Selection of currents patterns using SVMs for locating faults in radial power systems

The current regulatory framework demands better quality service to electric utilities, using as a measure of non scheduled power outages the SAIDI and SAIFI indexes. In this paper, a selection of characteristics from protective relays using a set of support vector machines (SVM) is proposed in order to solve the fault location problem in power distribution systems. To show the effectiveness of the proposed methodology, a large scale power distribution system was used. Results show lower validation errors which are associated to the high performance. Opportune fault location in power distribution systems plays an important role in the service improvement.