Fault Section Diagnosis with High Fault-Tolerance Performance for Distribution Networks Based on the Combination of Neural Logic Network Redundant Error Correct and FNN
Sun Yaming Du Hongwei Liao Zhiwei(Tianjin University 300072 China)
The principle and realization method of fault secti on diagnosis for distribution networks, based on the combination of multiple-valu ed neural logic network (MNLN) redundant error correct and feedfoward neural net works (FNN) are creatively presented. According to the redundant character of in formation which is collected by SCADA system of distribution networks and redund ant relation among information of Feeder Terminal Units (FTUs) which can acquire d from the correlativity of topology structure of distribution networks. The mod el and inference rules of information for redundant error correct based on the t heory of MNLN are proposed. Therefore, the error-corrected disposed information without distortion form the input vector of FNN model to be used as fault secti on diagnosis. In this paper, the proposed principle and method have widely gener ality for distribution networks, have high fault-tolerance performance and impo rtant practical worth.