Morlet小波变换结合MLP神经网络的齿轮箱故障检测方法

针对变速齿轮箱中的故障检测问题，提出了一种结合Morlet小波变换和多层感知器(MLP)神经网络的齿轮故障检测方法。利用角域技术，将时域中齿轮故障的非平稳振动信号转化为角域中的平稳信号。然后，利用进行Morlet小波变换并从小波系数中提取统计特征。同时根据最大能量与香农熵比来确定连续小波变换(CWT)的最优尺度，以此来缩减特征量，并将小波系数的能量和香农熵作为两个新特征添加到特征向量。最后，利用MLP神经网络对输入特征进行分类，从而检测故障。实验结果表明，该方法故障检测准确率高，且计算速度快。

A gearbox fault detecting method based on Morlet wavelet transform and MLP neural network

For the issues that the fault detection of gearbox, a method of gearbox fault diagnosis base on Morlet wavelet transform and MLP neural network is proposed. The angular domain technique is used to transform the non-stationary vibration signals of the gear fault in time domain into stationary signal in the angular domain. Then, Morlet wavelet is used to extract statistical features from wavelet coefficients, and determines the optimal scale of continuous wavelet transform (CWT) according to the maximum energy and Shannon entropy ratio, in order to reduce the amount of features, at the same time, the energy of the wavelet coefficients and Shannon entropy be regarded as two new features added to the feature vector. Finally, MLP neural network is used to classify the input features, so as to diagnose fault types. The experimental results show that this method has high accuracy of fault diagnosis and high calculation speed.