基于贝叶斯征兆解释度的链路故障定位算法

摘要

针对故障和征兆关系不确定的网络中故障定位算法检测率低和误检率高的缺陷,提出了一种基于贝叶斯征兆解释度的链路故障定位算法.该算法以概率加权的二分图作为故障传播模型,通过处理贝叶斯后验概率信息,定义一种新的参数贝叶斯征兆解释度,并基于该参数对可能链路故障进行判断,得出最优故障假设集合,实现链路故障定位.理论分析和仿真实验表明,该算法具有较低的计算复杂度,且在小规模不确定网络中具有较高的故障检测率和较低的故障误检率.%Aiming at the low detection rate and high false positive rate of fault localization algorithm in network of uncertainty relationship between fault and symptoms, this paper proposed a link failure localization algorithm based on Bayesian symptom explained degree. This algorithm took probabilistic weighted bipartite graph as fault propagation model, it defined a novel parameter Bayesian symptom explained degree by handling the Bayesian posterior probability, and dealt with the possible link failure based on the parameter, then obtained the optimal fault hypothesis set and realized link failure localization. The theory analysis and simulation results show that the algorithm has lower complexity, and it has higher fault detection rate and lower false positive rate in uncertainty small size network.