Fault Classification and Location for Distribution Generation Using Artificial Neural Networks

Abstract

With the proliferation of distributed generation (DG), the distributed network had become more complex. Such complexity will lead to difficulty for fault location in the distributed network. It may degrade the precision of existing fault location methods. Therefore, this paper will investigate the impact of distributed generation toward machine learning (ML) based fault location. Three categories of fault location had been tested which is fault type prediction, fault section prediction, and fault distance prediction with and without DG presence. The accuracy of machine learning based fault location is verified in IEEE 16 bus network and the impact due to the presence of DG, represented using photovoltaic (PV) generator is discussed in detail.