基于GAPSO-SVM的滚动轴承故障分类方法  

Rolling Bearing Fault Classification Method Based on GAPSO-SVM

作　　者：黄静 张招君 HUANG Jing;ZHANG Zhao-jun(School of Information Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学信息学院,浙江杭州310018

出　　处：《软件导刊》2021年第1期36-40,共5页Software Guide

基　　金：浙江省科技厅重点项目(2019C01114)。

摘　　要：针对滚动轴承故障分类准确率低的问题,提出一种利用遗传算法结合粒子群算法优化支持向量机分类器的故障诊断方法。实验通过提取滚动轴承不同故障状态下的振动信号,以转化成时域和频域组成的特征集为特征向量,利用粒子群生成二维粒子,即惩罚因子C、核函数参数G,并喂入支持向量机进行训练和交叉验证,取最优适应度对应的粒子,进而构建遗传粒子群改进支持向量机故障分类模型。实验证明,粒子群改进的支持向量机与遗传算法改进的支持向量机相比,该算法模型在滚动轴承故障分类中对时域、频域、时频域3个特征集的正确率均有明显改进。Aiming at the problem of low accuracy of rolling bearing fault classification,this paper proposes a fault diagnosis method based on genetic algorithm combined with particle swarm optimization to optimize the classifier of support vector machine.The vibra⁃tion signals of rolling bearing under different fault states are extracted.The two-dimensional particles,namely penalty factor C and ker⁃nel function parameter G,are generated by particle swarm optimization(PSO),which are transformed into the feature set composed of time domain and frequency domain.The two-dimensional particles are fed into SVM for training and cross validation.The particles cor⁃responding to the optimal fitness are selected,and then the fault classification model of SVM improved by genetic particle swarm opti⁃mization is constructed.It is proved that the accuracy of the proposed algorithm model in the fault classification of rolling bearings is higher than that of the particle swarm optimization and genetic algorithm(GA).

关 键 词：支持向量机 粒子群算法 遗传算法 滚动轴承 信号特征 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]