A comparative study of different signal processing techniques for fault location on transmission lines using hybrid Generalized Regression Neural Network

Travelling wave based fault location uses arrival times of the fault generated travelling waves at the terminals of the line. These arrival times are extracted by using different signal processing tools like Discrete wavelet transform (DWT), S-Transform (ST). The accuracy of fault location is highly influenced by the uncertainties in arrival time measurements. In this paper, an artificial neural network based fault locator is used to deal with such uncertainties. Based on the analysis of the arrival times estimated by different methods, a Generalized Regression Neural Network (GRNN) is trained for locating faults accurately. The performance of the compared technique is verified in MATLAB/Simulink environment by simulating different transmission line faults. Results obtained indicate the validity of Knowledge based solution using GRNN for fault location.