摘要
针对从行星齿轮箱非线性、非平稳振动信号中提取故障特征困难的问题,提出了一种基于精细时移多尺度模糊熵(refined time-shift multiscale fuzzy entropy,RTSMFE)、马氏距离的核正则化共面判别分析(Mahalanobis-kernel regularized coplanar discriminant analysis,M-KRCDA)以及郊狼优化算法优化支持向量机(coyote optimization algorithm-support vector machine,COA-SVM)的行星齿轮箱故障诊断方法。首先利用RTSMFE计算和组合行星齿轮箱原始故障信号的特征向量,构建原始高维故障特征集;然后采用M-KRCDA的特征筛选方法,减少了特征的维数并提高特征故障识别的准确性和效率;最后将低维特征输入到COA-SVM进行故障类型的判别。行星齿轮箱故障诊断试验结果分析表明,所提方法能够准确识别行星齿轮箱的常见故障,具有一定的应用前景。
Here,aiming at the problem of extracting fault features being difficult from nonlinear and nonstationary vibration signals of planetary gear box,a new planetary gearbox fault diagnosis method based on refined time-shift multiscale fuzzy entropy(RTSMFE),Mahalanobis-kernel regularized coplanar discriminant analysis(M-KRCDA) and coyote optimization algorithm-support vector machine(COA-SVM) was proposed.Firstly,RTSMFE was used to calculate and combine eigenvectors of original fault signals of planetary gearbox,and construct original high-dimensional fault feature set.Then M-KRCDA’s feature screening method was used to reduce features’ dimension number and improve accuracy and efficiency of feature fault identification.Finally,low-dimensional features were input into COA-SVM to distinguish fault types.The experimental results of planetary gearbox fault diagnosis showed that the proposed method can accurately identify common faults of planetary gearbox,and have a certain application prospect.
作者
戚晓利
崔创创
杨艳
程主梓
陈旭
QI Xiaoli;CUI Chuangchuang;YANG Yan;CHENG Zhuzi;CHEN Xu(School of Mechanical Engineering,Anhui University of Technology,Maanshan 243032,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2022年第21期109-120,共12页
Journal of Vibration and Shock
基金
安徽省自然科学基金(1808085ME152)。
作者简介
第一作者:戚晓利,男,博士,副教授,1975年生;通信作者:崔创创,男,硕士生,1997年生。
