基于多域划分的通信信息网络故障联合定位算法  

Network fault location algorithm based on multi-domain partitioning in communication and information networks

作　　者：何清素 周文婷[2] 许鸿飞 崔力民[2] 于忠迎 

机构地区：[1]国网信息通信产业集团有限公司,北京100053 [2]国网新疆电力公司信息通信分公司,新疆乌鲁木齐830002 [3]国网冀北电力有限公司信息通信分公司,北京100053

出　　处：《南京邮电大学学报（自然科学版）》2017年第1期79-84,共6页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：国家电网科研基金(526802150008)资助项目

摘　　要：现有信息通信网络缺乏有效地联合故障定位机制,文中通过对信息通信网络不同网络层面的故障类型分析,提出了基于多域划分的通信信息网络故障联合定位算法。该分析方法首先依据网络节点不同故障类型的发生概率,使用K-Means算法,将整体通信信息网络划分为多个子域,使发生相似故障的网络节点处在相同域内;然后对每个子域建立一个神经网络故障关联分析定位模型,并行地对多个子域内的故障进行分析定位,从而实现通信信息网络的高效联合故障定位。模拟实验结果表明,文中提出的故障定位算法具有较高的准确性和可靠性。To solve the fault localization problem in the national power grid information communication network, this paper proposes a novel algorithm to locate network faults in information and communication networks based on multi-domain partitioning theory. Firstly, the whole network is divided into several sub-domains on the basis of the failure probability of different network fault types by using the K-Means algorithm, in which network nodes with similar faults are partitioned into the same sub-domain. Then, a neural network fault location model is established in each domain, and a gradient descent method is used to locate the faults in every sub-domain simultaneously. The fault location of the whole network can be ef- ficiently achieved based on the proposed approach. Simulation experiment results demonstrate that the fault location algorithm has high accuracy and reliability.

关 键 词：故障联合分析 故障定位 多域划分 神经网络 K-MEANS 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]