基于PSO-VMD与贝叶斯网络的滚动轴承故障诊断  

Fault diagnosis of rolling bearing based on PSO-VMD and Bayesian network

作　　者：仝兆景[1] 芦彤 秦紫霓 TONG Zhaojing;LU Tong;QIN Zini(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454000,Henan,China)

机构地区：[1]河南理工大学电气工程与自动化学院,河南焦作454000

出　　处：《河南理工大学学报（自然科学版）》2021年第1期95-104,共10页Journal of Henan Polytechnic University（Natural Science）

基　　金：国家自然科学基金资助项目(U1504623);河南省高校基本科研业务费专项资金资助项目(NSFRF1615);国家安全监管总局关键技术科技项目(Henan-0008-2017AQ);河南省高等学校矿山信息化重点学科开放实验室开放基金资助项目(KY2015-07)。

摘　　要：为解决电机在变负载运行条件下滚动轴承振动信号故障的特征提取困难、故障诊断准确率低的问题,提出一种基于变步长粒子群的变分模态分解与贝叶斯网络相结合的滚动轴承故障诊断模型。通过变步长粒子群算法优化的变分模态分解与Hilbert变换,提取故障信息并离散化处理,构建贝叶斯网络故障诊断模型,对滚动轴承故障发生概率推理,并利用完备、不完备数据集以及噪声试验验证该方法的准确性。仿真结果表明,该方法能高效提取特征信息,实现对不确定信息的推理估计,提高滚动轴承故障诊断的准确率,在滚动轴承的故障诊断预测中具有较好的理论与应用前景。In order to solve the problems of difficulty in feature extraction and low fault recognition rate under variable load running condition of motor,a rolling bearing fault diagnosis model based on combination of variable mode decomposition function optimized by variable step size particle swarm and Bayesian networks was put forward.The fault information was extracted and discretized by the variational mode decomposition optimized by the variable step size particle swarm optimization algorithm and Hilbert transform.And then,a Bayesian networks diagnosis model with fault types was constructed,which inferred the probability of rolling bearing fault occurrence.The complete data set,the incomplete data set,and the noise experiments were used to verify the accuracy of the method.The simulation results showed that the proposed method could effectively extract feature information and predict estimation of uncertain information.It improved the accuracy of rolling bearing fault diagnosis,and was very promising in the fault diagnosis and prediction of rolling bearings.

关 键 词：变分模态分解 粒子群优化 贝叶斯网络 滚动轴承 故障诊断 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]