-
摘要:
针对冰雪气象下除冰保障过程的精细化管理及预测精度低下的问题，提出了一种基于时空关联动态贝叶斯网络的飞机地面除冰保障过程动态预测方法。在系统性分析除冰保障过程的基础上，设计了面向离港除冰队列的时空关联节点判别方法，基于K最近邻算法简化关联节点并构建了变结构动态贝叶斯网络模型。基于核注意力机制的除冰保障节点先验概率密度估计方法，结合条件概率更新结果构建了面向不同状态的飞机地面除冰保障过程动态预测方法。实验结果表明：所提方法在考虑除冰保障演化不确定性的基础上实现了各节点的动态预测，平均绝对误差为2.34 min，整体精度相比静态贝叶斯网络方法最大提升8.66%，为地面除冰运行战术组织及决策控制提供了有效依据。
-
关键词:
- 航空运输 /
- 动态预测 /
- 时空关联动态贝叶斯网 /
- 核注意力机制 /
- 地面除冰保障过程
Abstract:Aiming at the problem of fine management and low prediction accuracy of deicing operation process under ice and snow weather, a prediction method for aircraft ground deicing operation process based on the temporal and spatial correlation dynamic Bayesian network is proposed. A spatial-temporal correlation node identification method for departure deicing queue is developed based on a systematic analysis of the deicing operation process. The correlation node is then simplified using the K-nearest neighbor algorithm, and a dynamic Bayesian network model with variable structure is created. A priori probability density estimation method for deicing operation nodes based on kernel attention mechanism is studied. Combined with the conditional probability updating results, a dynamic prediction method for aircraft ground deicing support process for different states is constructed. A dynamic prediction approach for the aircraft ground deicing support process is built using the conditional probability updating findings in combination. The average absolute error is 2.34 min, and the whole accuracy is increased by 8.66% compared with static Bayesian network method, which can provide an effective decision-making basis for the tactical organization and control of ground deicing operations.
-
图 2 飞机地面除冰保障过程拓扑结构
Figure 2. Topological structure of aircraft ground deicing operation process
图 5 关键除冰保障过程节点先验概率模型
Figure 5. Prior probability model of key deicing operation process node
图 6 除冰保障节点条件概率模型演化过程
Figure 6. Evolution process for conditional probability model of deicing operation node
图 7 不同除冰保障过程预测结果
Figure 7. Prediction results of different deicing operation processes
图 8 多除冰保障过程预测结果误差分析
Figure 8. Error analysis of prediction results for multiple deicing operation process
图 9 不同概率更新方式预测结果的平均绝对误差对比
Figure 9. Comparison of mean absolute error of prediction results for different probability updating methods
图 10 不同方法的离港起飞节点预测结果对比
Figure 10. Comparison of departure & take-off node prediction results for different methods
表 1 不同模式的除冰保障过程
Table 1. Deicing operation process of different modes
模式 里程碑节点 A-CDM代码 空间状态 机位除冰 进入除冰队列，允许登机 PPT 机位 登机结束 EPT 机位 撤轮挡 OBT 机位 除冰开始 CZT 机位 除冰结束 EZT 机位 推出滑行 SAT 滑行道 离港起飞 TOT 跑道端头 集中除冰 进入除冰队列，允许登机 PPT 机位 登机结束 EPT 机位 撤轮挡/推出 SAT 机位 到达除冰等待点 WZT 滑行道 除冰开始 CZT 除冰位 除冰结束 EZT 除冰位 离港起飞 TOT 跑道端头 注： 若代码前缀为A，表示实际时间，即该节点已发生或正在发生；若代码为E，表示估计时间，即该节点未发生。 
下载: 导出CSV
表 2 除冰保障过程属性数据样例
Table 2. Data sample of aircraft deicing operation process attributes
属性 状态或数值 出港日期 2020-12-12 出港航班号 CA1805 机型 B738 出港机位 140 起飞跑道 36R 准点率/% 60 除冰模式 双车 作业模式 关车 除冰容量/（架次·h−1） 27 除冰类型 机位
下载: 导出CSV
表 3 除冰保障过程数据样例
Table 3. Data sample of aircraft deicing operation process
除冰保障过程节点 时刻 IBT 05:24 PPT 06:06 EPT 06:26 OBT 06:51 CZT 06:50 EZT 07:00 SAT 07:07 TOT 07:17
下载: 导出CSV
表 4 单除冰保障过程预测结果
Table 4. Prediction results of single deicing operation process
除冰保障节点 初始预测时间/min 最终预测时间/min 实际时间/min PPT 40.22 40.22 42 EPT 62.11 60.31 58 OBT 78.28 76.32 77 CZT 88.94 87.94 86 EZT 99.37 96.41 96 SAT 98.92 105.81 103 TOT 108.83 111.68 113
下载: 导出CSV
表 5 单除冰保障过程预测结果误差分析
Table 5. Error analysis of prediction results for single deicing operation process
除冰保障节点 平均绝对百分误差/% 平均绝对误差/min PPT 4.24 1.78 EPT 5.34 3.21 OBT 0.86 0.67 CZT 2.61 2.25 EZT 2.47 2.37 SAT 2.85 2.94 TOT 2.81 3.17
下载: 导出CSV
-
[1] 鲍帆, 蒋伟煜. 基于机场协同决策(A-CDM)的除冰管理研究[C]//第一届空中交通管理系统技术学术年会. 南京: 中国指挥与控制学会, 2018: 330-334.BAO F, JIANG W Y. Research on deicing management based on airport CDM[C]//The First Annual Conference on Air Traffic Management System Technology. Nanjing: Chinese Institute of Command and Control, 2018: 330-334(in Chinese). [2] 王立文, 李彪, 邢志伟, 等. 过站航班地面保障过程动态预测[J]. 北京航空航天大学学报, 2021, 47(6): 1095-1104. doi: 10.13700/j.bh.1001-5965.2020.0165WANG 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). doi: 10.13700/j.bh.1001-5965.2020.0165 [3] GUI G, LIU F, SUN J L, et al. Flight delay prediction based on aviation big data and machine learning[J]. IEEE Transactions on Vehicular Technology, 2020, 69(1): 140-150. doi: 10.1109/TVT.2019.2954094 [4] GARCIA-HERAS J, SOLER M, GONZALEZ-ARRIBAS D, et al. Robust flight planning impact assessment considering convective phenomena[J]. Transportation Research Part C: Emerging Technologies, 2021, 123: 102968. [5] LI M Z, KARTHIK G, HAMSA B. Graph signal processing techniques for analyzing aviation disruptions[J]. Transportation Science, 2021, 16(3): 1-22. [6] 王春政, 胡明华, 杨磊, 等. 基于Agent模型的机场网络延误预测[J]. 航空学报, 2021, 42(7): 324604.WANG C Z, HU M H, YANG L, et al. Airport network delay prediction based on Agent model[J]. Acta Aeronautica et Astronautica Sinica, 2021, 42(7): 324604(in Chinese). [7] BADRINATH S, BALAKRSHNAN H, MA J, et al. Comparative analysis of departure metering at United States and European airports[J]. AIAA Journal of Air Transportation, 2020, 28(3): 93-104. doi: 10.2514/1.D0179 [8] SALEH Y, SAMI E, FRANCOIS M. Computational fluid dynamics investigation of transient effects of aircraft ground deicing jets[J]. Journal of Thermophysics and Heat Transfer, 2019, 33(1): 117-127. doi: 10.2514/1.T5428 [9] 张政, 陈艳艳, 梁天闻. 基于网约车数据的城市区域出行时空特征识别与预测研究[J]. 交通运输系统工程与信息, 2020, 20(3): 89-94. doi: 10.16097/j.cnki.1009-6744.2020.03.014ZHANG Z, CHEN Y Y, LIANG T W. Regional travel demand mining and forecasting using car-hailing order records[J]. Journal of Transportation Systems Engineering and Information Technology, 2020, 20(3): 89-94(in Chinese). doi: 10.16097/j.cnki.1009-6744.2020.03.014 [10] 邢志伟, 刘洪恩, 李彪, 等. 基于时空关联网络的机场机位运行过程建模[J]. 系统工程与电子技术, 2021, 43(3): 722-730. doi: 10.12305/j.issn.1001-506X.2021.03.16XING Z W, LIU H E, LI B, et al. Modelling for airport gate operation process based on relational spatio-temporal network[J]. System Engineering and Electronics, 2021, 43(3): 722-730(in Chinese). doi: 10.12305/j.issn.1001-506X.2021.03.16 [11] FRIDERKOS O, OLIVE M, BARANGER E, et al. A non-intrusive space-time interpolation from compact stiefel manifolds of parametrized rigid-viscoplastic FEM problems[J]. Computational Mechanics, 2021, 68(4): 861-883. doi: 10.1007/s00466-021-02050-0 [12] TEHRANI A F, YEH H G, KWON S C. BER performance of space-time parallel ICI cancellation of OFDM in MIMO power line communications[J]. IEEE Systems Journal, 2021, 15(2): 1742-1752. doi: 10.1109/JSYST.2020.2968542 [13] HAN X, HSIEH C, KO S. Spatial modeling approach for dynamic network formation and interactions[J]. Journal of Business and Economic Statistics, 2021, 39(1): 120-135. doi: 10.1080/07350015.2019.1639395 [14] MONDO G, RODRIGUEZ M, CLARAMUNT C, et al. Modeling consistency of spatio-temporal graphs[J]. Data & Knowledge Engineering, 2013, 84(3): 59-80. [15] 杨晓玲, 冯山, 袁钟. 基于相对距离的反k近邻树离群点检测[J]. 电子学报, 2020, 48(5): 937-945. doi: 10.3969/j.issn.0372-2112.2020.05.014YANG X L, FENG S, YUAN Z. Outlier detection based on reversed k-nearest neighborhood MST of relative distance measure[J]. Acta Electronica Sinica, 2020, 48(5): 937-945(in Chinese). doi: 10.3969/j.issn.0372-2112.2020.05.014 [16] XU B, SUN F C. Composite intelligent learning control of strict-feedback systems with disturbance[J]. IEEE Transactions on Cybernetics, 2018, 48(2): 730-741. doi: 10.1109/TCYB.2017.2655053 [17] 娄文忠, 赵悦岑, 冯恒振, 等. 基于贝叶斯网络的MEMS安全系统可靠性分析[J]. 北京理工大学学报, 2021, 41(9): 952-960. doi: 10.15918/j.tbit1001-0645.2020.065LOU W Z, ZHAO Y C, FENG H Z, et al. Reliability analysis on MEMS S&A device based on Bayesian network[J]. Transactions of Beijing Institute of Technology, 2021, 41(9): 952-960(in Chinese). doi: 10.15918/j.tbit1001-0645.2020.065 [18] 廖华年, 徐新. 基于注意力机制的跨分辨率行人重识别[J]. 北京航空航天大学学报, 2021, 47(3): 605-612. doi: 10.13700/j.bh.1001-5965.2020.0471LIAO H N, XU X. Cross-resolution person re-identification based on attention mechanism[J]. Journal of Beijing University of Aeronautics and Astronautics, 2021, 47(3): 605-612(in Chinese). doi: 10.13700/j.bh.1001-5965.2020.0471 [19] KARDAKIS S, PERIKOS I, GRIVOKOPOULOU F, et al. Examining attention mechanisms in deep learning models for sentiment analysis[J]. Applied Sciences, 2021, 11(9): 3883. doi: 10.3390/app11093883 [20] YAN J, PENG Z, YIN H, et al. Trajectory prediction for intelligent vehicles using spatial-attention mechanism[J]. IET Intelligent Transport Systems, 2020, 14(4): 1855-1863. [21] LAFHRISSI F, DOUZI S, DOUZI K, et al. IDS-attention: An efficient algorithm for intrusion detection systems using attention mechanism[J]. Journal of Big Data, 2021, 8(1): 1-21. doi: 10.1186/s40537-020-00387-6 [22] PRECUP S A, GELLERT A, MATEI A, et al. Towards an assembly support system with dynamic Bayesian network[J]. Applied Sciences, 2022, 12(3): 985. doi: 10.3390/app12030985 [23] 沈琳, 于劲松, 唐荻音, 等. 图模型与学习算法结合的贝叶斯网络自动建模[J]. 北京航空航天大学学报, 2016, 42(7): 1486-1493. doi: 10.13700/j.bh.1001-5965.2015.0445SHEN L, YU J S, TANG D Y, et al. Automatic learning of Bayesian network structure using graph model and learning algorithm[J]. Journal of Beijing University of Aeronautics and Astronautics, 2016, 42(7): 1486-1493(in Chinese). doi: 10.13700/j.bh.1001-5965.2015.0445 -
下载:
点击查看大图
计量
- 文章访问数: 435
- HTML全文浏览量: 36
- PDF下载量: 12
- 被引次数: 0
