基于多层降噪技术及改进卷积神经网络的滚动轴承故障诊断方法

Rolling Bearing Fault Diagnosis Method Based on Multilayer Noise Reduction Technology and Improved Convolutional Neural Network

针对滚动轴承微弱故障在强噪声下难以实现有效诊断的问题,提出了一种基于多层降噪技术及改进卷积神经网络(Improved convolution neural network,ICNN)的轴承故障诊断新方法。首先,对滚动轴承的一维振动信号进行预处理,得到标签化的数据样本,分为训练集和测试集;然后采用奇异值分解(Singularvalue decomposition,SVD)处理训练样本,通过二分之均值法选择有效奇异值个数,获得原始降噪信号和带噪信号;为了避免丢失微弱故障细节特征,将带噪信号经过SVD进一步去噪消除模态混叠并输入经验模态分解(Empirical mode decomposition,EMD)得到内禀模态函数,根据方差贡献率大小选出IMF分量并与原始降噪信号叠加得到最终信号;将处理后的训练集数据输入到引入注意力机制(Attention mechanism,AM)的ICNN中进行学习;最后将得到的诊断模型应用于测试集,输出故障类别诊断结果。通过滚动轴承故障诊断模拟试验,在强噪声环境下进行测试,结果表明所提方法能更准确的在强噪声环境中实现轴承的故障诊断。

Aiming at the problem that the weak fault characteristics of rolling bearings are difficult to achieve effective diagnosis under strong noise, a new method of bearing fault diagnosis based on multilayer noise reduction technology and improved improved convolutional neural network(ICNN) is proposed. First, the vibration signal of the rolling bearing is pre-processed to obtain labeled data samples, which are divided into a training set and a test set. Then singular value decomposition(SVD) is used to process the training samples, the number of valid singular values is selected by the singular value mean method to obtain the original noise-reduced signal and the noisy signal. In order to avoid losing the details of weak faults, the noisy signal is further denoised by SVD to eliminate modal aliasing and input empirical mode decomposition to obtain the intrinsic mode function(IMF). The IMF component is selected according to the variance contribution rate and superimpose it with the original noise reduction signal to obtain the final signal. The processed training set data is input into the improved ICNN that introduces attention mechanism for learning. Finally, the obtained diagnostic model is applied to the test set, the fault category diagnosis results are output. Through the rolling bearing fault diagnosis simulation test, the test is performed in a strong noise environment, the results show that the proposed method can more accurately realize the bearing fault diagnosis in a strong noise environment.