基于信号稀疏表示和瞬态冲击信号多特征提取的滚动轴承故障诊断  

Fault Diagnosis of Rolling Bearing Based on Sparse Representation of Signals and Transient Impulse Signal Multifeature Extraction

作　　者：孟宗[1] 殷娜 李晶[1] MENG Zong;YIN Na;LI Jing(Key Laboratory of Measurement Technology and Instrumentation of Hebei Province,Yanshan University,Qinhuangdao,Hebei 066004,China)

机构地区：[1]燕山大学河北省测试计量技术及仪器重点实验室

出　　处：《计量学报》2019年第5期855-860,共6页Acta Metrologica Sinica

基　　金：国家自然科学基金(51575472,61873226,61873227);河北省高等学校科学研究计划重点项目(ZD2015049);河北省留学人员科技活动项目择优资助(C2015005020)

摘　　要：在滚动轴承故障信号特征分析中,针对瞬态冲击信号稀疏表示和特征提取问题,提出一种基于IChirplet原子的故障信号多重特征提取方法。在分析故障信号特点的基础上,构建IChirplet原子库,利用优化的OMP算法进行原子寻优,然后提取IChirplet原子的时频参数和重构信号的敏感特征作为特征参量,通过PSO_SVM实现故障分类。实验证明IChirplet原子与滚动轴承故障信号有较好的匹配性,且多重特征的提取能够有效表征故障信息,更准确地判断轴承故障类型。A multi-feature extraction method based on IChirplet atom is proposed to solve the problem of sparse representation and feature extraction of transient impact signal. According to the characteristics of the fault signal,a dictionary which is made up of IChirplet atoms is built and a improved pariticle swarm optimization algorithm is presented for searching the best atoms. And then sensitive parameters in time-frequency domain of the IChirplet atomics and characteristics of the reconstructed signal are extracted as the characteristic parameter. Finally,the fault classification is achieved by PSO_SVM. The experiment proves that the IChirplet atom can match the fault signal of rolling bearing well,and the multiple characteristics can better reflect fault information and the bearing failure type can be judged more accurately.

关 键 词：计量学 滚动轴承 故障诊断 稀疏分解 IChirplet原子 多重特征提取 

分 类 号：TB936[一般工业技术—计量学] TB973[机械工程—测试计量技术及仪器]