Algorithm for fault finding in electrical distribution systems using support vector machines

The aim of this work is to propose a system that allows the detection, classification and localisation of faults in a distribution power system with radial topology using a single measurement node. The proposed method is based on the use of Support Vector Machines (SVMs). The fault analysis algorithm consists of a multiclass classifier based on PSB to classify the type of fault and a second stage, also based on PSB, to locate the line in the power system where the fault occurs. The characteristic parameters are obtained by computing the RMS profile and deriving a two-level wavelet transform. A data mining process is applied to train and develop the classifier in order to select the characteristic parameters that best describe the disturbances in voltage and current oscillograms when a fault occurs in any grid line. One of the main advantages of the proposed algorithm is that the measurements are performed on a single network node, which is an important characteristic when calculating the cost and deployment time of a fault location system.