基于二叉树的LS-WSVM早期火灾多类分类研究  

Research on LS-WSVM Based on Binary Tree in Early Fire Multi-Class Classification

作　　者：庄哲民[1] 李卡麟[1] 张新峰[1] 李芬兰[1] 

机构地区：[1]汕头大学电子工程系,汕头515063

出　　处：《测试技术学报》2009年第6期482-486,共5页Journal of Test and Measurement Technology

基　　金：广东省科技计划资助项目(0711050600004)

摘　　要：针对早期火灾信息特点,提出了一种基于二叉树的最小二乘小波支持向量机(Least squareswavelet support vector machine,LS-WSVM)多类分类方法.该方法首先把主成份分析用于早期火灾信息的特征提取.然后,把二叉树结构和LS-WSVM相结合,提出了基于二叉树的LS-WSVM多类分类模型,不仅避免了盲目分类和不可分情况,而且提高了分类速度和泛化能力.最后,用该模型对特征信息进行处理,从而实现了对早期火灾的多类识别.早期火灾分类实验结果表明,该方法比采用径向基核函数的最小二乘支持向量机多类分类方法具有更好的识别效果和更快的分类速度.Aiming at the characteristics of early fire information, a least squares wavelet support vector machine （LS-WSVM） multi-class classification method based on binary tree was presented. The method first used principal component analysis to extract the feature of early fire information. And then it combined binary tree with LS-WSVM, so that the LS-WSVM multi-class classification model based on binary tree was proposed. The model not only avoids the condition of blind classification and impartibility, but also improves classification speed and generalization ability. Finally, the model was used to classify the extracted feature information, thus we accomplished the multi-class recognition of early fire. The early fire classification experiment results show the method has better recognition effect and faster classification speed compared with the least squares support vector machine multi-class classification method based on radial basis kernel function.

关 键 词：二叉树 最小二乘小波支持向量机 早期火灾 多类分类 主成份分析 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]