Distribution Network Feeder Fault Location Based On Data

Automatic positioning of distribution network fault is the prerequisite and basis of fault isolation and power restoration,and fault feeder sections’ positioning accurately can ensure the effectiveness of the fault isolation and the extend of blackout scope.The run-time distribution network’s topology is complicated and changeable, FTU of feeder terminal prones to information distortion or loss due to bad environment, so the distribution network fault location method must have general applicability and high fault tolerance. Therefore, fast, versatile and high fault tolerance is three basic characteristics of fault location methods to achieve real-time on-line fault location.In this paper the development, current situation and the problem existing in the distribution network fault isolation will be stated firstly. As the matrix method has no versatility and tolerance when fault loction high, an improvement for it is put forword as well as a new proposed method for fault location fit for its evaluation function. The simulation results showed that the improved matrix method is quite fit for the complicated topology multi-power distribution network location, though lack of switch information from the FTU or the information becomes distort, it still has a high accuracy and tolerance in fault location.In consideration of the problem of parameters sensitive, charge overflow and the particles are esay to trap into a local optimum inelectromagnetic algorithm, a method of particle component exchange search is brought in to improve the particle power, vector and movement step formula. And according to the flexible features in distribution network operation, the distribution network can be separated into several parts and every part intervals location model is built to improve the diagnose algorithm operation efficiency. The simulation results showed that the IEM algorithm has only a quite small scale of population and iteration in realizing the processing of FTU data and accurate fault location in distribution network in a high speed compared with standard EM alorithm.In order to improve the tolerance in IEM algorithm for fault location, combinationwith Bayesian Method will have a good effect. Firstly the expectaion swith fault information vector will be approached to switch fault information vector form FTU through the evaluation function to seek yhe fault intervals, and then the fault intervals vector probability will be calculated by the Bayesian method. At last the biggest probability fault intervals vector will be selected as yhe best fault interval vectors. The simulation results showed that the combination of IEM algorithm and Bayesian method has an obvious advantage in distribution network fault locatoin in high speed, versatility and tolerance.