基于模糊神经网络的电力变压器故障诊断

Fault Diagnosis Method for Power Transformer Based on Fuzzy Neural Network

针对传统的故障诊断技术具有模型复杂,数据采集方法比较繁琐,模糊规则难以确定等问题,提出了将模糊逻辑和神经网络技术相结合的方法;首先针对变压器的故障的特点,利用模糊理论描述不确定性信息,然后采用TOPSIS方法通过对DGA的特征量数据进行模糊处理从而实现优劣的排序;最后根据优化的数据输入到BP网络中,对不同的故障状态进行识别和诊断;同时采用无线网络进行数据传输,实现了远程数据采集与故障诊断功能;分析表明,这种方法能够实现对不同故障的诊断,有效地提高了故障模式的识别能力,较传统诊断方法有较大的优势。

According to the problems of traditional fault diagnosis technique, which are the complex model, tedious data acquisition method and the fuzzy rules are difficult to be determined, put forward a method of combining fuzzy logic and neural network technology. Firstly in the light of the characteristics of transformer fault, by using fuzzy theory describe uncertainty information, then through to the TOPSIS method to fuzzy processed the DGA characteristic data so as to realize the quality sorting. Finally, according to the optimized data input to BP network, to distinguish and diagnosis the different fault state. At the same time using the wireless network to transmission data, realize the remote data acquisition and fault diagnosis function. Analysis shows that this method can realize diagnosis to different breakdown, improve the efficiency of the fault mode recognition ability, have a greater advantage than traditional diagnostic methods.