Three phase transmission lines fault identification and location using multilayer artificial neural network

Abstract

Transmission lines are very important components of electric power infrastructure that connect generating stations and the load centres located over large geographical areas. Consequently, identification of the type of fault and their respective location is crucial to guarantee adequate protection of the power system lines. Various fault identification and classification techniques have been developed by different researchers. This paper is based on fault classification and location using artificial neural networks (ANN). A multilayer Feed-forward network was employed along with a back-propagation algorithm for each of the three phases in the Fault diagnostic process. The 245km Gombe - Yola 330kV transmission line was modeled and various fault types were simulated to obtain data for training. Analysis of neural networks with varying numbers of hidden layers and neurons per hidden layer has been carried out to validate the choice of the neural networks architecture and topology. Simulation results revealed that the faults detector has correctly differentiated between normal and faulty conditions on all the test data samples. Similarly, the classifier has achieved over 95% accuracy in classifying the fault type. More specifically, the location of all the five main categories of line faults; Line-Ground (L-G), Line-Line(L-L), Double Line-Ground (L-L-G), Three phase fault(L-L-L) and Three phase-Ground(L-L-L-G) were determined with a marginal error of between 0,02518%(62m) to 2.5% (6.125 km) overall minimum and maximum values respectively. This demonstrates the effectiveness of the proposed ANN based method towards achieving satisfactory transmission network fault diagnosis. As a possible extension to this work, a comparative study using other forms of neural network schemes such as RBF, SVM and ANFIS should be carried out so as to further ascertain the effectiveness of the proposed method.