针对滚动轴承振动信号易受环境噪声干扰及浅层学习模型依赖人工经验难以准确提取故障特征的难题，提出了一种优化自适应白噪声平均总体经验模态分解(OCEEMDAN)与卷积神经网络(CNN)联合的故障诊断方法。采用自适应白噪声平均总体经验模态分解(CEEMDAN)算法对原始信号进行分解，分形维数筛选最佳分量，奇异值(SVD)降噪优化，输入CNN实现故障诊断，分别与EMDCNN、EEMDCNN及CEEMDANCNN方法进行对比。结果表明：该方法在不同工况下均具有较高的识别率，突显了良好的鲁棒性与泛化性。

Aiming at the problems that vibration signals of rolling bearings are easily interfered by environmental noise and the shallow learning model relying on artificial experience is difficult to extract fault features accurately,a fault diagnosis method based on the combination of optimal complete ensemble empirical mode decomposition with adaptive noise (OCEEMDAN) and convolutional neural network (CNN) was proposed.CEEMDAN algorithm was used to decompose the original signal,the fractal dimension was used to screen the optimal component,the singular value decomposition (SVD) noise reduction was optimized,and the fault diagnosis was achieved by input CNN.The EMDCNN,EEMDCNN and CEEMDANCNN methods were compared respectively.The results show that the proposed method has a high recognition rate under different working conditions,which highlights the good robustness and generalization.

TH133

国家自然科学基金（51976131，52006148）；上海市“科技创新行动计划”地方院校能力建设项目（19060502200）