基于贝叶斯网络的可控飞行撞地事件量化研究  

Quantification of controlled flight into terrain event based on Bayesian network

作　　者：刘俊杰 叶英豪 杜尹岚 LIU Junjie;YE Yinghao;DU Yinlan(College of Safety Science and Engineering,CAUC,Tianjin 300300,China)

机构地区：[1]中国民航大学安全科学与工程学院,天津300300

出　　处：《中国民航大学学报》2023年第2期21-26,共6页Journal of Civil Aviation University of China

摘　　要：以可控飞行撞地(CFIT,controlled flight into terrain)事件类型信息为研究对象,依照基元事件分析法分析CFIT事件演化过程,建立了事件后果层—飞机状态层—诱发因素层三层贝叶斯网络结构模型。以2017—2019年中国民航收集的CFIT事件和航空安全网(ASN,Aviation Safety Network)近20年(2000—2019年)CFIT事件数据为样本,利用样本统计数据确立网络节点参数,采用GeNIe软件选取贝叶斯网络已知节点与目标节点进行量化分析,得到目标节点可能性排序、最大可能性目标节点及影响强度;最终得出该类事件的关键风险环节,即飞行高度控制、机组丧失情景意识、飞行保障及相应节点的量化结果,为加强此类事件风险控制提供数据支持。Choosing the information of controlled flight into terrain(CFIT)as the research object,a three-layer Bayesian network model including event consequence layer,aircraft status layer and inducing layer is established according to the analysis of the CFIT event evolution process based on basic element analysis method.Taking the CFIT event data collected by civil aviation of China from 2017 to 2019 and Aviation Safety Network(ASN)in recent 20 years(2000-2019)as samples,the parameters of network nodes are established.The known nodes and target nodes of Bayesian network are selected by GeNIe for quantitative analysis,to obtain the possibility ranking,most probable target nodes and impact intensity of target nodes.Finally,the key risk processes of the CFIT,namely flight altitude control,crew loss of situational awareness,flight support and corresponding quantitative results,are obtained to provide data support for strengthening the risk control of the CFIT events.

关 键 词：航空安全信息 信息量化 可控飞行撞地 贝叶斯网络 风险控制 

分 类 号：V328[航空宇航科学与技术—人机与环境工程] X949[环境科学与工程—安全科学]