基于改进诊断算法的空压机维护策略研究  

Research on Maintenance Strategy of Air Compressor Based on Improved Diagnosis Algorithm

作　　者：张炎亮[1] 马秋丽 雷俊辉 

机构地区：[1]郑州大学管理工程学院,河南郑州450001

出　　处：《工业工程与管理》2017年第6期32-39,共8页Industrial Engineering and Management

基　　金：国家自然科学基金资助项目(71271194);河南省高等学校重点科研项目计划资助项目(16A630035);河南省基础与前沿技术研究计划资助项目(162300410073)

摘　　要：建立了遗传算法(GA)优化多分类支持向量机(MSVM)的参数优化模型,解决了MSVM的惩罚参数c和核函数参数g选取中存在的盲目性问题;构建了基于GA-MSVM空压机故障诊断系统并在对比标准MSVM、PSO-MSVM和GS-MSVM的诊断结果的基础上,验证了GA-MSVM对空压机的故障诊断结果的准确性与稳定性。并基于FMEA分析提出了相应的预防性维护策略,为更好地空压机的故障状态诊断和空压机安全运行提供保障。The parameter optimization model was established based on multi class support vector machine（MSVM）optimized by genetic algorithm（GA）,and the blindness in the selection of MSVM penalty parameter C and kernel function parameter g were solved. The fault diagnosis system of air compressor based on GA-MSVM was built. Meanwhile, on the basis of comparing the diagnostic results of standard MSVM, PSO-MSVM and GS-MSVM, the accuracy and stability of the fault diagnosis of air compressor by GA-MSVM were verified. The corresponding preventive maintenance strategy was put forward based on the analysis of FMEA, and the guarantee was provided for better fault diagnosis and safe operation of air compressor.

关 键 词：故障诊断 遗传算法 多分类支持向量机 失效模式及后果分析 

分 类 号：TH45[机械工程—机械制造及自动化]