基于ICEEMDAN与GWO-LSSVM的滚动轴承故障诊断  

Fault Diagnosis of Rolling Bearings Based on ICEEMDAN and GWO-LSSVM

作　　者：丁伟 DING Wei(School of Automotive and Intelligent Transportation,Jiangsu Vocational College of Information Technology,Wuxi 214153,China)

机构地区：[1]江苏信息职业技术学院汽车与智能交通学院,江苏无锡214153

出　　处：《机械工程与自动化》2025年第2期142-144,151,共4页Mechanical Engineering & Automation

基　　金：江苏省高等学校基础科学(自然科学)研究面上项目(23KJD430006);江苏高校“青蓝工程”资助项目(苏教师函[2022]51号)。

摘　　要：针对不同状态下滚动轴承时域波形与幅值谱的差异不明显,难以准确判断轴承运行状况的问题,提出了一种新的滚动轴承状态分类方法。该方法首先使用改进自适应噪声完备经验模式分解方法(ICEEMDAN)对采集的轴承振动信号进行分解,筛选出包含主要特征频率的固有模态函数(IMF);然后通过计算IMF排列熵提取出故障特征信息,构建特征向量;最后通过灰狼算法(GWO)优化最小二乘支持向量机(LSSVM)模型中的惩罚因子和核参数,将其应用于滚动轴承的故障诊断。实验测试结果表明:将ICEEMDAN分解与GWO-LSSVM相结合的轴承故障诊断方法能够有效地识别轴承的状态,总的分类准确度可达98%,相较于传统的参数自选LSSVM方法,所提出的故障识别方法整体识别准确率提高了10.33%。A new rolling bearing state classification method is proposed to address the issue of unclear differences in the time-domain waveform and amplitude spectrum of rolling bearings under different states,making it difficult to accurately determine the operating condition of bearings.This method first uses the Improved Adaptive Noise Complete Empirical Mode Decomposition(ICEEMDAN)method to decompose the collected bearing vibration signals and select the Intrinsic Mode Functions(IMFs)that contain the main characteristic frequencies;Then,by calculating the IMF permutation entropy,fault feature information is extracted and feature vectors are constructed;Finally,the penalty factors and kernel parameters in the Least Squares Support Vector Machine(LSSVM)model were optimized using the Grey Wolf Algorithm(GWO)and applied to the fault diagnosis of rolling bearings.The experimental test results show that the bearing fault diagnosis method combining ICEEMDAN decomposition with GWO-LSSVM can effectively identify the status of bearings,with an overall classification accuracy of 98%.Compared with the traditional parameter self selection LSSVM method,the proposed fault recognition method has an overall recognition accuracy improvement of 10.33%.

关 键 词：滚动轴承 故障诊断 ICEEMDAN分解 灰狼优化算法 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TP306.1[自动化与计算机技术—控制科学与工程]