一种异步电动机故障诊断新方法

A new type of fault diagnosis method of asynchronous motor

针对基于RBF神经网络的异步电动机故障诊断方法存在参数确定较困难的问题,提出了一种基于差分进化算法优化RBF神经网络的异步电动机故障诊断方法。首先采用小波变换对异步电动机运行状态信号进行消噪处理,然后采用主元分析法与小波包分析法相结合方式提取消噪后的异步电动机运行状态信号特征,最后采用差分进化算法优化后的RBF神经网络对异步电动机运行状态信号特征进行诊断。实验结果表明,与未优化的RBF神经网络相比,采用差分进化算法优化后的RBF神经网络可有效识别出异步电动机故障。

In view of problem of difficult parameters determination existed in fault diagnosis method of asynchronous motor based on RBF neural network, the paper proposed a fault diagnosis method of asynchronous motor based on RBF neural network optimized by differential evolution algorithm. Firstly, the method uses wavelet transformation to make de-noising process for running state signal of asynchronous motor, then it uses principle component analysis method and wavelet packet analysis method to extract feature of the de-noised signal of asynchronous motor, and at last it uses RBF neural network optimized by differential evolution algorithm to diagnose the feature. The experiment result shows that the RBF neural network optimized by the differential evolution algorithm can identify faults of asynchronous motor effectively compared with RBF neural network without optimization.