基于最优二叉树的多故障分类器的设计  

The Design of Multi-fault Classification Based on Huffman Binary Tree

作　　者：朱学冬[1] 胡平[1] 

机构地区：[1]南京工业大学,南京210009

出　　处：《北京联合大学学报》2009年第2期26-29,共4页Journal of Beijing Union University

摘　　要：在分析常见多故障分类器的基础上,提出了基于故障优先级和核函数的聚合技术,利用这个方法生成一个基于最优二叉树的多故障分类器。该分类器是通过核函数具有将非线性问题线性化的特点生成聚类函数,然后通过各个故障的优先级生成最优二叉树。最后将此分类器应用在挖掘机故障诊断中。应用结果表明,此分类器不但能够保证故障诊断的正确率,而且可以让后果较严重的故障得到优先诊断。Based on the analysis of some existing methods of SVM muhi2class classification, a multi2class classification based on Huffman binary tree was put forward. This classification used Kernel function which can transfer non-linear into linear to get the clustering functions, then built a Huffman binary tree based on every fault＇ s priority. This method has been used in the fault diagnosis system of excavator. The result indicates that this classification can find not only faults, but the more important faults first.

关 键 词：支持向量机 最优二叉树 核函数 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]