基于证据合成与贝叶斯网络推理的航天器发射风险评估模型  

A risk assessment model based on evidence combination and Bayesian network reasoning

作　　者：董学军[1] 杜建洲 DONG Xuejun;DU Jianzhou(Technology Department,Jiuquan Satellite Launch Center,Jiuquan 732750,China;Launch Test Station,Jiuquan Satellite Launch Center,Jiuquan 732750,China)

机构地区：[1]酒泉卫星发射中心技术部,酒泉732750 [2]酒泉卫星发射中心发测站,酒泉732750

出　　处：《系统工程理论与实践》2019年第8期2170-2178,共9页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(70971131,71071156)~~

摘　　要：为解决航天器发射风险评估中统计数据较少问题,基于证据合成和Bayesian网络推理,构建了航天器发射风险评估模型.模型将灾难征兆信息转化成专家意见,然后使用证据合成技术对专家意见进行融合,并形成风险评估结论.模型针对专家中少数不同意见,使用网络推理技术将相关的灾难征兆转化成灾难发生可能性的判断.最后,通过STS-51L实例分析验证了模型的可行性.A risk assessment model based on evidence combination and Bayesian network reasoning was built in order to solve the problem of less statistical data in spacecraft launch.The model converts the signs of disaster into the opinions of multiple experts.Then,the model integrates those expert opinions and forms risk assessment conclusions by using evidence combination techniques.In addition,based on a few expert opinions,the model uses Bayesian network reasoning to convert disaster signs into judgments of the possibility of disasters.Finally,the feasibility of the model was verified by the STS-51 L case study.

关 键 词：航天器发射 证据合成 Bayesian网络推理 风险评估模型 灾难征兆 

分 类 号：V554[航空宇航科学与技术—人机与环境工程]