贝叶斯网络在科研试飞安全风险评估中的应用

详细信息    查看全文 | 推荐本文 |

英文篇名：Application of Bayesian network in the safety risk assessment of flight test 作者：孙碧晴 ; 杨旭 ; 吴双 ; 马德仲 英文作者：SUN Bi-qing;YANG Xu;WU Shuang;MA De-zhong;Quality and Airworthiness Department, Harbin Aircraft Industry Group Co., Ltd.;Higher Educational Key Laboratory for Measuring and Control Technology and Instrumentations,Harbin University of Science and Technology;Test Flight Station, Harbin Aircraft Industry Group Co., Ltd.; 关键词：风险评估 ; 因果贝叶斯网络 ; 专家判断 ; 灵敏度分析 英文关键词：risk assessment;;causal Bayesian network;;expert judgment;;sensitivity analysis 中文刊名：FHLX 英文刊名：Flight Dynamics 机构：哈尔滨飞机工业有限责任公司质量适航部;哈尔滨理工大学测控技术与仪器省高校重点实验室;哈尔滨飞机工业有限责任公司试飞站; 出版日期：2019-01-14 15:42 出版单位：飞行力学 年：2019 期：v.37;No.166 基金：黑龙江省自然科学基金资助(F201305) 语种：中文; 页：FHLX201902019 页数：5 CN：02 ISSN：61-1172/V 分类号：95-99

摘要

为提升直升机科研试飞安全水平,实现对试飞风险进行准确有效的评估,采用因果贝叶斯网络建立试飞安全风险评估系统。首先,根据直升机试飞的整体流程识别可能出现的风险事件并分类,通过风险因素间的内在关联构建贝叶斯网络拓扑结构;其次,结合专家调研数据,采用贝叶斯吐真剂算法确定各节点条件概率,利用因果贡献度和AgenaRisk软件进行灵敏度分析;最后,得到影响直升机科研试飞安全的关键风险事件。结果表明:利用因果贝叶斯网络建立的直升机科研试飞风险控制模型可以得到关键风险事件,依据分析得到的结果进行风险控制,可以满足试飞安全风险管理的需求。
        In order to improve the safety level of helicopter flight test and realize the accurate and effective evaluation of the flight test risk, this paper uses the causal Bayesian network to establish a flight test safety risk assessment system. Firstly, identify the possible risk events and classify them according to the overall process of helicopter flight test, and construct the Bayesian network topology through the internal relationship between risk factors. Then, combined with the expert survey data, use Bayesian procedural algorithm to determine the conditional probability of each node, and use causal contribution and AgenaRisk software for sensitivity analysis. Finally, the key risk events affecting the safety of helicopter scientific test flight are obtained. The results show that the risk control model of helicopter scientific research flight test established by causal Bayesian network can obtain key risk events, and the risk control based on the results obtained by the analysis can meet the requirements of flight test risk management.

引文

[1] 赵敏荣,刘兴堂,吴晓燕,等.建模与仿真中复杂假设的贝叶斯检验[J].飞行力学,2001,19(3):84-86.
    [2] 姬志伟.基于人为因素的民航飞行安全评估研究[D].南京:南京航空航天大学,2016.
    [3] 王爱文,杨敏,段华蕾.基于因果贝叶斯网络的风险建模与分析[J].系统工程与电子技术,2013,35(5):1023-1030.
    [4] 丁松滨,茹毅.基于贝叶斯网络的航空公司飞行安全系统评价[J].中国民航大学学报,2009,27(3):30-36.
    [5] 张连文,郭海鹏.贝叶斯网引论[M].北京:科学出版社,2006:21-30.
    [6] 孙旋,牛秦洲,徐和飞,等.基于贝叶斯网络的人因可靠性评价[J].中国安全科学学报,2006,16(8):22-27.
    [7] 杜红兵,王雪莉.基于贝叶斯网络的可控飞行撞地事故原因分析方法[J].安全与环境学报,2009,9(5):136-139.
    [8] 周晟瀚,杨敏,魏法杰,等.复杂装备试验安全风险评估与预警[M].北京:中国电力出版社,2016:224-235.