摘要
针对机载复杂系统动态多故障诊断中存在的诊断精度低等问题,提出一种基于遗传模拟退火算法的多故障诊断方法。首先根据隐马尔可夫过程描述复杂系统故障演化规律,并基于贝叶斯理论建立多故障诊断模型,将复杂系统的多故障诊断转化为递推求解动态规划问题。在此基础上,将遗传算法与模拟退火算法相结合,提出一种基于遗传模拟退火算法的动态多故障诊断方法,通过提高系统的全局搜索能力,可有效提高算法对动态多故障诊断的准确率。将ABR算法、DSA算法与该算法同时用于某型飞机机载电源系统的故障诊断中,结果表明该算法对该电源系统发生的多故障具有最高的诊断正确率。通过对3种算法进行比较分析发现,该算法适用于中、小规模复杂系统和对实时性要求不高的大规模复杂系统的动态多故障诊断。
Aiming at the problem of low diagnosis accuracy in airborne complex system dynamic multiple fault diagnosis,a dynamic multiple fault diagnosis （DMFD） method based on genetic simulated annealing algorithm is proposed.The fault evolution law of complex system is described according to hidden Markov process,and the dynamic multiple fault diagnosis model is established based on Bayesian theory,which is used to convert the DMFD to the recursively solving dynamic programming problem.And on this basis,a genetic simulated annealing algorithm （GSA） combining genetic algorithm with simulated annealing algorithm is proposed to improve the global searching capability,and efficiently increase the accuracy of dynamic multiple fault diagnosis.The ABR algorithm,DSA algorithm and the proposed algorithm were applied to the fault diagnosis of the airborne power supply system in a certain type of aircraft; the experiment result shows that this method can achieve the highest diagnosis accuracy for the multiple faults occurred in the power supply system.After comparative analysis of these three algorithms,it is found that the proposed algorithm is suitable for the dynamic multiple fault diagnosis in middle-scale,small-scale complex systems,and some large scale complex systems with low real time capacity requirement.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2014年第3期607-613,共7页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(51201182)
航空科学基金(20101996012)资助项目
关键词
动态多故障诊断
遗传模拟退火算法
机载电源系统
复杂系统
dynamic multiple fault diagnosis （DMFD）
genetic simulated annealing algorithm
airborne power supply system
complex system
