基于模糊关系与自组织竞争网络的变压器故障诊断

Transformer Fault Diagnosis Based on Fuzzy Relations and Self-Organizing Competitive Network

在当前变压器故障诊断研究现状基础上,综合变压器故障的传统诊断方法和人工智能网络诊断方法 ,构建了基于自组织竞争网络模型的电力变压器故障诊断方法。为实现变压器故障的实时在线诊断和高可靠性诊断提供了一种新的可行性思路。同时在针对变压器故障状态等级划分与特征参数提取,及故障状态数据模糊性和故障状态信息不完全所导致的不确定性问题,采用模糊理论中隶属度函数来对故障特征数据进行处理,得到电力变压器故障类型与故障数据的模糊关系。从模糊关系中提取出故障特征向量输入自组织竞争网络进行诊断分类,其提高了自组织竞争网络的故障分类性能和对故障数据的泛化能力。

On the basis of the current status of research on the transformer fault diagnosis, traditional diagnostic methods and artificial intelligence network fault diagnosis. This article constructes power transformer fault diagnosis method of self-organizing network model based on competition. To achieve high reliability and real-time online diagnosis transformer fault diagnosis a new feasible idea is provided. Also for transformer fault classification and feature parameter extraction and fault data in fuzzy and fault status information incomplete due to the uncertainty, fuzzy membership function theory is used to deal with the fault characteristic data for obtaining electricity fuzzy relations transformer fault type and fault data. The fault feature vector is extracted from the fuzzy relations to be diagnosed and classified by the self-organizing competitive network, it improves the performance of self-organizing fault classification competition network and fault data generalization.