一种改进的小波神经网络应用于电动机故障诊断

An Improved Wavelet Neural Network Used in Motor Fault Diagnosis

对基本的小波神经网络进行改进,用遗传算法取代传统的梯度下降法,对小波神经网络中的初始参数进行全局优化。将经过改进的小波网络应用于电动机的故障诊断,并对5组电动机故障数据进行验证。实验结果表明,该算法具有较快的收敛速度和较高的计算精度,证实了此方法应用于电动机故障诊断的正确性和有效性。

Global optimization to the initial parameters of the wavelet neural network is made by improving the basic wavelet neural network replacing the traditional gradient method with the genetic algorithms. The improved wavelet network is applied to the motor fault diagnosis, and five groups of motor fault are authenticated. The experiment shows fast convergence and high computation accuracy of the algorithm, and it confirm the correctness and effectiveness of the method used in motor fault diagnosis.