基于支持向量机的往复压缩机气缸活塞系统故障诊断技术Fault Diagnosis Technique Based on Support Vector Machines for Cylinder and Piston System in Reciprocating Compressor
王金东,赵海洋,于庆江
WANG Jin-dong,ZHAO Hai-yang,YU Qing-jiang(Daqing Petroleum Institute
摘要(Abstract)：
提出了一种基于支持向量机的往复压缩机气缸活塞系统故障诊断技术。压缩机P-V指示图是压缩机气缸部分工作状况的真实记录,该技术将采集的气缸压力信号经归一化处理后作为特征向量,输入到由多个支持向量机构造的多值分类器中进行故障模式分类。试验结果表明,该技术对多种气缸活塞系统故障诊断的准确率较高。与常用的BP神经网络相比,该诊断技术具有较强的适应性和更好的有效性、鲁棒性。
A fault diagnosis technique based on support vector machines for cylinder and piston system in reciprocating compressor is put forward.The P-V indicating diagram of compressor is a truthful record of working condition in the cylinder.According to the technique,the pressure signals collected from cylinder are normalized and treated as characteristic vector.And the vectors are inputted into a multi-class classifier composed of many support vector machines to identify fault modes.The experimental diagnosis results show that the technique can identify faults of cylinder and piston system more correctly.Furthermore,comparing with the traditional BP neural networks,the technology is more adaptive,efficient and robust.
关键词(KeyWords)：
故障诊断;支持向量机;往复压缩机;气缸
fault diagnosis;support vector machines;reciprocating compressor;cylinder
基金项目(Foundation): 黑龙江省骨干教师资助计划项目(1054G002)
作者(Author):
王金东,赵海洋,于庆江
WANG Jin-dong,ZHAO Hai-yang,YU Qing-jiang(Daqing Petroleum Institute
参考文献(References)：
- [1]Boser B,Guyon I,Vapnik B.A training algorithm for op-timal margin classifiers[A].In:Fifth Annual Workshopon Computational Learning Theory[C].Pittsburgh:ACMPress,1992.
- [2]Cortes C,Vapnik V.Support vector networks[J].Ma-chine Learning,1995,20:273-397.
- [3]时文刚,刘树林,张嘉钟.基于支持向量机的往复泵泵阀故障诊断方法[J].机械强度,2002,24(3):362-364.
- [4]王金东,张嘉钟,杨敬源.2D12往复压缩机气缸内压力检测技术[J].压缩机技术,2002(5):21-23.