基于多层训练干扰的CNN轴承故障诊断  

Fault Diagnosis Based on CNN with Multi-layer Training Interference for Bearings

作　　者：王连云 陶洪峰 徐琛[1,2] 杨慧中 WANG Lian-yun;TAO Hong-feng;XU Chen;YANG Hui-zhong(Key Laboratory of Advanced Process Control for Light Industry,Ministry of Education,Jiangnan University,Wuxi 214122,China;School of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China)

机构地区：[1]江南大学教育部轻工过程先进控制重点实验室,江苏无锡214122 [2]江南大学物联网工程学院,江苏无锡214122

出　　处：《控制工程》2022年第9期1652-1657,共6页Control Engineering of China

基　　金：国家自然科学基金资助项目(61773181);直升机传动技术国防科技重点实验室开放课题(HTL-O-22G05)。

摘　　要：针对滚动轴承故障诊断方法在低信噪比情况下抗噪性能差以及在不同工况下自适应性不足的问题,提出了一种基于多层训练干扰的卷积神经网络滚动轴承故障诊断方法。首先,通过快速傅里叶变换将原始信号从时域变换到频域并作为卷积神经网络(CNN)的输入,再对多层卷积层进行实时训练干扰,以不同程度随机破坏网络训练,提高模型的抗噪性能和域自适应能力;同时引入批量归一化(BN)方法,改进卷积神经网络结构,改善模型性能。实验结果表明,所提方法不仅在噪声干扰下能达到较高的诊断准确率,同时也能有效地解决工况变化问题。Aiming at the problem that the fault diagnosis method for rolling bearings has poor anti-noise performance under low signal-to-noise ratio(SNR) and insufficient self-adaptability under different working conditions, a fault diagnosis method based on convolutional neural network(CNN) with multi-layer training interference for rolling bearings is proposed. Firstly, the original signal is transformed from time domain to frequency domain by fast Fourier transform(FFT) and used as the input of the CNN. Then, the real-time training interference is performed on multiple convolutional layers to randomly destroy the network training to different degrees, so as to improve the anti-noise performance and domain adaptive ability of the model.Meanwhile, the batch normalization(BN) method is introduced to improve the structure of the CNN and the model performance. The experimental results show that the proposed method achieves high diagnostic accuracy under noise interference and solves the problem of working condition change.

关 键 词：卷积神经网络 故障诊断 抗噪声 变工况 滚动轴承 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]