一个基于并行集团的系统级故障概率诊断算法  

An System-Level Fault Probabilistic Diagnosis Algorithm Based on Parallel Group

作　　者：张大方[1] 陈兴华[1] 宣恒农[2] 

机构地区：[1]湖南大学计算机与通信学院,湖南长沙410082 [2]南京经济学院计算机系,江苏南京210003

出　　处：《电子学报》2003年第2期180-182,共3页Acta Electronica Sinica

基　　金：国家自然科学基金 (No .699730 1 6);广东省自然科学基金 (No.0 1 0 4 75)

摘　　要：概率诊断算法是系统级故障诊断研究的一个重要方面 ,本文提出了一种基于并行集团的概率诊断算法 PGSFPD算法 ,并设计了一个系统级故障诊断软件仿真系统 ,对诊断算法进行仿真 ,分析比较各算法的性能 ,仿真结果表明PGSFPD算法性能优于经典的概率诊断算法 Somani&Agrawal算法 ,可在只需较少测试数的情况下 ,在保持很高诊断正确率的同时 。Consider a system composed of N nodes that can be faulty or fault free.The purpose of system level fault diagnosis is to have each fault free node determine the state of all nodes of the system.Probabilistic diagnosis algorithm is very important in the system level fault diagnosis research.A system level fault probabilistic diagnosis algorithm based on parallel group was proposed,PGSFPD algorithm,and the software emulation system of system level fault diagnosis was presented.Experimental results are given through simulation of the algorithm for large systems of from 200 nodes to 1000 nodes under the emulation system.The results indicate the PGSFPD algorithm has better performance than the Somani & Agrawal algorithm,which is a classic probabilistic algorithm in system level fault diagnosis.

关 键 词：系统级故障诊断 集团 概率诊断 诊断算法 软件仿真 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]