Fault location in distribution systems using mathematical analysis and support vector machine / Sophi Shilpa Gururajapathy

Distribution systems are continuously exposed to fault occurrences due to various reasons, such as lightning strike, failure of power system components due to aging of equipment and human error. These phenomena affect the system reliability and results in expensive repairs, damaged work in process,...

Full description

Saved in:
Bibliographic Details
Main Author: Sophi Shilpa, Gururajapathy
Format: Thesis
Published: 2017
Subjects:
Online Access:http://studentsrepo.um.edu.my/7446/1/All.pdf
http://studentsrepo.um.edu.my/7446/6/sophi.pdf
http://studentsrepo.um.edu.my/7446/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Distribution systems are continuously exposed to fault occurrences due to various reasons, such as lightning strike, failure of power system components due to aging of equipment and human error. These phenomena affect the system reliability and results in expensive repairs, damaged work in process, lost productivity and power loss to customers. Due to this, various intelligent methods have been developed to locate fault in distribution system. However, fault location using intelligent methods is challenging since it requires training data for processing. The training data is commonly created by simulation, which is time consuming. Therefore, in this work, a fault location method based on previous work is proposed using limited simulation data. The existing method was improved by estimating voltage sag data using support vector machine, thus limiting the simulated data. Faulty section is identified by comparing the actual voltage sag data with the simulated and estimated voltage sag data. An improved ranking and Euclidean distance approach for fault distance is also presented. A method using SVM is also proposed to identify the faulty phase, fault type, faulty section and fault distance. By having these features, a more accurate and effective fault location can be obtained. The method identifies faulty phase and fault type using support vector classification analysis. Meanwhile, the faulty section and the fault distance are identified using support vector regression analysis. The effectiveness of the proposed method was tested on an actual TNB distribution network from Malaysia and SaskPower distribution network from Canada. The test cases were conducted for all types of fault and for various fault resistances. The test results have proven the effectiveness of the proposed method in locating fault under various conditions. It has shown improvement over the existing trigonometric methods in locating different types of faults and may serve as an alternative technique for estimating fault location in distribution networks.