用聚类支持向量机模型监测环保设备运行状态  

Running condition monitoring for environmental protection equipment based on a clustering support vector machine

作　　者：曹晓莉[1,2] 甘思源[1] 江朝元[2] 

机构地区：[1]重庆工商大学废油资源化技术与装备教育部工程研究中心,重庆400067 [2]重庆工商大学计算机科学与信息工程学院,重庆400067

出　　处：《重庆大学学报（自然科学版）》2009年第6期674-679,共6页Journal of Chongqing University

基　　金：重庆市教委资助项目(KJ080708);废油资源化技术与装备教育部工程研究中心重大项目(07011301)

摘　　要：针对环保设备多状态监测与识别问题,提出了一种基于多分类聚类支持向量机的分类算法模型。该算法模型充分利用多个相关联被监测设备及其状态属性特征,采用神经网络聚类分析获得关联特征的聚类子空间,再对聚类子空间构造二叉树支持向量机进行多分类识别。该算法模型避免了盲目分类,与单纯的支持向量机算法比较具有更好的分类性能。通过对日本大晃船用污水处理装置实测样本的训练、检验表明,该算法具有98%以上的分类识别率。A multi-classification clustering model based on Support Vector Machines （SVM） is proposed to address problems of condition monitoring and identification in environmental protection equipment. Based on the associated attributes of monitoring equipment and their condition features, a neural network is adopted for clustering analysis to obtain clustering subspaees. The clustering subspace is employed to design a binary tree architecture based multi-classification SVM for multi-classification recognition. Compared with the traditional SVM learning algorithms, the proposed algorithm improves the classification performance while avoiding blind classification. The model is applied to train and examine the measured samples for a ship sewage treatment device made in Japan. The results show that the classified recognition rate of this algorithm exceeds 98% and is valuable for practical applications.

关 键 词：支持向量机 设备状态 多分类 聚类 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]