基于LMD鄄ICA降噪的滚动轴承故障特征提取方法-北京邮电大学学报.PDF

2017年2月 北 京 邮 电 大 学 学 报 Feb.2017 第40卷 第1期 Journal of Beijing University of Posts and Telecommunications Vol.40 No.1 摇 摇 文章编号:1007鄄5321(2017)01鄄0111鄄06 DOI:10.13190/ j.jbupt.2017.01.020 基于LMD鄄ICA降噪的滚动轴承故障特征提取方法研究 许同乐,摇 王营博,摇 郑店坤,摇 陈摇 康,摇 刘同义 (山东理工大学 机械工程学院,山东 淄博255049) 摘要:在滚动轴承进行故障识别中,针对局部均值分解(LMD)方法分析非平稳、非线性含噪信号时,存在端点效 应,易产生虚假分量和单通道独立成分分析(ICA)盲源分离时的欠定问题,提出了基于LMD鄄ICA 降噪的振动信号 特征提取算法. 首先对原始信号进行LMD,并抑制端点效应,得到n个瞬时频率具有物理意义的乘积函数(PF)之 和;然后对得到的PF分量以连续的3 阶PF分量为一序列组合进行ICA,可以得到n-2个重构分量;最后利用n- 2个分量进行重构,得到降噪后的故障信号,并再次进行LMD或功率谱计算,提取故障特征. 经验证,该方法可有 效识别滚动轴承的多类故障. 关摇 键摇 词:局部均值分解;独立成分分析;故障特征提取;故障诊断 中图分类号:TH133郾33;TH165郾3摇 摇 摇 摇 文献标志码:A Research of the Rolling Bearing Fault Signal Feature Extraction Method Based on the LMD鄄ICA Noise Reduction XUTong鄄le,摇 WANG Ying鄄bo,摇 ZHENG Dian鄄kun,摇 CHEN Kang,摇 LIUTong鄄yi (Mechanical Engineering School,Shandong University of Technology,Shandong Zibo255049,China) Abstract:Aiming at the problem of existing the end effect,easily producingfalse component. The local mean decomposition (LMD)wasusedin analysisof non鄄stationary andnonlinear signalscontainingnoise and the underdetermined problem of the single channel independent component analysis (ICA) blind source separation,the feature extraction algorithm of the vibration signals based on the noise reduction LMD鄄ICAwasproposed. By applying the algorithm,thefault of rolling bearing isdiagnosed,firstly,the original signals are decomposed by LMD,the n instantaneous frequency product function (PF) compo鄄 nentswith physical significance are obtained and the end effect islimited. Then the obtained PF compo鄄 nents which are arranged into one sequence combinationper constant third鄄order PFcomponentsare