基于强化学习方法的机场跑道最优起降窗口研究

唐欣珂; 张舵; 张昆

doi:10.12382/bgxb.2025.1032

您当前的位置：

首页 >

文章列表页 >

基于强化学习方法的机场跑道最优起降窗口研究

更新时间：2026-04-20

- 基于强化学习方法的机场跑道最优起降窗口研究
- Research on Optimal Takeoff and Landing Windows of Airport Runways Based on Reinforcement Learning
- 兵工学报 2026年
- 作者机构：
  
  国防科技大学理学院，湖南长沙410073、TakeoffandLandingWindow，TLW
- 作者简介：
- 基金信息：
- DOI：10.12382/bgxb.2025.1032
  中图分类号： O38
- 收稿：2025-11-24，
  
  网络首发：2026-04-20，
- 稿件说明：
移动端阅览
唐欣珂，张舵，张昆. 基于强化学习方法的机场跑道最优起降窗口研究[J/OL]. 兵工学报, 2026(2026-04-20). https://doi.org/10.12382/bgxb.2025.1032.

TANG X K, ZHANG D, ZHANG K. Research on optimal takeoff and landing windows of airport runways based on reinforcement learning[J/OL]. Acta Armamentarii, 2026(2026-04-20). https://doi.org/10.12382/bgxb.2025.1032. (in Chinese)
唐欣珂，张舵，张昆. 基于强化学习方法的机场跑道最优起降窗口研究[J/OL]. 兵工学报, 2026(2026-04-20). https://doi.org/10.12382/bgxb.2025.1032. DOI：

TANG X K, ZHANG D, ZHANG K. Research on optimal takeoff and landing windows of airport runways based on reinforcement learning[J/OL]. Acta Armamentarii, 2026(2026-04-20). https://doi.org/10.12382/bgxb.2025.1032. (in Chinese) DOI：

摘要

机场跑道封锁是毁伤评估领域的重要研究方向，其核心在于依据起降窗口(Takeoff and Landing Window，TLW)判据高效定位不同毁伤场景下的最优TLW，而经典遍历算法虽可通过遍历离散位置与角度获取全局最优解，但其精度与效率均受离散尺寸制约。引入强化学习算法，通过构建考虑弹坑分布与跑道几何约束的状态-动作空间，并设计分段多层奖励函数，采用近端策略优化算法实现智能体的策略训练。对比结果表明，在多种毁伤度场景下，强化学习方法在连续跑道空间内获得的TLW长度与遍历算法最优解的误差均不超过0.49%，角度偏差不大于1.1°，展现出良好的鲁棒性与精度。通过奖励函数对比与训练曲线分析证明，所提出的分段多层奖励模型相较于单值基础奖励模型，能够显著提升复杂场景下的精度与稳定性。研究结果表明，强化学习方法在机场跑道封锁毁伤问题研究中具有重要价值。

Abstract

Airport runway blockade is an important research direction in the field of damage assessment. Its core lies in efficiently locating the optimal Takeoff and Landing Window (TLW) under different damage scenarios according to TLW criteria. Although classical traversal algorithms can obtain the global optimal solution by traversing discrete positions and angles

their accuracy and efficiency are both restricted by the discrete size.This paper introduces a Reinforcement Learning (RL) algorithm

which constructs a state-action space model considering crater distribution and runway geometric constraints

designs a Multi-Level Reward Function (MRF)

and adopts the Proximal Policy Optimization (PPO) algorithm to implement policy training. Comparison results show that under various damage degree scenarios

the error between the TLW length obtained by the RL method in the continuous runway space and the optimal solution of the ES algorithm is no more than 0.5%

and the angular deviation is no more than 1.1°

exhibiting excellent robustness and accuracy. Comparisons of reward functions and analysis of training curves demonstrate that the proposed MRF can significantly improve the policy convergence speed and stability compared with the Baseline Reward Function (BRF). The research results indicate that RL has important value in the research on airfield runway closure damage assessment.

关键词

Keywords

references

杨云斌, 李小笠. 机场跑道目标易损性分析方法研究[J]. 弹箭与制导学报, 2010, 30(2): 141-144. YANG Y B, LI X L. Study on vulnerability analysis method for runway[J]. Journal of Projectiles, Rockets, Missiles and Guidance, 2010, 30(2): 141-144. (in Chinese)

苏高峰,王宝树,苗启广. 机场跑道的打击效果评估研究[J]. 火力与指挥控制,2008,33(11): 46-47,52. SU G F, WANG B S, MIAO Q G. A study of the runway battle damage assessment[J]. Fire Control & Command Control,2008,33(11): 46-47,52. (in Chinese)

梁振刚, 刘阔, 潘浙平, 等. 子母弹对机场跑道的封锁效能评估[J]. 兵器装备工程学报, 2021, 42(1): 35-38. LIANG Z G, LIU K, PAN Z P, et al. Evaluation of blockade effectiveness of submunition against airport runway[J]. Journal of Ordnance Equipment Engineering, 2021, 42(1): 35-38. (in Chinese)

仵可, 李浩, 陈博, 等. 重型侵彻子母弹对机场跑道的封锁效能仿真研究[J]. 现代应用物理, 2025, 16(2): 021001. WU K, LI H, CHEN B, et al. Simulation study on blocking effectiveness of heavy penetration cluster bombs against airport runways[J]. Modern Applie Physics, 2025, 16(2): 021001. (in Chinese)

张兵, 王晓琨. 机场跑道打击预案评估方法研究[J]. 信息化研究, 2023, 49(3): 25-30. ZHANG B, WANG X K. Research on the evaluation method of attack plan to airport runway[J]. Informatization Research, 2023, 49(3): 25-30. (in Chinese)

刘云辉, 冯源, 高月光, 等. 子母弹反机场跑道封锁概率快速计算方法[J]. 火力与指挥控制, 2022, 47(9): 159-162,169. LIU Y H, FENG Y, GAO Y G, et al. Research on fast calculation method of probability of airstrips blockaded by cluster munitions[J]. Fire Control & Command Control, 2022, 47(9): 159-162,169. (in Chinese)

梁杰, 任君, 李磊, 等. 基于典型几何形状精确回归的机场跑道检测方法[J]. 兵工学报, 2020, 41(10): 2045-2054. LIANG J, REN J, LI L, et al. Airport runway detection agorithm based on accurate regression of typical geometric shapes[J]. Acta Armamentarii, 2020, 41(10): 2045-2054. (in Chinese)

计宏磊, 杨清文. 基于图像分析的机场跑道功能性毁伤评估[J]. 计算机应用, 2016, 36(增刊1): 165-168,173. JI H L, YANG Q W. Airport runway functional damage assessment based on image analysis[J]. Journal of Computer Applications, 2016, 36(S1): 165-168,173. (in Chinese)

王志军, 陈超. 机载布撒器对机场跑道封锁效率的monte-carlo仿真[J]. 弹箭与制导学报, 2001,21(2): 73-76,80. WANG Z J, CHEN C.The monte-carlo simulation of interdiction effectiveness of airborne dispenser upon runway[J]. Journal of Projectiles, Rockets, Missiles and Guidance, 2001,21(2): 73-76,80. (in Chinese)

舒健生, 王运吉, 朱昱. 子母弹抛撒半径对机场毁伤效果影响分析[J]. 火力与指挥控制, 2001,26(3): 49-50. SHU J S, WANG Y J, ZHU Y. The effect of the cluster warhead dispersion radius on airport damage efficiency analysis[J]. Fire Control & Command Control, 2001,26(3): 49-50. (in Chinese)

操龙平, 陈谋, 周同乐. 基于改进卷积神经网络的机场跑道封锁效能评估[J]. 航空兵器, 2025, 32(2): 128-136. CAO L P, CHEN M, ZHOU T L. Evaluation of airport runway blockade effectiveness based on improved convolutional neural network[J]. Aero Weaponry, 2025, 32(2): 128-136. (in Chinese)

张舵, 宋歌, 陈琨, 等. 基于功能恢复时间的机场跑道目标毁伤评估方法[J]. 指挥控制与仿真, 2023, 45(3): 94-98. ZHANG D, SONG G, CHEN K, et al. Damage assessment method for airport runway targets based on functional recovery time[J]. Command Control & Simulation, 2023, 45(3): 94-98. (in Chinese)

韦兰柯, 陈榕, 赵英清, 等. 基于毁伤效能理论的弹药需求测算[J]. 舰船电子工程, 2023, 43(4): 131-135. WEI L K, CHEN R, ZHAO Y Q, et al. Ammunition demand measurement based on the theory of damage effectiveness[J]. Ship Electronic Engineering, 2023, 43(4): 131-135. (in Chinese)

黄龙华, 冯顺山, 樊桂印. 封锁型子母弹对飞机从机场跑道强行起降的封锁效能分析[J]. 弹箭与制导学报, 2008,28(4): 139-142. HUANG L H, FENG S S, FAN G Y. Analysis on the interdiction effectiveness of interdiction-target submunition on airplane force to fly or land from airport runway [J]. Journal of Projectiles, Rockets, Missiles and Guidance, 2008,28(4): 139-142. (in Chinese)

于志刚, 邓可教, 刘娟娟, 等. 基于最小封控时间的机场跑道毁伤评估方法[J]. 火力与指挥控制, 2024, 49(5): 158-164,171. YU Z G, DENG K J, LIU J J, et al. Airport runway damage assessment method based on minimum blocking time[J]. Fire Control & Command Control, 2024, 49(5): 158-164,171. (in Chinese)

WEI W L, CHEN Y Q, WANG Z Q, et al. Comparative study on damage effects of penetration and explosion modes on airport runway[J]. Construction and Building Materials, 2024, 411: 134169.

侯鹏, 裴扬, 赵倩, 等. 典型机场目标的毁伤效果研究[J]. 航空科学技术,2023,34(12): 100-110. HOU P, PEI Y, ZHAO Q, et al. Research on the damage effect of typical airport targets[J]. Aeronautical Science & Technology,2023,34(12): 100-110. (in Chinese)

于志刚, 邓可教, 刘娟娟, 等.面向机场的反跑道火箭弹弹着点分配模型[J].运筹与模糊学,2024,14(1):1130-1137. YU Z G, DENG K J, LIU J J, et al. A model for allocation of impact points of anti runway rocket for airports[J]. Operations Research and Fuzziology,2024,14(1):1130-1137. (in Chinese)

YE L T, CHI C, LIU Z W,et al. Online actuator selection and controller design for linear quadratic regulation with an unknown system model[J]. IEEE Transactions on Automatic Control, 2025, 70(1): 18-33.

LI Y, WANG Y H, TAN X Y.Self-imitation guided goal-conditioned reinforcement learning[J]. Pattern Recognition, 2023, 144: 109845.

LI Y, WANG Y H, TAN X Y. Highly valued subgoal generation for efficient goal-conditioned reinforcement learning[J]. Neural Networks, 2024, 181: 106825.

SILVER D, HUANG A, MADDISON C J, et al. Mastering the game of go with deep neural networks and tree search[J/OL]. Nature, 2016, 529(7587): 484-489.

SILVER D, SCHRITTWIESER J, SIMONYAN K, et al. Mastering the game of go without human knowledge[J]. Nature, 2017, 550(7676): 354-359.

LUO B, WU H N, HUANG T W. Off-policy reinforcement learning for H∞ control design[J]. IEEE Transactions on Cybernetics, 2015,45(1): 65-76.

DUAN Y, CHEN X, HOUTHOOFT R, et al. Benchmarking deep reinforcement learning for continuous control: arXiv:1604.06778[R].Ithaca,NY,US:Cornell University,2016:1604.06778.

HEYSE J, VEGA M T, DE BACKERE F, et al. Contextual bandit learning-based viewport prediction for 360 video[C]//Proceedings of 2019 IEEE Conference on Virtual Reality and 3D User Interfaces. Osaka, Japan: IEEE, 2019: 972-973.

HSU J Y, TSENG W K, HSIEH J Y, et al. The rec-ommending agricultural product sales promotion mode in ecommerce using reinforcement learning with contextual multiarmed bandit algorithms[J]. Mathematical Problems in Engineering, 2020, 2020: 1-10.

JIANG R, MA C.Batched nonparametric contextual bandits[J].IEEE Transactions on Information Theory, 2025, 71(6): 4537-4555.

CHOUKALI M A, MIRBOLOUK S, VALIZADEH M,et al. Deep contextual bandits-based energy-efficient beamforming for integrated sensing and communication [J].Physical Communication, 2025, 68: 102576.

BOUDLAL A, KHAFAJI A, ELABBADI J, et al. Entropy adjustment by interpolation for exploration in proximal policy optimization (PPO)[J]. Engineering Applications of Artificial Intelligence, 2024, 133: 108401.

LI Y, TAN X Y. Candidate ratio guided proximal policy optimization[J]. Engineering Applications of Artificial Intelligence, 2025, 152: 110576.

浏览量

下载量

CNKI被引量

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于蚁群优化算法引导深度Q网络的移动机器人路径规划算法

基于DDQN-D³PG的无人机空战分层决策

一种元学习和强化学习结合的多飞行器协同制导律

基于人工势场和改进强化学习的自主式水下潜航器避障和航迹规划

一种探索率自适应设置的强化学习雷达干扰决策方法