摘要
在分布式系统中,系统的规模庞大,结构复杂,这使得故障数据的收集十分困难;为了对分布式系统进行故障诊断,文章提出了一种基于不完全故障数据的故障诊断方法;首先,应用非对称比较方法进行系统的故障诊断;其次,将系统的故障诊断转化为一个二分类问题;最后,根据系统运行的部分症状数据,应用线性支持向量机模型对系统中的节点状态进行分类;模拟实验表明,文章提出的方法能有效的识别出系统中的故障;此外为了进一步提高该方法的执行效率,可以采用系统的部分症状数据将故障节点限定在一定的范围内。
In distributed systems,the scale is very large,and the structure is complex,which make the collection of fault data of the systems very hard.In order to diagnose the distributed systems efficiently,this paper proposed an incomplete fault data based fault diagnosis approach in distributed systems.Firstly,we diagnosed the distributed systems with the asymmetric comparison-based method.Secondly,we transformed the fault diagnosis problem into a binary classification problem.Finally,we applied the linear support vector machine to classify the nodes into two statuses based on partial syndromes.The simulation experiments show that,the proposed approach is efficient in recognizing fault nodes.In addition,in order to improve the execution efficiency of the proposed approach,we can use incomplete syndromes to refine the fault nodes in a certain sets.
出处
《计算机测量与控制》
北大核心
2014年第11期3539-3542,共4页
Computer Measurement &Control
基金
国家自然科学基金项目(61127443)
关键词
不完全故障数据
分布式系统
故障诊断
支持向量机
incomplete fault data
distributed system
fault diagnosis
support vector machine
