Optimized support vector machine & wavelet transform for distribution grid fault location

The precise knowledge of fault locations is very important as longer period interruption caused by faults are responsible for most of the customer minute losses in the distribution grid. The objective of this paper is to locate different types of faults in a distribution grid by extracting the features of three phase fault currents employing wavelet transform (WT) and feeding the statistical measures of the extracted features as inputs of optimized support vector machine (SVM). Backtracking search algorithm is employed to optimize the parameters of the SVM. Finally, the results of the proposed model are compared with the results of a general/non-optimized support vector machine to validate the efficacy of the proposed model.