-
摘要:
对航班地面保障过程进行精准预测是实现航班精细化管理、提升机场协同决策（A-CDM）系统管理效能的关键。为此，提出一种基于级联多输出梯度提升回归树模型的航班地面保障多节点动态预测方法。通过搭建级联框架实现了不同保障进度之间预测信息的传递和预测结果的更新，基于可进行多节点预测的梯度提升回归树设计了地面保障过程动态预测算法，以典型繁忙机场为对象构建了航班基础属性与层级信息传递两大类特征集。结果表明：所提方法有效实现了地面保障各关键节点完成时间的动态预测，初始预测各节点±5 min预测精度均达到80%以上，随着保障过程推进模型预测性能逐步提升，超过60%的节点±5 min最终预测精度超过95%，为提升航班运行的可预测性和机场多主体协同决策能力提供有效方法支撑。
Abstract:Accurate prediction of flight ground service is the key to achieving fine flight management and improving management efficiency of the airport collaborative decision making (A-CDM) system. Therefore, a multi-node dynamic prediction method for flight ground service based on a cascaded multi-output gradient boosting regression tree model was proposed. The cascaded framework was built to realize the prediction information transmission and result updates between different service schedules. The dynamic prediction algorithm of flight ground service was designed based on gradient boosting regression tree which could be used for multi-node prediction. By taking a typical busy airport as an object, a feature set was constructed, covering flight basic attributes and level information transmission. The results show that the proposed method can effectively realize the dynamic prediction of key node completion time in flight ground service. The initial prediction accuracy of each node within ±5 min reaches more than 80%, and the prediction performance gradually improves as the flight ground service continues. The final prediction accuracy of over 60% of nodes within ±5min exceeds 95%. It provides effective method support for improving the flight operation predictability and the collaborative decision making ability of multi-agents in airports.
-
图 2 级联框架与地面保障多节点动态预测过程的映射关系
Figure 2. Mapping relationship between cascaded framework and multi-node dynamic prediction of flight ground service
图 3 级联多输出梯度提升回归树算法框架
Figure 3. Algorithm framework for cascaded multi-output gradient boosting regression tree
图 4 过站时间对地面保障节点完成时间的影响
Figure 4. Effect of turn-round time on node completion time in flight ground service
图 5 航班密度对地面保障节点完成时间的影响
Figure 5. Effect of flight density on node completion time in flight ground service
表 1 航班地面保障运行数据
Table 1. Data of flight ground service operations
字段 样例 字段 样例 日期 2021-3-1 上轮挡 14:32 航班号 9C6136|9C8569 开客门 14:36 航班属性 正班 下客开始/结束 14:38|14:43 STA 14:35 保洁开始/结束 14:51|15:02 STD 15:50 供油开始/结束 15:24|15:37 机型 A320 配餐开始/结束 15:28|15:29 机位 55|61 登机口开启 15:34 航线性质 国内|国际 登机口关闭 15:50 地面代理 CQH 关客门 15:57 旅客总数 175|165 撤轮挡 16:07 注：STA为计划到港时间，STD为计划离港时间，CQH为国际民用航空组织规定的春秋航空三字代码。 
下载: 导出CSV
表 2 航班地面保障节点完成时间转化结果
Table 2. Conversion result of node completion time in flight ground service
节点 转化后节点完成
时间/min节点 转化后节点完成
时间/min上轮挡 0 配餐结束 57 开客门 4 登机口开启 62 下客结束 11 登机口关闭 78 保洁结束 30 关客门 85 供油结束 65 撤轮挡 95
下载: 导出CSV
表 3 第1层级关键节点完成时间预测结果
Table 3. Prediction results of key node completion time at Level 1
关键节点 MAE/min R2 ±3 min预测精度/% ±5 min预测精度/% 开客门 1.60 0.85 85.20 95.39 下客结束 2.46 0.80 67.32 87.98 供油结束 7.66 0.72 26.40 41.61 登机口开启 2.93 0.96 56.36 82.47 登机口关闭 2.96 0.97 54.79 82.37 关客门 2.94 0.97 55.44 83.23 撤轮挡 3.11 0.97 52.97 80.04 注：预测精度是指模型输出的时间与实际滑入时间的差值在某一设定范围内的数量与总预测样本数之比。
下载: 导出CSV
表 4 层级序号与进行该层级时的对应保障进程
Table 4. Level number and service process at corresponding levels
层级序号 对应保障进程 层级序号 对应保障进程 1 上轮挡 4 登机口开启 2 开客门 5 登机口关闭 3 下客结束 6 关客门
下载: 导出CSV
表 5 不同模型在第1层级时的撤轮挡时间预测结果对比
Table 5. Comparison of off-block time prediction results of different models at Level 1
预测模型 MAE/min R2 ±3 min预测精度/% ±5 min预测精度/% CNN 5.77 0.87 34.83 53.64 DT 3.62 0.85 59.58 73.00 SVR 5.31 0.53 63.60 70.23 RF 3.83 0.77 55.03 78.42 GBRT 3.11 0.97 52.97 80.04
下载: 导出CSV
表 6 所有节点在不同层级的动态预测结果
Table 6. Dynamic prediction results for all nodes at different levels
保障进程 层级 MAE/min R2 ±3 min预测精度/% ±5 min预测精度/% 开客门 1 1.60 0.85 85.20 95.39 下客结束 1 2.46 0.80 67.32 87.98 2 1.92 0.87 79.65 95.27 登机口开启 1 2.93 0.96 56.36 82.47 2 2.93 0.96 57.41 83.06 3 2.92 0.96 58.37 83.14 登机口关闭 1 2.96 0.97 54.79 82.37 2 3.02 0.96 55.91 81.11 3 2.99 0.97 55.69 82.19 4 2.56 0.97 64.05 88.61 关客门 1 2.94 0.97 55.44 83.23 2 3.01 0.97 55.89 81.91 3 2.97 0.97 56.88 82.21 4 2.5 0.98 66.06 88.93 5 0.96 0.98 97.04 99.00 撤轮挡 1 3.11 0.97 52.97 80.04 2 3.11 0.97 54.59 79.96 3 3.11 0.97 54.02 79.89 4 2.62 0.98 64.46 86.60 5 2.11 0.98 75.22 94.54 6 2.01 0.98 77.34 95.29
下载: 导出CSV
-
[1] MILLER T, LU B, STERLING L, et al. Requirements elicitation and specification using the agent paradigm: the case study of an aircraft turnaround simulator[J]. IEEE Transactions on Software Engineering, 2014, 40(10): 1007-1024. doi: 10.1109/TSE.2014.2339827 [2] GARCÍA ANSOLA P, GARCÍA HIGUERA A, OTAMENDI F J, et al. Agent-based distributed control for improving complex resource scheduling: application to airport ground handling operations[J]. IEEE Systems Journal, 2014, 8(4): 1145-1157. doi: 10.1109/JSYST.2013.2272248 [3] 冯霞, 张鑫, 陈锋. 飞机过站上客过程持续时间分布[J]. 交通运输工程学报, 2017, 17(2): 98-105. doi: 10.3969/j.issn.1671-1637.2017.02.011FENG X, ZHANG X, CHEN F. Boarding duration distribution of aircraft turnaround[J]. Journal of Traffic and Transportation Engineering, 2017, 17(2): 98-105(in Chinese). doi: 10.3969/j.issn.1671-1637.2017.02.011 [4] SCHULTZ M, EVLER J, ASADI E, et al. Future aircraft turnaround operations considering post-pandemic requirements[J]. Journal of Air Transport Management, 2020, 89: 101886. doi: 10.1016/j.jairtraman.2020.101886 [5] POHLING O, SCHIER-MORGENTHAL S, LORENZ S. Looking into the crystal ball: how automated fast-time simulation can support probabilistic airport management decisions[J]. Aerospace, 2022, 9(7): 389. doi: 10.3390/aerospace9070389 [6] EVLER J, ASADI E, PREIS H, et al. Airline ground operations: schedule recovery optimization approach with constrained resources[J]. Transportation Research Part C: Emerging Technologies, 2021, 128: 103129. doi: 10.1016/j.trc.2021.103129 [7] SHEIBANI K. Scheduling aircraft ground handling operations under uncertainty using critical path analysis and Monte Carlo simulation[J]. International Journal of Business Strategy and Automation, 2020, 1(1): 37-45. doi: 10.4018/IJBSA.2020010103 [8] 邢志伟, 吴兵, 罗晓, 等. 基于AFSA-RVM的航班保障服务时间状态估计[J]. 计算机仿真, 2020, 37(8): 35-39. doi: 10.3969/j.issn.1006-9348.2020.08.008XING Z W, WU B, LUO X, et al. State estimation about service time of flight support based on AFSA-RVM[J]. Computer Simulation, 2020, 37(8): 35-39(in Chinese). doi: 10.3969/j.issn.1006-9348.2020.08.008 [9] 邢志伟, 韩大浩, 罗谦. 基于改进GA的神经网络航班保障时间估计[J]. 计算机工程与设计, 2020, 41(1): 107-114.XING Z W, HAN D H, LUO Q. Estimation of flight support time based on improved GA neural network[J]. Computer Engineering and Design, 2020, 41(1): 107-114(in Chinese). [10] 王立文, 李彪, 邢志伟, 等. 过站航班地面保障过程动态预测[J]. 北京亚洲成人在线一二三四五六区学报, 2021, 47(6): 1095-1104.WANG L W, LI B, XING Z W, et al. Dynamic prediction of ground support process for transit flight[J]. Journal of Beijing University of Aeronautics and Astronautics, 2021, 47(6): 1095-1104(in Chinese). [11] MA Y L, JAMES HOUSDEN R, FAZILI A, et al. Real-time registration of 3D echo to X-ray fluoroscopy based on cascading classifiers and image registration[J]. Physics in Medicine and Biology, 2021, 66(5): 055019. doi: 10.1088/1361-6560/abe420 [12] TIAN M Z, LIU L, LU J Y, et al. Vehicle recognition based on haar features and adaboost cascade classifier[J]. Journal of Physics: Conference Series, 2022, 2303(1): 012052. doi: 10.1088/1742-6596/2303/1/012052 [13] 王宁, 曹萃文. 基于XGBoost模型的炼油厂氢气网络动态多输出预测模型[J]. 华东理工大学学报(自然科学版), 2020, 46(1): 77-83.WANG N, CAO C W. A dynamic multi-output prediction model of the hydrogen network in a real-world refinery based on XGBoost model[J]. Journal of East China University of Science and Technology, 2020, 46(1): 77-83(in Chinese). [14] ALHAKEEM Z M, JEBUR Y M, IMRAN H, et al. Prediction of ecofriendly concrete compressive strength using gradient boosting regression tree combined with gridsearchCV hyperparameter-optimization techniques[J]. Materials, 2022, 15(21): 7432. doi: 10.3390/ma15217432 [15] HUANG Y F, LIU Y H, LI C H, et al. GBRTVis: online analysis of gradient boosting regression tree[J]. Journal of Visualization, 2019, 22(1): 125-140. doi: 10.1007/s12650-018-0514-2 [16] WANG T, HU S H, JIANG Y. Predicting shared-car use and examining nonlinear effects using gradient boosting regression trees[J]. International Journal of Sustainable Transportation, 2021, 15(12): 893-907. doi: 10.1080/15568318.2020.1827316 [17] 中国民用航空局. 机场协同决策系统技术规范: MH/T 6125—2022[S]. 北京: 中国民用航空局, 2022.Civil Aviation Administration of China. Technical specifications for airport collaborative decision-making system: MH/T 6125—2022[S]. Beijing: Civil Aviation Administration of China, 2022(in Chinese). -
下载:
点击查看大图
计量
- 文章访问数: 582
- HTML全文浏览量: 117
- PDF下载量: 14
- 被引次数: 0
