DETECTION, IDENTIFICATION AND LOCATION OF FAULTS IN DISTRIBUTION SYSTEM

Abstract

Electrical power distribution network is an integral part of electrical power systems since it is the last stage in the delivery of electricity to customers. Analogous to humans’ circulatory system, if transmission system can be termed as the arteries of a human body, then the distribution system are the capillaries. The distribution network is responsible for distributing power to consumers at desired voltage levels with higher reliability. Alternating Current (AC) three-phase four-wire structure is the standard distribution system that exists throughout the world. With the growth in urban population and development of industries, distribution grids now consider a considerable amount of power. The large number of lines in a distribution system experience regular faults which lead to high values of line currents. A fault which occurs on a distribution network is defined as an irregular condition of circuit that results in energy being dissipated in a manner other than the serving of the intended load. As compared to transmission system lesser research works has been carried out for detection, identification, classification and location of faults in distribution system. The algorithms which have been developed for transmission system cannot be directly applied to distribution system because of certain constraints. Several techniques have been developed in order to get an errorless fault detection, identification and location. Traditionally, impedance based methods were developed but it suffered from the problem of multiple estimation. With the evolution of time, travelling based approaches also found its existences, but the measuring device required high sampling devices. With the introduction of artificial intelligence, there is a need for development of hybrid algorithm which not only detects, identify faults but also locate them accurately. Feature extraction is the basic need for development of protection algorithms using digital signal processing tools. It transforms data of high dimension to a lower dimension. But at the same time, the embedded information content is kept intact. Also the dimensionality of data is reduced. Further, the complexity for the purpose classification or regression is decreased. This chapter presents a brief concept of the tools used for feature extraction. It covers a brief overview of the different signal processing tools involved in the development of algorithm. Wavelet Transform, Wavelet Packet Transform, Gabor Transform, M – Band Wavelet Transform and Complex Dual v Tree Wavelet transform have been dealt. Also, it presents a brief overview of Artificial Neural Network meant for the classification and regression. The thesis gives an introduction of distribution system and the need for identification, classification and location of fault. Further, an extensive literature reviewed throughout this research work. During this the focus was on fault detection, identification and its location. Further, the detailed literature of signal processing tools used for feature extraction such as Wavelet Transform, Wavelet Packet Transform, Gabor Transform, M Band Wavelet Transform, Complex Dual Tree Wavelet Transform and the neural network employed for carrying out classification and location can be seen in next chapter. In the next work wavelet transform and wavelet packet transform has been used in order to collect the features. It has been extensively used by power system engineers for fault detection and location. However, it should be kept in mind that only approximations of the signal are decomposed to next level. It fails to accurately capture high frequency information in signals. Hence, in order to capture more information content in a signal, Wavelet Packet Transform is used which is an expansion of classical wavelet decomposition. It represents high frequency information in a better manner as compared to wavelet transform. A comparative result of wavelet and wavelet packet transform over different “daubechies” family“db1, db2, db4 and db8” has been presented. The result justifies the use of Wavelet Packet Transform which gives more accurate result than Wavelet Transform. Further, another technique based on the use of Gabor Transform for collecting features is presented in the next chapter. In Gabor Transform, one obtains the coefficients through high pass filter only. It gives detail information which is required to trap the sudden changes in the fault signal. In the present work, a level decomposition is used for fault classification and four levels decomposition have been used for fault location. A comparison is presented between Gabor Transform and Wavelet Packet Transform. Results obtained are very promising for Gabor Transform. In the subsequent work, M- Band Wavelet Transform has been used to extract the feature. M-band decomposition gives both logarithmic and linear frequency resolution. Further, its decomposition yields large number of sub bands which further helps more information about the signal. In MBand transform for one level decomposition one obtains the one low pass filter decomposition and two high pass filter decomposition. A comparison is presented between Gabor Transform and Mvi Band Wavelet Transform. Results obtained are very promising for both the databases in case of MBand Wavelet Transform. Another technique based on extraction of features of Dual Tree Complex Wavelet Transform is presented. The Dual-tree complex wavelet transform is used to overcome the two fundamental problems of wavelet transform as already discussed, while retaining the properties of nearly shift invariance and directionally selectivity. A comparison is presented between M- Band Wavelet Transform Dual Tree Complex Wavelet Transform. Results obtained for both the databases are very good that it proves that it is the best transform which give almost errorless result. At, the end an alternative solution to the problems associated with interruptions by means of a statistical modeling of current sample database applied to determine the fault location in power distribution systems in order to reduce the system restoration time. The current samples collected from the sample distribution systems are subjected to FCM to obtain clusters and fed to expectation maximization algorithm Eventually, the summary of work in the thesis is presented and also focuses on the scope for future work. An attempt has been made in the thesis to give an almost errorless fault location with the use of various digital signal processing tools and statistical based methods. It gives an edge over the conventional impedance based methods and also the problem of multi estimation has been successfully deal