基于低秩约束的旋转机械内激励激发的振源信号分离研究  

Separation of vibration source signals excited by internal excitation of rotating machinery based on low rank constraint

作　　者：贺志洋 赵德尊 娄乐 程卫东[1] HE Zhiyang;ZHAO Dezun;LOU Le;CHENG Weidong(School of Mechanical,Electronic and Control Engineering,Beijing Jiaotong University,Beijing 100044,China;Department of Materials and Manufacturing,Beijing University of Technology,Beijing 100084,China;Beijing Institute of Computer Technology and Application,Beijing 100854,China)

机构地区：[1]北京交通大学机械与电子控制工程学院,北京100044 [2]北京工业大学材料与制造学部,北京100124 [3]北京计算机技术及应用研究所,北京100854

出　　处：《振动与冲击》2022年第17期152-159,共8页Journal of Vibration and Shock

基　　金：北京市自然科学基金资助项目(3192025);科技部国家重点研发计划(2016YFB1200601-B24;2016YFB1200602-26);国家自然科学基金(51905292);中国博士后科学基金(2019M660615;2020T130348)。

摘　　要：随着机械设备向自动化程度更高的方向发展,机械系统变得更加复杂,给基于振动的健康监测带来了巨大的挑战。振源信号分离对于机械系统的噪声控制、准确的状态监测及故障诊断具有重要意义。现有的方法中的关键之一是要已知振源信号的统计特性、源的个数等。但这些有效的特征在工程应用中很难获得。对旋转机械和往复式机械的工作特性进行了分析,发现内部激励激发的信号具有低秩性。并建立了低秩噪声分离模型,提出了一种多低秩约束的振源信号分离方法。仿真信号和试验信号验证了该方法的有效性。与基于独立分量分析的盲源分离(BSS-ICA)方法相比,该方法分离后的振源信号具有更好的分类效果和更低的噪声。With development of mechanical equipment to a higher level of automation,mechanical systems become more complex,and bring great challenges to vibration-based health monitoring.Vibration source signals separation is of great significance for noise control,accurate condition monitoring and fault diagnosis of mechanical system.One of keys of the existing methods is to know statistical characteristics of vibration source signals and number of sources.However,these effective features are difficult to obtain in engineering applications.Here,working characteristics of rotating machinery and reciprocating machinery were analyzed.It was found that signals excited by internal excitation have low rank.Then,a low rank noise separation model was established,and a multi-low rank constrained vibration source signals separation method was proposed.Simulated signals and test ones verified the effectiveness of the proposed method.Compared with the blind source separation method based on independent component analysis(BSS-ICA),separated vibration source signals obtained with the proposed method have better classification effect and lower noise.

关 键 词：故障诊断 低秩约束 振动信号 信号分离 

分 类 号：TP17[自动化与计算机技术—控制理论与控制工程] TP206[自动化与计算机技术—控制科学与工程] TH133.3[机械工程—机械制造及自动化]