A REFINED GENETIC ALGORITHM FOR THE FAULT SECTIONS LOCATION

The fault sections location in distribution networks based on the information of the short circuit current may get false results due to the signalsdistortion. So the genetic algorithm (GA) is proposed for its high fault-tolerance performance. But the construction of the fitness function is a main bottleneck for the application of GA in the fault sections location. In this paper, a new mathematical model of the fault sections identification is presented and a Refined Genetic Algorithm (RGA) using the model as the fitness function is developed. The new model includes the following two charac-ters: ① it improves the conventional fitness function so as to avoid the false results of the identification and its fault-tolerance ability is stronger than the conventional GA ② it is able to locate multiple fault sections in multi-source distri-bution networks. The RGA in the paper is more suitable for complex fault diagnosis in distribution networks and the simulation results have verified its correctness.