【摘要】 针对滚动轴承故障诊断过程中，存在提取故障特征困难、识别故障准确率低以及速度慢等问题，提出了基于改进的CEEMDAN与贝叶斯算法优化支持向量机（BO-SVM）相结合的滚动轴承故障诊断方法。首先利用ICEEMDAN对原始振动信号进行分解，得到若干的模态函数分量（IMF），采用相关系数法筛选有用的IMF分量重构信号，将重构信号的多尺度排列熵作为特征向量输入到BO-SVM故障诊断模型进行训练和测试。研究结果表明：采用该方法能够有效地提取特征信息，ICEEMDAN-BO-SVM故障诊断模型可以实现对滚动轴承快速、准确地诊断，诊断时间为21.26 s，准确率达到了99.38%，与网格搜索法（GS）、遗传算法（GA）优化的SVM模型相比，该方法的诊断时间和准确率具有一定的优越性。更多还原

【Abstract】 In view of the difficulties in extracting fault features,low accuracy of fault identification and slow speed in the pro-cess of rolling bearing fault diagnosis,a rolling bearing fault diagnosis method based on the combination of improved CEEMDANand Bayesian optimized support vector machine(BO-SVM)is proposed. Firstly,ICEEMDAN is used to decompose the original vi-bration signal to obtain a number of intrinsic mode functions(IMF). The correlation coefficient method is used to screen the usefulIMF component reconstruction signal,and the multi-scale permutation entropy of the reconstructed signal is input as the feature vec-tor to the BO-SVM fault diagnosis model for training and testing. The research results show that this method can effectively extractfeature information,and ICEEMDAN-BO-SVM fault diagnosis model can realize rapid and accurate diagnosis of rolling bearings.The diagnosis time is 21.26 s,and the accuracy rate reaches 99.38%. Compared with the SVM model optimized by grid search meth-od(GS)and genetic algorithm(GA),this method has certain advantages in diagnosis time and accuracy rate.更多还原