On-line Fault Detection Method Based on Modified SVDD for Industrial Process System

摘要：

To tackle problems of on-line fault detection inindustrial process system,a method based on modifiedSVDD(Support Vector Data Description)is presentedSince the performance of SVDD is strongly influencedby kernel parameter selection and hyper-sphereclass-boundary.A new criterion is presented tooptimize the kernel parameter,which is performed bymeasuring the non-Gaussian value of kernel samplevector.Moreover,when the kernel sample vector iswith uneven distribution,there are certain risks inusing hype-sphere as class-boundary in comparisonwith hype-ellipse class-boundary.KPCA(Kernel PCA)is employed to adjust the hyper-sphere to acquire morereasonable class-boundary,in which the length ofellipses major axis in each principal componentdirection is computed and then equalize each majoraxis by scaling method.The feasibility andeffectiveness of proposed approach is illustrated by theapplication of standard data and fault diagnosissimulation experiment.

作者: JinFa Zhuang Jian Luo YanQing Peng ChangQing Wu

作者单位: Department of automation,Xiamen University,Xiamen 361005,China Department of automation,Xiamen University,Xiamen 361005,China;Department of Electronic and Electric

会议类型: 国际会议

会议名称: 2008 3rd International Conference on Intelligent System and Knowledge Engineering(第三届智能系统与知识工程国际会议)(ISKE 2008)

会议地点: 厦门

会议语种:英文

页码: 754-760

在线出版日期: 2008-11-17（万方平台首次上网日期，不代表论文的发表时间）