改进的小波神经网络在异步电机故障诊断中的应用

The application of the improved wavelet neural network in asynchronous motors fault diagnosis

根据异步电机的复杂故障特点,结合小波变换技术,提出了一种改进的小波神经网络用于异步电机的故障诊断。利用小波变换技术提取异步电机特征信号作为小波神经网络的输入向量,并对小波神经网络算法进行优化,提出了动量系数和学习率自适应调整的小波神经网络算法,给出了动量系数和学习率的调整方法。通过实际测试数据的诊断结果说明该方法的有效性和可行性,具有诊断准确率高、收敛速度快、泛化能力强等优点。

According to asynchronous motor＇s complex fault characteristics, and combining wavelet transform technique, an improved wavelet neural network for fault diagnosis of asynchronous motor was proposed in this paper. Taking wavelet transform technique as wavelet neural network the input vector of picking up asynchronous motor＇s the characteristic signal, and the wavelet neural network algorithm was optimized. The self- adaptive wavelet neural network algorithm about adjusting momentum vector and learning rate was proposed and given the momentum factor and the learning rate adjustment method. The actual testified results show that the method is effective and feasible,has a better diagnostic accuracy, fast and generalized performances.