不确定条件下航空不安全事件灵敏度分析的Monte-Carlo方法

为解决不确定条件下航空不安全事件灵敏度分析的难题，基于Bow-tie模型提出了多模式下航空安全性指标及其灵敏度测度，以轮胎爆破事件为例，采用Monte-Carlo方法计算得到安全性指标、基本事件的全局灵敏度及其分布参数的局部灵敏度。根据轮胎爆破事件仿真结果，两类灵敏度指标均随着飞行时间的增加而发生变化，尤其是在500~600 h时变化最为显著，但灵敏度重要性排序保持不变；基本事件类型是影响灵敏度的一个主要因素，电子类基本事件灵敏度测度远远小于机械类基本事件；同类型基本事件中，平均故障间隔时间不是主导因素，灵敏度大小还与失效传递的逻辑关系密切相关。算例结果表明:航空安全水平随着飞行时间逐步下降，应重点关注飞行时间为500~600 h时航空组件失效导致事故发生的可能性；航空组件的灵敏度重要性不会随着飞行时间变化，提高灵敏度较高的基本事件的可靠性水平是防范航空事故的关键。 

Sensitivity analysis for aviation insecure event using Monte-Carlo method under uncertain conditions

To solve the problem of sensitivity analysis of aviation insecure events under uncertain conditions, this paper proposes the aviation safety index and its sensitivity measurement based on the Bow-tie model. Taking the tire burst accident as an example, we calculate the aviation safety index, the global sensitivity for basic event and its local sensitivity for distribution parameters using Monte-Carlo method. According to the simulation results of the tire burst accident, both types of sensitivity indexes vary with the increasing flight hour, and the most significant change appears during 500-600 h, but with the same order of index importance. The type of basic events is the main factor affecting sensitivity, for the sensitivity of electronic events is far less than the mechanical events. In the uniform type of basic events, the mean time before failure is not the leading factor affecting the sensitivity, which has a close relationship with the failure transferring logic. The results of this example demonstrate that the safety index descends with the flight hour, and the focus for improving aviation safety is to pay attention to accident caused by aviation components failures in 500-600 h. The importance of sensitivity will not change with the flight hour, and the key of preventing aviation accident is to improve the degree of reliability for basic events with a higher sensitivity.