基于改进卷积神经网络的机场跑道封锁效能评估  

Evaluation of Airport Runway Blockade Effectiveness Based on Improved Convolutional Neural Network

作　　者：操龙平 陈谋[1] 周同乐 Cao Longping;Chen Mou;Zhou Tongle(College of Automation Engineering,Nanjing University of Aeronautics&Astronautics,Nanjing 211106,China)

机构地区：[1]南京航空航天大学自动化学院,南京211106

出　　处：《航空兵器》2025年第2期128-136,共9页Aero Weaponry

基　　金：国家自然科学基金青年科学基金项目(62203217);江苏省基础研究计划自然科学基金青年基金项目(BK20220885);天元实验室基金项目(24-JSKY-ZZKT-29)。

摘　　要：针对传统机场跑道封锁效能评估方法因需要遍历所有跑道造成效率慢且难以采用图像数据的问题,本文提出一种基于改进卷积神经网络算法的机场跑道封锁效能评估计算模型。建立了机场跑道毁伤数据与是否封锁成功之间的非线性关系模型,避免直接采用循环求解所带来时间过长的问题。根据跑道毁伤图像特点,对卷积核的大小及数量进行相应的改进,同时引入批量归一化处理层和Mish激活函数,解决训练过程中梯度消失的问题。仿真结果表明,该算法不仅能够有效判别跑道是否成功封锁以及计算某一组瞄准点下的封锁概率,而且识别速度相比于传统算法更具有优势。In response to the inefficiency and inability to utilize image data inherent in traditional airport runway blockade effectiveness assessment methods,this paper presents an improved convolutional neural network algorithm for assessing the blockade effectiveness of airport runways.A nonlinear model is established between the damage data of the airport runway and the success of blockade,avoiding the excessive time consumption caused by direct iteration.The size and number of convolutional kernels are modified according to the characteristics of the runway damage images,and batch normalization layers and Mish activation functions are introduced to address the issue of gradient disappea-rance during training.Simulation results demonstrate that the algorithm can effectively determine whether the runway is successfully blocked and calculate the blockade probability for a set of aiming points,and it has a significant advantage in recognition speed compared to traditional algorithms.

关 键 词：机场跑道 封锁效能评估 封锁概率 跑道毁伤 卷积神经网络 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]