Fault classification on a power transmission line using discrete wavelet transform and artificial neural networks

Other Title(s)

تصنيف الأعطال الخاصة بخطوط نقل القوى الكهربائية باستخدام تحويل المويجة المتقطع و الشبكات العصبية الاصطناعية

Dissertant

Matar, Mustafa Ghassan

Thesis advisor

Muhammad, Umar R.

University

Princess Sumaya University for Technology

Faculty

King Abdullah II Faculty of Engineering

Department

Department of Electrical Engineering

University Country

Jordan

Degree

Master

Degree Date

2018

English Abstract

The main objective of this thesis is to utilize the techniques of Wavelet Transform and Artificial Neural Network (ANN) for classification of major types of faults on power transmission line on different locations.

Classifying the disturbances in power transmission lines have been challenging tasks for accurate and reliable protection of power transmission line.

Over a wide range of literature survey, there are many approaches that have been published with emphasis on different faults and system characteristics.

This thesis deals with the design of a Discrete Wavelet Transform (DWT) and Artificial Neural Network (ANN) based classification system for transmission line fault classification.

The output of the DWT has been used as trainer for the ANN that enhances the system performance, mainly, the accuracy of classification.

The accuracy of classification is crucial feature for practical classification system.

The performance of the proposed trained feed forward neural network has been promising and this offers the potential for implementation in a digital relay for transmission line protection.

The results of the proposed ANN methodology are found to be accurate under the conditions of different fault location.

Applying any form of Discrete Wavelet Transforms produces details coefficients, which are related to high frequency components, as well as approximations coefficients, which are related to low frequency components at each level of resolution.

In the case of applying the Wavelet transform, the details coefficients are down sampled by 2 and filtered with the same filter bank at each level of resolution.

A simulation setup is developed for collecting various phase currents waveforms for purposes of selecting the best mother Wavelet and the number of levels of resolutions.

The Daubechies (db5) Mother Wavelet with one level of resolution are found to be the best in providing adequate information to classify transmission lines faults.

The proposed algorithm is simulated using MATLAB/Simulink on a data collected from a typical transmission line.

The results of these tests show difference in its ability to classification between the Feedforward and Elman architecture in terms of accuracy under the conditions of different fault location.

The scope of the proposed approach is to classify series and shunt types of faults.

The series types of faults have been rarely considered in the literature and have been fairly investigated classified with along shunt faults in this thesis.

The main contribution of the thesis could be briefly outlined in two significant points: First, a comparative study has been conducted between the types of ANN architectures which are Elman and Feedforward ANN in terms of accuracy for classifying the fault.

Both architectures have been trained by DWT coefficients that plays an important role in classification procedure.

Second, the system is able to classify the shunt and series types of faults over a wind range of fault locations.

Main Subjects

Electronic engineering

Topics

• Electric power
• Neural networks(Computer science)

No. of Pages

37

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-890486