzzy logic computations and Deffuzification.

In our case, the crisp inputs to the fuzzy logic controller shall be the load voltage error (e) and

current tap position. These parameters shall then be fuzzified, i.e. made into membership

functions so that the fuzzy controller can understand. The output shall be the direction of the tap

position and the time delay which shall be made into member functions as well. The parameters

shall then be tested in the rule base to come up with the appropriate output to initiate the optimal

tap location.
INPUTS AND OUTPUTS TO THE FLC

As far as this paper is concerned, for optimization of the transformer taps using Fuzzy logic

controller two inputs and two outputs were considered:

The inputs are;

i. Voltage error (V error )

ii. Original tap position (n)

The outputs are:

i. Direction of the tap position (n•)

Voltage error

A transformer is a machine that is used to step up /down voltage and current to suit the power

requirement of the consumers. These transformers are normally rated to receive and deliver fixed

values of current and voltages, however these machines will always output a voltage, current or

power slightly higher or lower than the rated value. The range of output voltage (V O) variation

from the rated voltage¿ ¿) is called Voltage error (V error ).

V error =¿ ¿ ¿)

Tap position

The voltage control in transformers is performed by changing the turn’s ratio. This is done by

provision of taps in the winding. Irrespective of the end use for which tapping is put to; taps are

provided on the HV winding. If we consider the tap position on the primary side of the

transformer winding as (n) in relation to the number of turns, that is the value of (n) will keep on
varying and the number of turns on the secondary side of the transformer is ( N s ).Therefore

keeping the relationship between the voltage ratio and the turn ratio in the transformer;

n Vp
( ) =( )
Ns Vs

Vs = ( Vn ) N
p
s

If N s is considered to be fixed,to keep the value of the secondary voltage (V s ) constant when the

primary voltage varies, either (V p) or (n) must be adjusted. Since the objective is to keep (V s )

back to nominal, then (n) must be adjusted considerably.

Membership function

Implementation of membership function is vital in the fuzzification and defuzzification steps of a

FLS, to evaluate the non-fuzzy input values to fuzzy linguistic terms and vice-versa. A

membership function is implemented to measure the linguistic term. For instance,

membershipfunctions for the linguistic terms of voltage error are plotted.

As gathered in Tables 3.0 and Table 3.1, the control algorithm is defined by 15 control rules for

the direction of tap change and another 15 rules for the time delay to be applied at each control

step.
 VOLTAGE ERROR (e)

DIRECTION

OF TAP HNEG NEG ZERO POS HPOS

POSITION (^n)

HIGH UP UP UC DOWN DOWN

TAP

LOCATION

(n) NOR UP UP UC DOWN DOWN

LOW UP UP UC DOWN DOWN

Table 3.0 Rules for tap location

RULES

The following are the rules that are to be adhered to in the rule base of the fuzzy logic inference

section. The criteria for formulating these rules is the use of the IF THEN statements. This

criterion was founded by Mamdani, 1977; and they are fifteen of them as follows:
1. IF the voltage error is HNEG AND the tap location is high THEN move tap position

UP.

2. IF voltage error is NEG AND tap location is high THEN move tap position UP.

3. IF the voltage error is ZERO AND tap location is high THEN tap position is UC.

4. IF the voltage error is POS AND the tap location is high THEN move tap position

DOWN.

5. IF the voltage error is HPOS AND the tap location is high THEN move tap position

DOWN.

6. IF the voltage error is HNEG AND the tap location is normal THEN move tap position

UP.

7. IF the voltage error is NEG AND the tap location is normal THEN move tap position

UP.

8. IF the voltage error is ZERO AND the tap location is normal THEN the tap position is

UC.

9. IF the voltage error is POS AND the tap location is normal THEN move the tap position

DOWN.

10. IF the voltage error is HPOS AND the tap location is normal THEN move the tap

position DOWN.

11. IF the voltage error is HNEG AND the tap location is low THEN move the tap position

UP.

12. IF the voltage error is NEG AND the tap location is low THEN move tap position UP.

13. IF the voltage error is ZERO AND the tap location is low THEN the tap position is UC.
 14. IF the voltage error is POS AND the tap location is low THEN move tap position

DOWN.

15. IF the voltage error is HPOS AND the tap location is low THEN move tap position

DOWN.

The figure below indicates the general approach of the fuzzy logic:
 INPUTS; Voltage error (e) and Current tap
position (n)
 OUTPUTS: Direction of the tap position .
 Fuzzification of crisp inputs and output
 Determine number of fuzzy partitions
 Choosing number of fuzzy partitions
 Determination of fuzzy control rule base
 Definition of inference engine
 Choosing of deffuzification method
PARAMETERS TUNING
FUZZY LOGIC CONTROLLER OPERATION
1. Mapping of membership functions.
1. Fuzzification
2. Fuzzy inference rules. 2. Fuzzy inference
3. Defuzzification
3. Scaling factors

SIMULATION AND TESTING

OK

END

Figure3.0 Fuzzy logic approach flow chart

THE ARTIFICIAL NEURAL NETWORK TECHNIQUE

A neural network is a massively parallel distributed processor made up of simple processing

units that have a natural tendency for storing experiential knowledge and making it available for
us. Artificial neural network (ANN) is a type of Artificial Intelligence technique that mimics the

behavior of the human brain. ANNs have the ability to model linear and non-linear systems

without the need to make assumptions implicitly as in most traditional statistical approaches .The

Neural Network Toolbox is one of the commonly used, powerful, commercially available

software tools for the development and design of neural networks. The software is user-friendly,

permits flexibility and convenience in interfacing with other toolboxes in the same environment

to develop a full application. Neural computing requires a number of neurons, to be connected

together into a neural network, neurons are arranged in layers. Each neuron within the network is

usually a simple processing unit which takes one or more inputs and produces an output. At each

neuron, every input has an associated weight which modifies the strength of each input. The

neuron simply adds together all the inputs and calculates an output to be passed on.

The activation function is generally non-linear. Linear functions are limited because the output is

simply proportional to the input. The other available activation functions include: symmetric

Hard limit, Log-sigmoid,Tan-sigmoid,Radial basis and satlin transfer function.The figure below

indicates the general structure:

Designing ANN models follows a number of systemic procedures. In general, there are five

basics steps:

i. Collecting data.

ii. Preprocessing data.

iii. Building the network

iv. Train, and

v. Test performance of model

The flow chart below indicates how these steps are followed in the implementation of the

Artificial Neural Network (ANN).

 DATA COLLECTION

DATA
PREPROCESSING

BUILDING THE
NETWORK

TRAIN THE
NETWORK

TEST FOR PERFORMANCE

Figure 3.1.3.0: The Artificial Neural Network flow chat.

The Artificial Neural Network that we shall implement shall be made up of the following inputs;

voltage error (e) and the current tap location (tc) and one output. The results obtained from the

fuzzy logic are the ones that will be used to test the ANN. The ANN basically is made of 2

inputs, one intermediate layer and one output.

The diagram below shows the connection of the ANN:

Hidden layer
Inputs

Voltage error

Output

New tap
location

Current tap
location

Figure 3.1.3.1: The expected ANN network

The five major steps followed in the implementation of the ANN is as follows;

Data collection

Collecting and preparing sample data is the first step in designing ANN models. A set of

examples for training the network is assembled. Each case consists of a problem statement

(which represents the input into the network) and the corresponding solution (which represents

the desired output from the network). The input data is entered into the network via the input

layer.In the case of this project these data was collected from the results obtained from the fuzzy

logic implementation, both the inputs and the output.

Data pre-processing

After data collection, three data preprocessing procedures are conducted to train the ANNs more

efficiently. These procedures are: solve the problem of missing data, normalize data and

randomize data. Normalization procedure before presenting the input data to the network is

generally a good practice, since mixing variables with large magnitudes and small magnitudes

will confuse the learning algorithm on the importance of each variable and may force it to finally

reject the variable with the smaller magnitude. Each neuron in the network processes the input

data with the resultant values steadily "percolating" through the network, layer by layer, until a

result is generated by the output layer.

Building the network

At this stage, the designer specifies the number of hidden layers, neurons in each layer, transfer

function in each layer, training function, weight/bias learning function, and performance

function. In this work, two inputs were selected; the voltage error (Verror)and the current

transformer tap position(n).However the network contains only one hidden layer that has a

number of neurons and a single output (n•).

Training the network

During the training process, the weights are adjusted in order to make the actual outputs

(predicated) close to the target (measured) outputs of the network. The actual output of the

network is compared to expected output for that particular input. This results in an error value..

The connection weights in the network are gradually adjusted, working backwards from the

output layer, through the hidden layer, and to the input layer, until the correct output is produced.
Fine tuning the weights in this way has the effect of teaching the network how to produce the

correct output for a particular input, i.e. the network learns.

Testing the network

The next step is to test the performance of the developed model. At this stage unseen data are

exposed to the model.

CHAPTER FOUR

EXPECTED RESULTS

The main objective of this project was to optimize the on-load transformer taps so that the

voltage variations are reduced. Therefore at the end of this project it is expected that the voltage

variations at the output of the transformer are reduced to values less than 10% of the nominal

voltage. This indicates that for transformers that are rated to output a voltage of 240 V, then the

voltage variations are expected to range between -24 to +24 V.

However the results obtained through the Fuzzy logic system (FLS) implementation are expected

to be very close to the results obtained via the Artificial Neural Network (ANN) implementation.
FINAL YEAR PROJECT BUDGET

The table below contains the quantity and the approximated cost of the requirements of the

project. The total cost amount to Ksh.15, 500.

REQUIREMENTS QUANTITY COST (Ksh)

1 MATLAB/ Simulink software 1 3,000

2 Field Study 3 4,000

3 Typing and printing # 2,500

4 Internet # 6,000

TOTALS

Ksh.15,500
THE WORK PLAN AND SCHEDULE

The table below is the work plan and schedule towards the realization of this project:

MONTH

ACTIVITIES JUNE JULY AUG SEPT OCT NOV DEC

Topic for the

project

Concept note

presentation

Literature

review

Proposal

drafting
Mini-

presentation

Proposal

defense

Software

purchasing

Simulation

process

Simulation

testing

Results

analysis

Final

documentation
REFERENCES:

1. Yee Ming Chena and Wen-Shiang Wang A particle swarm approach to solve

environmental/economic dispatch problem.

2. Huynh Thi Thanh Binh Hybrid Particle Swarm Optimization For Solving Multi -Area

Economic Dispatch Problem.

3. P.R.Sujin, Dr.T.Ruban Deva Prakash and M.Mary Linda, Particle Swarm Optimization

Based Reactive Power Optimization, Journal Of Computing, Volume 2, Issue 1, January

2010, Issue No. 2151-9617.

4. Harinder Pal Singh, Yadwinder Singh Brar, D. P. Kothari Combined Active and Reactive

Power Dispatch Using Particle Swarm Optimization.

5. M. T. Hagan, H. B. Demuth, and M. Beale, Neural Network Design,PWS Publishing Co.,

Boston, 1995.
6. P. Okanik, B. Kurth, J.H. Harlow, “An update on the Paralleling of OLTC Powers

Transformers,” Transmission and Distribution Conference, 1999 IEEE, Volume: 2, pp.

871 –875, 1999.

7. Ye Li; Nair, N.C.; Sing-Kiong Nguang; , "Improved coordinated control of On-load Tap

Changers," Universities Power Engineering Conference (AUPEC), 2010 20th

Australasian ,vol., no., pp.1-6, 5-8 Dec. 2010.