基于AP-KSVD-OMP重构算法的电机轴承故障诊断  

Bearing Fault Diagnosis Based on AP-KSVD-OMP Reconstruction Algorithm

作　　者：杨敬娜[1] 冯贺平[1] 朱雪珣 郝克明 吴旭琳 YANG Jingna;FENG Heping;ZHU Xiaoxun;HAO Keming;WU Xulin(Department of Intlligent Engineering,Hebei Software College,Baoding Hebei 071000,China;College of Power Engineering,North China Electric Power University,Baoding Hebei 071000,China;Hwa Create Corporation,Beijing 100000,China)

机构地区：[1]河北软件职业技术学院智能工程系,河北保定071000 [2]华北电力大学动力工程学院,河北保定071000 [3]北京华力创通科技股份有限公司,北京100000

出　　处：《机械设计与研究》2022年第5期130-133,共4页Machine Design And Research

基　　金：河北省教育厅重点科技研究课题(ZD2018020)。

摘　　要：为增强轴承故障信号对于分块压缩感知的重构性能,设计了一种通过自适应分块方式来完成匹配追踪故障诊断算法。此算法综合运用信号分块自适应(Adaptive partitioning,AP)以及K奇异值分解(K-singular value decomposition,KSVD)方法,对正交匹配追踪(Orthogonal matching pursuit,OMP)算法实施了改进,进行重构时对支撑集原子通过二次筛选进行处理,获得更合理的支撑集原子结构,显著改善重构效果。研究结果表明:对轴承故障信号重构后AP-KSVD-OMP算法达到了最低重构相对误差和更短时间,可以保留OMP算法无需对信号稀疏度进行预估的特征,实现了信号重构精度的显著提升。对混合故障信号重构后可以达到对原始信号故障特征准确重现要求,与促使信号形成良好匹配状态。AP-KSVD-OMP算法具备高效运行的优点,实现了重构信号精度的显著提升。In order to enhance the reconstruction performance of bearing fault signals of block compressive sensing,an adaptive block approach is designed to complete the matching pursuit fault diagnosis algorithm.This algorithm integrates signal partitioning(adaptive partitioning,AP)and k-singular value decomposition(KSVD)methods.The orthogonal matching pursuit(OMP)algorithm is improved,and the support set atoms are processed by secondary screening during reconstruction to obtain more reasonable support set atomic structure and improve the reconstruction performance significantly.The results show that AP-KSVD-OMP algorithm achieves the lowest reconstruction relative error and shorter time for the reconstruction of bearing fault signals,and can retain the feature of OMP algorithm that does not need to predict signal sparsity,thus significantly improving the signal reconstruction accuracy.After the reconstruction of mixed fault signals,the requirement of accurate reappearance of fault features in the original signals can be achieved,and the signal can form a good matching state.AP-KSVD-OMP algorithm has the advantages of efficient operation,and achieves significant improvement of reconstructed signal accuracy.

关 键 词：压缩感知 自适应分块 重构算法 稀疏字典 

分 类 号：TH137[机械工程—机械制造及自动化]