基于RS与LS-SVM多分类法的故障诊断方法及其应用  

Fault diagnosis method based on rough set and least squares support vector machine and its application

作　　者：蒋少华[1,2] 桂卫华[1] 阳春华[1] 戴贤江[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083 [2]韶关学院计算机科学学院,广东韶关512024

出　　处：《中南大学学报（自然科学版）》2009年第2期447-451,共5页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金重点资助项目(60634020);国家自然科学基金资助项目(60574030);教育部博士点科研基金资助项目(20050533016)

摘　　要：针对密闭鼓风炉熔炼过程工艺复杂、工况变化较大的特点,提出应用粗糙集(RS)和最小二乘支持向量机(LS-SVM)进行鼓风炉故障诊断的方法。该方法首先利用RS对炉子的故障样本进行知识约简,获得各故障类型的征兆最小条件属性作为特征向量,然后,输入到由多个最小二乘支持向量机构成的多故障分类器中进行故障识别和分类。研究结果表明:该方法具有较强的泛化能力,诊断准确率达到90%以上。Aiming at the complex and variable reaction of Pb-Zn smelting in imperial smelting furnace （ISF）, a novel method for the furnace fault diagnosis based on rough set （RS） and least squares support vector machine（LS-SVM） was put forward. According to the method, the original fault examples were reduced by using the rough set theory to get a simple rule collection as eigenvectors, and then these eigenvectors were input into multiple fault classifiers of LS-SVM to identify faults. The experimental results show that the method has better classification performance and its classification precision reaches more than 90%.

关 键 词：粗糙集 最小二乘支持向量机 多类分类器 故障诊断 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]