Abstract: For the traditional method, the fault feature extraction effect of rolling bearings is still limited, and the parameters of the least squares support vector machine classifier are not easy to determine, which reduces the accuracy of fault diagnosis. Least squares support vector machine based on the energy moment of the eigenmode function and the Bayesian framework is proposed for fault diagnosis of rolling bearings. In this method, the original signal is decomposed into multiple eigenmode functions by empirical mode decomposition, and then the integral of the time axis is used as the eigenmode function to obtain the energy-moment feature fault vector of the intrinsic mode function. Bayesian inference method is used to perform three-level hierarchical inference and the parameters of the least squares support vector machine classifier have arbitrariness and uncertainty, and the parameters are optimized. The simulation results of the rolling bearing show that this method can effectively and accurately diagnose the fault, and the diag-nostic accuracy rate reaches 98.75%.

Key words: fault diagnosis, empirical mode decomposition, energy moment of eigenmode function, Bayesian, least squares support vector machine, parameter optimization

摘要： 针对传统方法对滚动轴承故障特征提取效果尚有局限和最小二乘支持向量机分类器的参数不易确定，从而降低了故障诊断的准确性的问题，提出基于本征模函数能量矩和贝叶斯框架下的最小二乘支持向量机实现滚动轴承的故障诊断。在该方法中，通过经验模态分解将原始信号分解为多个本征模函数，之后将本征模函数作时间轴的积分，得到本征模函数能量矩特征故障向量。采用贝叶斯推理方法进行三级分层推断，解决最小二乘支持向量机分类器的参数具有任意性和不确定性的问题，实现参数优化。对滚动轴承的仿真结果表明，该方法能对故障进行有效、准确的诊断，诊断正确率达到98.75%。

关键词: 故障诊断, 经验模态分解, 本征模函数能量矩, 贝叶斯, 最小二乘支持向量机, 参数优化