遗传算法在航电故障诊断规则优化中的应用

APPLYING GENETIC ALGORITHM IN OPTIMIZATION OF AVIONICS FAULT DIAGNOSIS RULES

航空电子系统的日趋复杂化,使得用来进行故障诊断的模糊规则集合越来越庞大、冗余度越来越高,不能满足飞机故障诊断准确实时的需要。针对这一问题,采用一种基于Pareto优胜关系的多目标遗传算法——MOGA-Ⅱ对模糊规则集合进行去冗余优化,在保证诊断出的故障数目尽量多的前提下使得所用的规则数目尽量少。仿真结果表明,与传统的规则优化方法——权重系数法相比,MOGA-Ⅱ在不能获得足够的专家知识或操作经验时可以得到更加紧凑无损的模糊规则集合,更适合于应用在航电故障诊断规则优化工作中。

The fuzzy rule sets used for diagnosing the faults of avionic devices have an increasing size and redundancy due to the growing complicity of the avionics system, this lead to dissatisfying the demand of accurate and real-time diagnosis on aircraft faults. To solve the prob- lem, in this paper, a Pareto dominance-based Multiple Objective Genetic Algorithms, in short MOGA-]I , was used to optimise the redundancy removal of the fuzzy rule sets, in order to use as less as possible the rules in premise of ensuring as many as possible the faults to be diag- nosed. Simulation results indicate that compared with traditional rules optimization method, the weight coefficient method, MOGA- Ⅱ is able to acquire more compact and lossless fuzzy rule sets without sufficient expert knowledge or operation experience, and is more applicable for the optimisation of avionics fault diagnosis rules.