增强的最小相关广义Lp/Lq范数反卷积在旋转机械故障诊断中的应用  

Application of enhanced minimum correlated generalized Lp/Lq deconvolution in fault diagnosis of rotating machinery

作　　者：谭翠 黄晨光 易彩 周秋阳 林建辉[3] TAN Cui;HUANG Chenguang;YI Cai;ZHOU Qiuyang;LIN Jianhui(Weifang Institute of Technology,Weifang 261101,China;Weichai Power Company Ltd.,Weifang 261041,China;Traction Power State Key Laboratory,Southwest Jiaotong University,Chengdu 610031,China)

机构地区：[1]潍坊理工学院,山东潍坊261101 [2]潍柴动力股份有限公司,山东潍坊261041 [3]西南交通大学牵引动力国家重点实验室,四川成都610031

出　　处：《中国测试》2024年第10期157-166,共10页China Measurement & Test

基　　金：国家重点研发计划资助项目(52272355);四川省科技计划资助(2022NSFSC1918)。

摘　　要：用于提取滚动轴承故障周期冲击特征的最小相关广义Lp/Lq范数反卷积(MCG-Lp/Lq-D)存在对先验周期参数的精确度要求过高的问题。因此有必要提出一种新的周期估计方法-具有约束的自适应周期估计(APEC),该方法可以在强噪声条件下依然能够稳定的给出周期信息的真实估计或其近似值,随后将APEC引入到MCG-Lp/Lq-D构造了新的盲解卷积算法EMCG-Lp/Lq-D,该方法不仅能在强噪声条件下通过APEC自适应调整故障周期,同时还继承了MCG-Lp/Lq-D对噪声和异常值鲁棒的特点。由于APEC是一种需要先验周期(不要求精确周期)指导的周期估计方法,这使得其在低信噪比(SNR)条件下具有更好的鲁棒性来估计故障周期,从而使EMCG-Lp/Lq-D的运用范围相较于MCG-Lp/Lq-D的更广泛。最后,仿真和实验验证了EMCG-Lp/Lq-D的有效性和优越性。Minimum correlated generalized Lp/Lq deconvolution(MCG-Lp/Lq-D) used to extract the fault periodic impact characteristics of rolling bearings has the problem that its requirement for the accuracy of prior period parameter is too high.Therefore,a new period estimation method-Adaptive period estimation with constraint(APEC) is proposed.This method can still give the true estimation of period information or its approximate value stably under strong noise conditions.Subsequently,APEC is introduced into MCG-Lp/Lq-D to design a new blind deconvolution algorithm,EMCG-Lp/Lq-D.This method can not only adaptively adjust the fault period through APEC under strong noise conditions,but also inherits the robustness of MCG-Lp/Lq-D to noise and outliers.Since APEC is a period estimation method that requires a prior period(not requiring precise period) guidance,it has better robustness to estimate the real fault period under low SNR conditions,so that the application range of EMCG-Lp/Lq-D is wider than that of MCG-Lp/Lq-D.Finally,simulations and experiments verify the effectiveness and superiority of EMCG-Lp/Lq-D.

关 键 词：最大相关峭度反卷积 自适应故障周期估计 故障诊断 机器状态监测 局部故障 

分 类 号：TB9[一般工业技术—计量学] TH133.33[机械工程—测试计量技术及仪器] TN911.7[电子电信—通信与信息系统]