基于小波包信息熵和小波神经网络的异步电机故障诊断

采用一种基于小波包信息熵和小波神经网络的方法对异步电机进行故障诊断。将故障信号进行小波包预处理,并在此基础上提取信号的小波包能谱熵和小波包系数熵,构成信号的信息熵特征向量。训练小波神经网络使其在输入特征向量后能有效检测并输出故障模式,以实现对单一故障和复合故障的诊断。通过内嵌的方式把小波变换融入神经网络,具有良好的自适应分辨率和容错能力,可以有效避免局部最小值以及收敛速度过于缓慢的问题。试验表明,基于小波包信息熵和小波神经网络的方法能很好地进行异步电机的故障诊断,且该方法优于同参数下的BP神经网络模型。

Fault diagnosis of asynchronous motor based on wavelet packet entropy and wavelet neural network

A method based on wavelet packet entropy and wavelet neural network was presented for asynchronous motor to realize fault diagnosis. The signal with faulty information was pretreated by wavelet packet, the wavelet packet energy spectrum entropy and coefficient entropy was extracted. The feature vector of information entropy was constructed. When the feature vector was input into the wavelet neural network, we trained it to detect and output the fault mode, so as to realize the fault diagnosis. This method had good adaptive resolution and fault tolerance, and it could avoid local minima and slow convergence effectively. The experiment results showed that this method could be used for fault diagnosis of induction motors, which was better than BP neural network model with the same parameters.