铁道科学与工程学报2025，Vol.22Issue(2)：887-899,13.DOI:10.19713/j.cnki.43-1423/u.T20240637
基于最优参数VMD和改进散布熵的轴承亚健康状态识别
Bearing sub-health state identification based on optimal parameter VMD and improved dispersion entropy
摘要
Abstract
To address the challenges of noise interference,mode mixing,and difficulty extracting state features in bearing sub-health conditions,a method based on optimal parameter Variational Mode Decomposition(VMD)and improved dispersion entropy was proposed for bearing sub-health state recognition.First,an Improved Sparrow Search Algorithm(ISSA)was designed to search for the optimal VMD decomposition parameters adaptively,improving the efficiency and quality of VMD decomposition.Then,by using these optimal parameters,the signal was decomposed by VMD to obtain a series of Intrinsic Mode Functions(IMF).Next,the Pearson Correlation Coefficient(PCC)between each IMF and the original signal was calculated.IMFs with a correlation coefficient greater than 0.3 were selected to reconstruct the signal,achieving noise reduction and enhanced state features.Second,to better characterize the complexity and irregularity of bearing signals and effectively distinguish between healthy and sub-health states,time-shifted multi-scale analysis and fractional calculus were introduced into dispersion entropy.This allowed for the extraction of fine-state features of bearings at multiple scales.Finally,the Euclidean distance was used to characterize the bearing state curve,and a sub-health threshold was set based on the Chebyshev inequality.When the Euclidean distance exceeded this threshold,a corresponding warning was issued,completing the bearing sub-health state recognition.Experimental results on the bearing datasets of XJTU-SY and IMS show that the ISSA algorithm has higher convergence speed and accuracy than other optimization algorithms.The optimal parameter VMD effectively eliminates the mode mixing problem,and the improved dispersion entropy accurately extracts fine features of the bearing's full life state.The proposed algorithm can accurately identify the sub-health state of bearings and provide warnings without the need for model training,facilitating better maintenance of bearing operating conditions by maintenance personnel.关键词
轴承/亚健康状态识别/最优参数VMD/改进麻雀搜索算法/时移多尺度分数阶散布熵Key words
bearings/sub-health state identification/optimal parameter VMD/improved sparrow search algorithm/time-shifted multi-scale fractional dispersion entropy分类
机械工程引用本文复制引用
魏文军,甘洁..基于最优参数VMD和改进散布熵的轴承亚健康状态识别[J].铁道科学与工程学报,2025,22(2):887-899,13.基金项目
国家自然科学基金资助项目(61863023) （61863023）
甘肃省自然科学基金资助项目(23JRRA868) （23JRRA868）
