唐贵基, 丁傲, 王晓龙,等.基于ARLMD和IMOMEDA的滚动轴承早期微弱故障诊断[J].河南理工大学学报（自然科学版）,2023,42(1):102-115.

TANG G J, DING A, WANG X L, et al.Incipient weak fault diagnosis for rolling bearing based on ARLMD and IMOMEDA[J].Journal of Henan Polytechnic University(Natural Science) ,2023,42(1):102-115.

基于ARLMD和IMOMEDA的滚动轴承早期微弱故障诊断

唐贵基1, 丁傲1,2, 王晓龙1, 张晔2, 姜超2, 李海明2

1.华北电力大学 能源动力与机械工程学院，河北 保定 071003；2.国网吉林省电力有限公司 长春供电公司，吉林 长春 130021

摘要:为实现强噪声干扰下滚动轴承早期微弱故障损伤特征的准确识别，提出一种自适应鲁棒局部均值分解（adaptive robust local mean decomposition，ARLMD）和改进多点最优最小熵解卷积（improved multipoint optimal minimum entropy deconvolution adjusted，IMOMEDA）相融合的诊断新方法。首先，利用基于斯皮尔曼相关系数的ARLMD算法对原始信号进行处理，按照线性峭度（L-kurtosis）最大原则筛选出蕴含丰富故障信息的最佳分量，实现原始信号信噪比的有效提升；然后，针对多点最优最小熵解卷积（multipoint optimal minimum entropy deconvolution adjusted，MOMEDA）效果受滤波长度影响较大的特点，提出基于余量自相关能量比（residual autocorrelation energy ratio，RAER）的最优滤波长度选取策略，对最佳分量进行IMOMEDA处理，实现周期性冲击特征强化放大；最后，计算解卷积信号的包络谱，从中提取出特征频率信息。仿真、实验及工程信号分析结果表明，所提方法可有效提取强噪声下的微弱故障特征，实现轴承损伤的精确诊断。

关键词:滚动轴承;早期故障;鲁棒局部均值分解;解卷积;余量自相关能量比

doi:10.16186/j.cnki.1673-9787.2021060085

基金项目:国家自然科学基金资助项目（52005180）；河北省自然科学基金资助项目（E2020502031，E2022502003）；中央高校基本科研业务费专项资金资助项目（2021MS069）

收稿日期:2021/06/22

修回日期:2021/10/20

出版日期:2023/01/25

Incipient weak fault diagnosis for rolling bearing based on ARLMD and IMOMEDA

TANG Guiji1, DING Ao1,2, WANG Xiaolong1, ZHANG Ye2, JIANG Chao2, LI Haiming2

1.School of Energy，Power and Mechanical Engineering，North China Electric Power University，Baoding  071003，Hebei，China；2.Changchun Power Supply Company，State Grid Jilin Electric Power Company Co.，Ltd.，Changchun  130021，Jilin，China

Abstract:In order to realize the weak fault identification of rolling bearing under the working condition of background noise interference，a new diagnosis method combining adaptive robust local mean decomposition（ARLMD）and improved multipoint optimal minimum entropy deconvolution adjusted（IMOMEDA）was proposed.Firstly，in order to effectively improve the SNR of original signal，ARLMD algorithm based on Spearman correlation coefficient was used to process the original signal，and then the best component containing rich fault information was screened out by L-kurtosis maximum principle.To solve the limitation which the accuracy of MOMEDA was affected by filter length，an improved MOMEDA method was proposed based on residual autocorrelation energy ratio（RAER）.IMOMEDA was executed on the optimal component to enhance and amplify the periodic impact features.Finally，the characteristic frequency information was extracted from the envelope spectrum of deconvolution signal.The simulation，experiment and engineering signal analysis results showed that the proposed method could effectively extract the weak fault features under strong noise environment，and could realize the accurate diagnosis of bearing damage.

Key words:rolling bearing;incipient faults;robust local mean decomposition;deconvolution;residual autocorrelation energy ratio

 013_2021060085_唐贵基_H.pdf