Rough set theory for fault location on electrical distribution systems based on geographic information systems

It's reported that faults in electric distribution systems are responsible for 95% of shortages to customers. Accurate fault location which is the foundation and premise for fault diagnosis is significant for improving service quality of distribution systems. Various methods including expert system, ANN, fuzzy theory and Bayesian network have been proposed to settle fault location while each method has its own disadvantage. In this paper rough set theory is proposed for solving this problem on the basis of geographic information systems(GIS). We use topology rules based on GIS to analyze distribution network connectivity and network switches status through information derived from GIS database. In fact, customer trouble calls are presently the major source of information about abnormal network status, especially for distribution system without advanced supervise and remote control facilities. Hence, trouble calls are regard as the condition attribute of the information table for fault location. The process of fault location implementation based on rough set theory: initial decision table is established, thus reduction rules according to rough set theory can distill effective information to locate faults, the result of fault location will be flickered on the distribution system GIS. The simulation test demonstrates that rough set theory for fault location based on trouble calls can obtain the promising result. (5 pages)