Fault Classification and Location Identification in a Smart Distribution Network Using ANN

This paper presents a novel approach to classify and locate different types of faults in a smart distribution network (DN). The proposed method is able to classify all types of faults that can occur in a DN and then based on fault type, it can identify the approximate fault location (FL) with a high accuracy. The method is based on artificial neural networks pattern recognition which uses data from mu PMUs/smart meters placed at different locations in a DN. The proposed technique needs fault-on voltages of all the nodes connected to the end of line/branches in order to classify and locate different types of faults. The method is tested on a modified IEEE-37 bus system with distributed generation along with dynamic loading conditions and varying fault resistances. Both balanced and unbalanced fault types are applied to the system. An accurate classification of 100% is achieved when classifying all fault types and above 99% accuracy is achieved when identifying the approximate fault location.