基于随机集贝叶斯网的航班延误分析  

Flight Delay Analysis Based on Random Set Bayesian Network

作　　者：贡英杰 刘君强[1] 黄亮 

机构地区：[1]南京航空航天大学民航学院,南京210016

出　　处：《武汉理工大学学报（交通科学与工程版）》2017年第5期844-848,853,共6页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：国家自然科学基金与民航联合基金项目资助(U1533128)

摘　　要：针对鲜有研究的航空联盟对航班延误的影响以及航空联盟与延误成因的组合关系,基于贝叶斯网络原理和随机集理论,分别在完整样本和仅有统计信息的情况下对航空公司的航班运行情况进行贝叶斯建模.通过对加入航空联盟前后的延误情况进行对比分析、敏感性分析及与其他影响航班延误的关键因素进行组合分析,指出基于随机集贝叶斯建模的方法能为延误分析提供支持;加入航空联盟后,航班延误的整体情况得到改善;相对其他不可控的影响因素,航空联盟、航空公司自身因素的组合改善能有效地减少延误.In order to evaluate the impacts of airline alliances on flight delays and the combined effects of alliances with different factors which were barely investigated,according to Bayesian network and random set theory,the Bayesian models of flight delay analysis were separately built based on abundant samples and only statistics.Throughout the research,the extent of flight delays before and after joining the alliance was compared and their sensitivity was analyzed,as well as the combination of other key factors.The results indicate that the proposed method can strongly support the delay analyse and decision.The flight delay status tends to improve after joining the alliance.The combined improvements of the alliance and airlines can effectively ameliorate the impacts of flight delays relative to other uncontrollable factors.

关 键 词：航空联盟 航班延误 贝叶斯网络 随机集理论 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]