基于核主元分析和最小二乘支持向量机的中速磨煤机故障诊断  

Fault Diagnosis of Medium Speed Mill Based on KPCA and LSSVM

作　　者：刘定平[1] 叶向荣[1] 陈斌源[1,2] 汤美玉[1] 

机构地区：[1]华南理工大学电力学院,广州510640 [2]广东珠海金湾发电有限公司,珠海519000

出　　处：《动力工程》2009年第2期155-158,共4页Power Engineering

摘　　要：为了对火电厂磨煤机作出早期故障预测并有效判别其故障类型,提出了基于核主元分析(KPCA)和最小二乘支持向量机(LSSVM)的磨煤机故障诊断新方法,并采用该方法对某电厂的HP碗式中速磨煤机的故障特征数据进行了仿真试验.结果表明:该方法可提取变量的特征信息,以有效地捕捉变量间的非线性关系,从而能有效地处理故障征兆与故障类型之间的不确定性,具有很好的分辨力,而且故障诊断的正确率很高.In order to obtain prediction of mill faults, as well as the faults type, a new fault diagnosis method based on kernel principal component analysis （KPCA） and least square support vector machines （LSSVM） was given. A simulation by this method was carried out based on fault feature data of the mill. Results show that variables characteristic can be extracted effectively to describe the nonlinear relationship of original datasets. Indeterminacy between fault symptom and fault type can be disposed availably, and the method has good discrimination capability and right fault diagnosis.

关 键 词：中速磨煤机 故障诊断 最小二乘支持向量机 核主元分析 

分 类 号：TK223[动力工程及工程热物理—动力机械及工程]