TRANSMISSION LINE FAULT CLASSIFICATION USING ARTIFICIAL NEURAL NETWORK BASED FAULT CLASSIFIER

Transmission line systems are the indispensable link between consumers and generating stations. The intrusion caused to the consumer because of transmission line faults is massive or consequential. A proper and secure protection scheme is mandatory for uninterrupted power supply to the consumer end. This study comes up with a new development of Artificial Neural Network based Fault Classifier (ANNFC) using Discrete Wavelet Transform (DWT); for classification of distinctive faults on three phase transmission line. The proposed ANNFC is trained using discrete sets of data achieved from wavelet analysis taking Db8 as mother wavelet and addition of fifth level detail coefficients of fault transients both for relay terminal and far end terminal for fault classification. The back propagation algorithm along with feed forward neural network is used and a detailed analysis with two hidden layer, one input and one output layer has been performed in order to justify the choice of neural network. The feasibility of the proposed algorithm of fault classification is tested on a 11 KV, 100 MVA, 100 Km long transmission line under various possible fault types and fault impedances in MATLAB environment. The result shows enhanced accuracy and efficient in identifying L-G, L – L, L – L – G, L – L – L faults. Furthermore the performance of ANNFC is compared with the algorithm based fault classifier (AFC) and the results shows that AFC is able to classify the fault with an accuracy of 98% but the proposed ANNFC algorithm, can classify the fault with an efficiency of almost 100%.