基于模糊积分融合技术的变压器故障识别

Power Transformer Fault Diagnosis Based on Fuzzy Integral Fusion

提出了一种基于模糊积分融合技术的变压器故障诊断新方法。首先建立三比值的隶属函数并对三比值数据进行模糊预处理,再用4个不同的径向基函数神经网络进行故障识别,最后利用模糊积分融合方法将4组诊断结果进行融合处理,得出诊断结果。仿真实例表明这种方法比单个人工神经网络具有更高的故障诊断性能。

A new method to diagnose power transformer based on fuzzy integral fusion is presented. Firstly, the membership function of the value of three ratio is presented and the data of value of three ratio are processed by fuzzy method, then the faults are identified by using four different radial basis function neural network, finally foru results of diagnosis from four different neural network are fused by fuzzy integral fusion and form the final diagnosis. The simulation results of presented method prove to be more effective than single neural network in fault diagnosis.