Fault Detection and Location for Reinforcement of Smart Distribution Systems Restoration, Using Discrete Orthogonal Stockwell Transform and Regression ANN

Abstract

Distribution system plays an important role in supplying consumers with electric energy. Due to their structural complexity, these systems have always had difficulties in locating faults for faster restoration. Restoration in distribution systems includes these steps: Fault Location, Isolation, and Service Restoration (FLISR) using manual and automatic switches, which is essential to reduce interruption duration thus to improve system reliability. Therefore, one of the first and the most important steps of reliability improvement is timely locating the fault, thus reinforcing the restoration (FLISR) process. In this article, a fault detection and location method is proposed, using frequency spectrum analysis by Discrete Orthogonal Stockwell Transform (DOST) to extract fault characteristics from the fault current waveform at a certain place of measurement, followed by fault location estimation by regression ANNs. The proposed method first identifies all the possible fault locations using DOST coefficients. Then, the actual fault location is estimated, using the knowledge from distribution system status. The proposed method is tested on IEEE 16-bus distribution test system, revealing that the method is capable to locate various fault types, in a fast and accurate manner, and easily implementable. The effect of the following parameters on the accuracy of the proposed method is analytically discussed: fault location distance from the main bus, faults type, fault inception angle, fault resistance and load variation, where the first four parameters negligibly affect, while the last two might significantly affect the accuracy.