基于改进粒子群优化的SVM故障诊断方法  

Support Vector Machine Fault Diagnosis Method Based on Improved Particle Swarm Optimization

作　　者：杨柳松[1] 何光宇[2] 

机构地区：[1]东北林业大学机电工程学院,哈尔滨150040 [2]空军工程大学工程学院,西安710038

出　　处：《计算机工程》2013年第3期187-190,196,共5页Computer Engineering

基　　金：中央高校基本科研业务费专项基金资助项目(DL11BB32);黑龙江省科技厅自然科学基金资助项目(F201028)

摘　　要：针对支持向量机(SVM)分类模型参数选取困难的问题,提出基于遗传免疫的改进粒子群优化算法,克服传统粒子群算法前期收敛快、后期易陷入局部最优的缺陷。将该算法与优化支持向量机分类模型相结合,建立基于遗传免疫粒子群和支持向量机的诊断模型,并用于轴承故障诊断中。结果表明,基于遗传免疫粒子群算法优化的SVM可实现对SVM分类模型参数的自动优化,并能提高SVM分类模型的故障诊断精度,对分散程度较大、聚类性较差的故障样本分类有较强的适用性。In order to resolve the difficulty that the choice of parameters influence the accuracy of Support Vector Machine（SVM） fault diagnosis model, a genetic-immune Particle Swarm Optimization（PSO） algorithm based on genetic evolution algorithm and immune selection algorithm is presented and used to optimize model parameters of SVM. The forecasting model based on a genetic-immune PSO algorithm and SVM is proposed and used to diagnose bearing fault. The results show that diagnosis model of SVM optimized by genetic-immune PSO algorithm can achieve automatic optimization of parameters, increase diagnosis accuracy of the conventional cross-validation algorithm, and is more fitting to classify the faulty samples scattered greatly.

关 键 词：支持向量机 故障诊断 粒子群优化 遗传免疫 轴承 交叉验证 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]