基于优化变分模态分解与混沌分形融合的滚动轴承故障识别  

Fault Identification of Rolling Bearings Based on Optimized Variational Mode Decomposition and Chaotic Fractal

作　　者：孙康 金江涛 李春[1,2] 许子非 SUN Kang;JIN Jiangtao;LI Chun;XU Zifei(School of Energy and Power Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China;Shanghai Key Laboratory of Multiphase Flow and Heat Transfer in Power Engineering,Shanghai 200093,China)

机构地区：[1]上海理工大学能源与动力工程学院,上海200093 [2]上海市动力工程多相流动与传热重点实验室,上海200093

出　　处：《动力工程学报》2022年第10期951-959,985,共10页Journal of Chinese Society of Power Engineering

基　　金：国家自然科学基金资助项目(52006148,51976131,52106262);上海“科技创新行动计划”地方院校能力建设资助项目(19060502200)。

摘　　要：为精确识别滚动轴承故障类型,针对强非线性及非平稳性信号,分析其混沌特性,结合Lyapunov指数提出优化变分模态分解(OLVMD)方法,利用该方法实现降噪并选取敏感分量重构故障信号。引入分形理论,采用拟合偏差平方和方法对传统的关联维数计算方法进行改进,计算轴承不同状态下的混沌关联维数,并分析了损伤轴承实验数据。结果表明:OLVMD方法可有效剔除无关分量,消除冗余影响;不同状态轴承的关联维数具有显著差异,关联维数可作为轴承工作状态监测与诊断的依据,且该方法有良好的鲁棒性和泛化性。In order to accurately identify the fault types of rolling bearings, the chaos characteristics of strong nonlinear and non-stationary signals were analyzed. Combined with Lyapunov index, an optimized variational mode decomposition(OLVMD) method was proposed to realize noise reduction and select sensitive components to reconstruct fault signals. The fractal theory was introduced to improve the traditional method to calculate correlation dimension by using the square sum of fitting deviation. Then the chaotic correlation dimensions of bearings in different states were calculated by this improved method, and the experimental data of damaged bearings were analyzed. Results show that OLVMD method can effectively eliminate the irrelevant components and the redundant effects. The correlation dimensions of bearings in different states are significantly different, which can be used as the basis for monitoring and diagnosis of bearing working conditions, and the method has good robustness and generalization.

关 键 词：滚动轴承 故障诊断 混沌 LYAPUNOV指数 变分模态分解 分形 关联维数 拟合偏差平方和 

分 类 号：TH133.33[机械工程—机械制造及自动化]