Prediction for flight cabin service process based on synergy of multi-Markov chains
邢志伟 1刘鹏 1李彪 1罗谦2
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 中国民航大学电子信息与自动化学院,天津 300300
2. 中国民用航空局第二研究所工程技术研究中心,四川成都 610041
折叠
摘要
为感知航班客舱保障过程各节点的动态演化机理,提出一种多马尔可夫链协同(synergy of multi-Markov chains,SMMC)的航班客舱保障过程预测方法.根据航班客舱保障的实际流程及相互约束关系,构建一种客舱保障过程节点协同的马尔可夫模型;基于历史数据作为样本并改进DBSCAN(density-based spatial clustering of applications with noise)聚类算法,设计面向客舱保障过程的DBSCAN-SMMC预测方法.选取国内某大型机场航班运行保障过程的实际运行数据开展仿真验证.研究结果表明,所提方法实现了各节点发生时刻的动态精准预测,其平均绝对误差的均值为0.606 min,均方根误差的均值为1.133 min,与其它方法相比平均绝对百分误差最少降低2％,拟合优度最大提升0.14,能够为机场运行精细化管理提供决策依据.
Abstract
To perceive the dynamic evolution mechanism of each node in the flight cabin service process,a synergy of multi-Markov chains(SMMC)for flight cabin service process prediction method was proposed.According to the actual process and mutual constraint relationship of flight cabin service,a Markov model of cabin service process node coordination was constructed.Based on historical data as samples and to improve the DBSCAN clustering algorithm,a DBSCAN-SMMC prediction method for cabin service process was designed.The actual operation data of the flight operation service process of a large domestic airport were selected for simulation verification.Results show that the proposed method realizes the dynamic and accurate prediction for the occurrence time of each node.The mean absolute error is 0.606 min,and the mean root mean square error is 1.133 min.Compared with other methods,the mean absolute percentage error is reduced by at least 2％,and the goodness of fit is improved by 0.14.It can provide decision-making basis for the refined management of airport operations.
关键词
航空运输/客舱保障/节点预测/马尔可夫链/聚类算法/动态预测/仿真
Key words
air transportation/cabin service/node prediction/Markov chain/clustering algorithm/dynamic prediction/simula-tion
维普
万方数据