An accurate fault classification algorithm using a minimal radial basis function neural network

Distance relaying techniques based on the measurement of impedance at the fundamental frequency between the fault location and the relaying point have attracted widespread attention. The sampled voltage and current data at the relaying point are used to classify the type of fault involving the line with or without fault resistance present in the fault path. The accuracy of the fault classification also depends on the amplitude of the dc offset and harmonics in comparison to the fundamental component. Fourier transforms, Differential equations, Waveform modelling, and Kalman filters are some of the techniques used for wellestablished fault detection and location calculations. In recent times due to strong pattern recognition capabilities, ANNs (Artificial Neural Networks) are considered a viable alternative to the conventional fault classification approaches. Several ANN architectures have been considered [1-7 ] for the classification of faults on a power transmission line from the sampled voltage and current signals at the relaying point. Amongst these the multilayered perceptrons with backpropagation learning strategy and the radial basis function neural network present very attractive solutions in classifying faults and estimating the fault distances accurately. The BP algorithm, however, has a serious drawback of overlearning and thus can produce erroneous result when the fault data falls outside the trained patterns. Similarly the conventional RBF network [5] considered for distance relaying has the demerits of selecting the number of hidden units on a trial and error approach and determination of parameters of the Gaussian function by K-means clustering algorithm. Further a large An accurate fault classification algorithm using a minimal radial basis function neural network