基于改进YOLOv8的飞机起降飞行姿态风险检测  

Risk Detection of Aircraft Takeoff and Landing Flight Attitude Based on Improved YOLOv8

作　　者：罗凤娥[1] 杜裕鑫 刘玉妍 朱子垚 吴小林 韩晓彤 LUO Feng-e;DU Yu-xin;LIU Yu-yan;ZHU Zi-yao;WU Xiao-lin;HAN Xiao-tong(Civil Aviation Flight University of China,Guanghan 618000,China)

机构地区：[1]中国民用航空飞行学院,四川广汉618000

出　　处：《航空计算技术》2024年第4期7-12,共6页Aeronautical Computing Technique

基　　金：民航局教育人才类“基于能力导向培养的航空运行控制虚拟仿真实验中心建设”项目资助(0252108)。

摘　　要：飞机在执行飞行任务时,其起飞和降落两个阶段是整个飞行过程中最关键和最具挑战性的时刻之一,其起降飞行姿态的变化程度关系到整个飞机的安全性。为了降低飞机在起降过程的风险性,提出了一种基于改进YOLOv8的飞机起降飞行姿态的风险检测方法。该检测方法综合利用了CBAM注意力机制、GSConv轻量化卷积模块和HIOU准则函数等深度学习和目标检测技术,实现了对飞机机翼、起落架等关键结构部位的准确检测,以便利用识别到的飞行姿态特征参数来对飞机当前的安全状态进行分析和评估。经实验验证表明,该改进方法对于飞机起降阶段中的飞行姿态风险状况能够进行精准识别和检测,具有较高的可靠性。The take-off and landing stages are among the most critical and challenging moments in the entire flight process,as the degree of attitude change directly impacts the overall aircraft safety.To mitigate risks during takeoff and landing,a risk detection method based on an improved YOLOv8 flight attitude is proposed.This detection method effectively combines deep learning and object detection technologies,including CBAM attention mechanism,GSConv lightweight convolutional module,and HIOU criterion function to achieve precise identification of key structural components such as aircraft wings and landing gear.By utilizing identified flight attitude characteristic parameters,it enables analysis and evaluation of the current safety state of the aircraft.Experimental results demonstrate that this enhanced method accurately identifies and detects flight attitude risks during takeoff and landing stages with high reliability.

关 键 词：改进YOLOv8 目标检测 特征提取 起降飞行姿态 风险检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]