文献类型：期刊文献

中文题名：CEEMDAN-FuzzyEn-PPCA与IWOA-SVM优化的轴承故障诊断

英文题名：Bearing Fault Diagnosis of CEEMDAN-FuzzyEn-PPCA and IWOA-SVM Optimization

作者：韩正功 周知进 李玢

第一作者：韩正功

机构：[1]贵州理工学院机械工程学院,贵州贵阳550003;[2]贵阳市建筑设计院有限公司,贵州贵阳550081

第一机构：贵州理工学院机械工程学院

年份：2022

期号：7

起止页码：108-113

中文期刊名：机械设计与制造

外文期刊名：Machinery Design & Manufacture

收录：CSTPCD;;北大核心:【北大核心2020】；

基金：国家自然科学基金项目(51479073);贵州省科技计划项目(黔科合基础[2019]1152号)。

语种：中文

中文关键词：支持向量机;参数优化;改进鲸鱼优化算法;故障诊断;滚动轴承

外文关键词：Support Vector Machine;Parameter Optimization;Improved Whale Optimization Algorithm;Fault Diagnosis;Rolling Bearing

摘要：针对滚动轴承故障问题,提出一种基于CEEMDAN-FuzzyEn-PPCA的特征提取方法和一种IWOA-SVM优化模型用于轴承故障诊断。首先,使用自适应白噪声的完整集成经验模态分解方法(CEEMDAN)对振动信息进行消噪与分解,得到若干个平稳的IMF模态分量;其次,提取模糊熵(FuzzyEn)值,进行概率主成分分析(PPCA)特征处理,作为支持向量机(SVM)的输入;最后,在用样本训练SVM时,采用改进鲸鱼优化算法(IWOA)对惩罚参数C和径向基核函数参数σ进行寻优,实现对滚动轴承故障类型的辨识。研究结果表明:与GA-SVM、PSO-SVM、WOA-SVM相比,基于CEEMDANFuzzyEn-PPCA与IWOA-SVM优化的轴承故障诊断模型能以更快的收敛速度和更高的准确率达到全局最优,具有较强的工程实用性。
A feature extraction method based onCEEMDAN-FuzzyEn-PPCA and an IWOA-SVM optimization model are proposed for bearing fault diagnosis.Firstly,by means of CEEMDAN method,vibration signal could be de-noised and decomposed,which obtained several stable IMF modal components.Furthermore,several FuzzyEn values were extracted for PPCA feature processing as input to SVM.In the end,when training SVM with samples,IWOA was used to optimize the penalty parameters C and radial basis kernel function parametersσ,so as to identify the fault types of rolling bearing.The experimental results show that,compared with GA-SVM,PSO-SVM and WOA-SVM,the bearing fault diagnosis model based on CEEMDANFuzzyEn-PPCA and IWOA-SVM optimization can achieve global optimization with faster convergence rate and higher accuracy,which could be used in engineering application to a large extent.