基于多目标遗传优化的容差电路故障屏蔽诊断  

Tolerance-Circuit Fault Screen Diagnosis Based on Multiobjective Genetic Algorithm

作　　者：彭敏放[1] 何怡刚[1] 沈美娥[2] 贺建飚[3] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]北京信息工程学院计算机科学与工程系,北京100101 [3]中南大学信息科学与工程学院,长沙410083

出　　处：《电工技术学报》2006年第3期118-122,共5页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金(50277010);湖南省自然科学基金(04JJ6034);教育部新世纪优秀人才支持计划(NCET-04-0767);湖南省科技计划(04FJ2003;03GKY3115;05FJ3008)资助项目。

摘　　要：将多目标遗传优化算法与故障屏蔽原理相结合,提出了一种新的容差模拟电路故障屏蔽诊断方法。通过分析计算故障与容差效应对端口特征量的影响,建立了模拟电路具容差情况下的故障屏蔽模型,创建了目标函数与相应的适应值函数,用以搜索能极大限度屏蔽故障支路的最优激励,根据最优激励的Pareto解集及所提隶属度函数来实现故障定位。模拟试验结果表明:所提方法避免了故障效应与容差效应的相互交叠,故障定位准确率高。In order to solve the difficult problem of searching for the optimal excitation in tolerance-circuit fault screen diagnosis and improve the diagnosis accuracy, a fault diagnosis strategy for analog circuits with tolerance is proposed by combining the principle of fault screen with the genetic algorithm for multi-objective optimization. After analyzing and calculating faults and tolerance effort to terminal performance, the mathematical model of fault screen for tolerance circuits is established based on statistics theory. The objective function and the corresponding fitness function are presented, by which Pareto optimal solutions of the excitation vectors screening fault branches are searched out. Based on the gained optimal excitation, a fault attaching function is proposed and applied to fault location. The experiment results show that the proposed strategy avoids the superposition effects of tolerance and faults and achieves satisfactory accuracy.

关 键 词：故障诊断 遗传算法 多目标优化 故障屏蔽 模拟电路 

分 类 号：TM131[电气工程—电工理论与新技术]