摘要
研究了基于极值理论(EVT)的低频高危事件定量评估方法.构建了考虑驾驶员响应的飞控系统故障后评估模型,介绍了角速率传感器故障后极值样本的获取方法.利用非线性优化模型对极值理论中常用的线性模型进行了改进,针对极值样本分布模型中参数的辨识,对比了几种优化算法对文中评估模型的适用性.采用四种优化算法对模型参数进行了对比辨识以寻求飞行风险条件概率最优解,得出了自适应粒子群算法对文中评估模型适应度最高的结论.最后将传感器故障风险概率加入有驾驶员响应环节的马尔科夫过程模型对飞控系统风险概率进行动态定量评估.其最终结果可为定量评估某型机操纵系统的动态可靠性提供理论依据.
Flight risk has the characters of small probability and great hazard. In order to assess it quantitatively, extreme value theory (EVT) was adopted to analyze the distribution of decisive parameters. Firstly the fault model that takes pilot response into consideration was built, then a method of acquiring the extreme sample when the angular rate sensor breaks down was introduced. After that, the distribution of the decisive parameters was obtained using the simulation system; then non-linear model was used to replace the inaccurate linear model that is widely used in the process of identifying the extreme value distribution. In order to solve the uncertainty in the fitting process, four optimization algorithms were taken to identify the model parameters contrastively and the adaptive range particle swarm optimization (ARPSO) was found to be the best suitable algorithm. The acquired risk probability was then taken into Markov model that involves pilot mode to evaluate the compositive flight risk of control system quantitatively. The results can evaluate the dynamic reliability in some certain airplanes' control system.
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2013年第2期538-544,共7页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(60572172
61074007)
关键词
飞行风险概率
极值理论
人-机系统
角速率传感器
自适应区间粒子群
flight risk probability
extreme value theory (EVT)
pilot-aircraft system
angular rate sensor
adaptive range particle swarm optimization (ARPSO)