Genetic Algorithm을 이용한 부분방전 패턴인식 최적화 연구

Abstract

Since insulation failures in service of GIS (Gas-Insulated Switchgear) cause fatal accidents, a reliable risk assessment technique is required. Meanwhile the domestic insulation diagnosis technique especially recognition of PD (Partial Discharge) pattern by use of the UHF (Ultra High Frequency) PD detection and analysis system has made rapid strides in recent years. Nevertheless, there are lacking in the published results of research work related to the optimization of neural network as a popular tool for recognition of PD pattern. Accordingly, in this paper, in order to optimize the structure of neural network, the genetic algorithm combined with neural network has been proposed for finding the best value of parameters of neural network. For the pattern learning, database was established by use of self-designed defects which are reported can be occurred and can be mostly critical in GIS as for the input of neural network. Database was analyzed to distinguish patterns by means of PRPD (Phase Resolved Partial Discharge) method and stored to the form with to unite the average amplitude of PD pulse and the number of PD pulse. There are two ways to be the most suitable value of parameters for trial-and-error method and genetic algorithm. First, the recognition success rate of defects was 93.2% and time consuming was very high by means of trial-and-error method. Second, the recognition success rate of defects was 100% by applying the genetic algorithm at neural network and it took a short time to find the best solution of parameters for optimization. Especially, it could be possible that the scrupulous parameters was obtained by genetic algorithm. Therefore, with these results, it can be considered that the genetic algorithm combined with neural network is the most promising method of a recognition of PD pattern.