摘要
作为一种基于正定核的学习方法,传统支持向量机(Support Vector Machine,SVM)能较好地解决小样本、非线性、过学习、维数灾和局部极小等问题,从而广泛应用于模式识别、回归估计等领域。当前,核方法及其在故障诊断中的应用引起了人们的广泛重视并成为研究热点。为解决传统支持向量对核函数正定性的限制及求解速度不高的缺陷,通过引入最小二乘支持向量机分类算法提高学习速度,采用隐核特征映射技术实现核函数的进一步扩展,提出了一种新的隐核最小二乘分类器(HKLSC)算法。将其应用于实际工业过程的故障诊断中并根据采集的滚动轴承数据进行了仿真。结果表明,该隐核分类器具有很好的故障诊断性能,为故障诊断提供了一种新的有效途径。
As a general positive kernel - based learning machine, Support Vector Machine （ termed SVM） can solve the problems such as small samples, nonlinear, over fitting, curse of dimensionality and local minima, and it has been widely used in pattern recognition, regression estimation, etc. Currently, kernel - based learning method and its application have attracted more and more researchers and become a new active area in the field of fault diagnosis. However, the standard SVM is basically restricted to static problems, due to the fact that it is a very stringent requirement of positive kernel function and its high computational complexity. By combining the advantages of traditional least square SVM classifier and the extension of kernel via hidden kernel map method, a novel hidden kemel based least square classifier （termed HKLSC） is presented for fault diagnosis of industrial process in this paper. Numerical experiments of rolling bearing are given to illustrate the effectiveness of the proposed method , and the result shows that it might offer a new opportunity in the area of fault diagnosis.
出处
《计算机仿真》
CSCD
北大核心
2009年第9期153-155,280,共4页
Computer Simulation
基金
广西自然科学基金项目(桂科自0832074)
关键词
支持向量机
隐核函数
分类
故障诊断
Support vector machine
Hidden kernel function
Classifier
Fault diagnosis
