基于SVM的建筑电气系统无中心故障诊断算法研究

Research on the decentralized fault diagnosis algorithm for building electrical system based on SVM

针对当前集中式建筑智能化系统存在的诸多问题,以及建筑电气系统安全运行的重要性,在无中心、扁平化的建筑智能化平台上提出一种基于支持向量机(Support Vector Machine,SVM)的建筑电气系统故障诊断算法。因为SVM具有在小样本情况下分类效果较好的特性,所以本算法将SVM插入到每一个节点中,分布式并行地解决建筑电气系统故障诊断问题。实验结果表明提出的算法具有实时性好、实施方便等优点,而且对单个节点或多个节点故障均能做出正确的判定。

In view of the existing problems in the centralized building intelligent system,and the importance of the safety operation of the building electrical system,a fault diagnosis algorithm for building electrical system based on Support Vector Machine（SVM） was proposed on the decentralized and flattening building intelligent platform. Due to SVM has a better classification effect in small sample cases,this algorithm inserted SVM into each node and solved the fault diagnosis problem of the building electrical system in a distributed parallel way. The experimental results showed that the proposed algorithm had the advantages of timely and convenient implementation,and can correctly judge the faults of single node or multiple nodes.