基于贝叶斯网络的试飞风险评估方法研究

Researchon Risk Assessment Method of Flight Test Basedon Bayesian Network

建立了基于故障树的贝叶斯网络的风险评估模型;综合运用正向推理与反向推理方法,实现了对试飞风险的分析;结合湾流宇航公司G650公务机试飞事故的实例开展风险概率的计算,并借助贝叶斯建模软件GENIE实现图形化描述,找出了试飞风险的关键致因为人的因素与湾流宇航公司的管理因素,与采用传统故障树分析法得到的结论相同,从而验证了贝叶斯方法的可靠性,最后提出了改进措施来减少试飞科目的风险,提高试飞安全性。

Bayesian network risk assessment model based on fault tree was established based on the system engineering theory of“man-machine-management-environment”.By using forward reasoning and backward reasoning methods,the risk of flight test was analyzed.Combining with the example of G650 civil aircraft flight test accident of Gulfstream Aerospace Company,this paper calculated the risk probability,and realized the graphical description with the help of GENIE,a Bayesian modelling software.It finds out that the key causes of flight test risk are human factors and management factors of Gulfstream Aerospace Company.The conclusion is the same as that obtained by traditional fault tree analysis method.The reliability of Bayesian method was verified.Finally,some improvement measures were put forward to reduce the risk of flight test subjects and improve the safety of flight test.