基于LMD与改进SVM的轴承故障诊断方法

Bearing Fault Diagnosis Method Based on LMD and Improved SVM

针对轴承振动信号故障特征信息实际提取困难的问题,基于局部均值分解(LMD)与改进支持向量机(SVM)提出了轴承故障诊断方法。对所采集的轴承振动信号进行局部均值分解,得到若干乘积函数的分量。计算各乘积函数的能量,选取能量百分比值作为识别故障的特征值。针对支持向量机不能自适应选择核函数参数和惩罚因子的问题,利用细菌觅食优化算法对支持向量机进行参数优化。将特征值输入改进支持向量机模型,对轴承故障状态进行识别。试验结果表明,相对于传统支持向量机模型和隐马尔可夫模型,采用所提出的轴承故障诊断方法,对轴承故障的识别准确率提高7个百分点以上,由此验证了所提出的轴承故障诊断方法的可靠性。

Aiming at the issue of difficulty in extracting fault signature information of bearing vibration signal,a bearing fault diagnosis method was proposed based on LMD and improved SVM.LMD on the collected bearing vibration signals is conducted to obtain the components of several product functions.The energy of each product function is calculated and the energy percentage value is selected as the characteristic value for identifying the fault.Aiming at the problem that SVM cannot adaptively select the kernel function parameters and the penalty factor,the bacterial foraging optimization algorithm is used to optimize the parameters of SVM.The characteristic values are input to improved SVM model to identify the fault state of the bearing.The test results show that,compared with traditional SVM model and hidden Markov model,the proposed bearing fault diagnosis method can improve the identification accuracy of bearing fault by more than 7 percentage points,thus verifying the reliability of the proposed bearing fault diagnosis method.