Fault Location and Classification in Power System

Chew, Zerahny Kia Yuan (2020) Fault Location and Classification in Power System. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Chew Kia Yuan Zerahny.pdf Restricted to Registered users only Download (5MB)

Abstract

Power transmission systems are the core of power transmission that enables the usage of electrical devices. It is essential to maintain the health of the power system to minimize power outages and to reduce the down time when faults occur in power systems. Simulation model of long transmission line was made for simulation and collection of fault signals. The voltage and current waveform of the fault signal holds the information of its fault type and its location. Using Discrete Wavelet Transform (DWT), the important characteristics of the fault type and its location can be extracted. There are various mother wavelets to be selected as well as various levels of decompositions to be considered. These extracted wavelet coefficients are then processed using mathematical operations. Principal Component Analysis (PCA) is then used to reduce the dimension of the data. This is done to reduce the training time of machine learning models. The resulting output was used as features to train machine learning models for location and classification of faults. Several models are considered in order to compare the performance and suitability of the models. Fault estimations are also carried out using the features extracted. The transmission lines can span over great lengths leading to unnecessary time wasted to find the fault. By relying on the fault estimation, the search area for the fault can be reduced concurrently reducing the time needed to locate the fault. The Feedforward Backpropagation Neural Network performed very well for fault classification having up to 100% accuracy. The same model was used for fault location and the accuracy obtained was 95.9%. Other machine learning models perform slightly poorer than neural network but had good accuracy for location and classification of fault. Fault estimation was also able to perform with acceptable overall estimation error of 3.54% when tested with fault signals at various points on the transmission line. The results obtained considered single-phase to ground, two-phase, two-phase to ground and three-phase to ground faults. The faults occurred at various fault inception angles. The faults also included low and high fault impedances. The results indicate that this approach managed to detect and locate the fault zone with good accuracy on a long transmission line model.