基于IDE-LSSVM的柴油机气门故障诊断算法研究  

Diesel engine valve fault diagnosis algorithm is based on IDE-LSSVM

作　　者：曹龙汉[1] 秦恺[1] 张迁[1] 牟浩[1] 文迪[2] 

机构地区：[1]重庆通信学院控制工程重点实验室,重庆400035 [2]重庆邮电大学汽车电子与嵌入式系统研究所,重庆400065

出　　处：《自动化与仪器仪表》2014年第5期20-24,共5页Automation & Instrumentation

基　　金：重庆市应用开发计划项目(cstc2013yyk-fA40004)资助

摘　　要：最小二乘支持向量机在柴油机故障样本少的情况下具有优越的学习能力和泛化能力,特别适合故障诊断,但同时也存在着参数设置等难题。针对最小二乘支持向量机参数设置问题,提出了一种改进微分进化算法优化最小二乘支持向量机方法。利用柴油机气门振动信号实测数据,经小波变换作为诊断模型的故障特征,建立了基于IDE-LSSVM的柴油机气门间隙故障诊断模型,并与基于遗传和基于粒子群算法的LSSVM模型进行比较,结果表明:基于IDE优化的LSSVM模型收敛速度快,故障识别率较高。Least squares support vector machine under the condition of diesel engine fault diagnosis few samples has excellent learning ability and generalization ability , especially for fault diagnosis, but also there is parameter settings and other problems. For least squares support vector machine parameters problem, The mixed stategy of differential evolution algorithm to optimize the parameters of the Least squares support vec-tor machine is proposed. According to diesel engine valve vibration signals measured data and as the character-istic values of the diagnostic model after wavelet transform,the diesel engine valve clearance fault diagnosis model of LSSVM based on IDE is established and compared with the GA-LSSVM and PSO-LSSVM model. The results showed that：IDE-LSSVM has perfect accuracy and convergence speed in the diagnosis classifica-tion.

关 键 词：柴油机 故障诊断 微分进化算法 最小二乘支持向量机 

分 类 号：TP368.1[自动化与计算机技术—计算机系统结构]