一种基于PMC模型下的概率性矩阵诊断算法

发布时间：2018-06-10 00:46

  本文选题：系统级故障诊断 + 概率性矩阵诊断算法　； 参考：《南京理工大学学报》2017年04期

【摘要】：系统级故障诊断是提高多处理器系统可靠性的必要手段。为了有效定位多处理系统中的故障单元,该文建立了一种基于PMC模型t可诊断条件下的概率性矩阵诊断算法。首先对一般概率性矩阵诊断算法进行仿真分析获悉其具有较高的误检率,在诊断过程中引进绝对故障基和节点集团思想,通过计算绝对故障基以寻找系统中的部分故障处理机,集团用于将不确定状态的节点单元分类以补充正常节点集合,改善了原诊断的限制条件。仿真实验验证:改进后的概率性矩阵诊断算法保持了很高的检测精度,并且随着节点数的增多极大地降低了误检率,提高了诊断效果,使得该算法具有广泛的适用性。
[Abstract]:System level fault diagnosis is an essential means to improve the reliability of multiprocessor systems. In order to effectively locate the fault units in multiprocessing systems, a probability matrix diagnosis algorithm based on PMC model t diagnostics is established in this paper. Firstly, the general probability matrix diagnosis algorithm is simulated and found that it has a high false detection rate. In the process of diagnosis, the idea of absolute fault base and node group is introduced, and the absolute fault base is calculated to find part of the fault processor in the system. The cluster is used to classify the node units of uncertain states to supplement the normal set of nodes, which improves the limit condition of the original diagnosis. Simulation results show that the improved probabilistic matrix diagnosis algorithm has high detection accuracy, and with the number of nodes increasing, the false detection rate is greatly reduced, and the diagnosis effect is improved, which makes the algorithm widely applicable.
【作者单位】： 西安电子科技大学数学与统计学院;
【基金】：国家自然科学基金(71271165)
【分类号】：TP332
，

本文编号：2001394