航空学报 > 2026, Vol. 47 Issue (12): 332898-332898   doi: 10.7527/S1000-6893.2025.32898

基于深度集成学习的机载视觉感知鲁棒性设计与分析

马赞1,2, 张同杰1, 白杰2(), 陈勇3, 田毅1,2   

  1. 1.中国民航大学 安全科学与工程学院,天津 300300
    2.中国民航大学 民用航空器适航审定技术重点实验室,天津 300300
    3.中国商用飞机有限责任公司 上海飞机设计研究院,上海 200216
  • 收稿日期:2025-10-13 修回日期:2025-11-10 接受日期:2025-12-02 出版日期:2025-12-29 发布日期:2025-12-25
  • 通讯作者: 白杰 E-mail:jbai@cauc.edu.cn
  • 基金资助:
    国家重点研发计划(2022YFB3904300)

Robustness design and analysis of airborne visual perception based on deep ensemble learning

Zan MA1,2, Tongjie ZHANG1, Jie BAI2(), Yong CHEN3, Yi TIAN1,2   

  1. 1.College of Safety Science and Engineering,Civil Aviation University of China,Tianjin 300300,China
    2.Key Laboratory of Civil Aircraft Airworthiness Certification Technology,Civil Aviation University of China,Tianjin 300300,China
    3.Shanghai Aircraft Design and Research Institute,Commercial Aircraft Corporation of China,Shanghai 200216,China
  • Received:2025-10-13 Revised:2025-11-10 Accepted:2025-12-02 Online:2025-12-29 Published:2025-12-25
  • Contact: Jie BAI E-mail:jbai@cauc.edu.cn
  • Supported by:
    National Key Research and Development Program of China(2022YFB3904300)

摘要:

基于机器学习的视觉感知功能对航空器在复杂环境下提升态势感知或自主飞行能力至关重要,其性能对飞行安全具有重大影响。而机器学习技术固有的概率特性对其满足适航安全目标造成巨大挑战,阻碍其在机载端的应用。因此,形成一种基于深度集成学习的机载视觉感知鲁棒性优化设计方法。首先,基于运行设计域生成高代表性数据集,提出基于CW-SSIM的K折交叉验证方法,在少量数据情况下提高训练集与验证集间独立性。其次,基于YOLO架构,引入深度可分离卷积,设计3种优化基学习器,分别通过多尺度特征融合、小目标检测的专注性提升和细粒度特征提取应对不同检测需求。最后,设计集成学习方法,采用加权自适应融合策略动态调整基学习器权重,提升模型的精度及鲁棒性。实验结果表明:集成学习模型比NMS、WBF等检测框融合算法具有更好的性能表现,并且在IoU≥0.7时,集成模型比单个模型平均的P值、R值和F1分数至少提高11.36%、2.06%和6.78%,在IoU≥0.75时,AP值至少提高约3%,所提方法在复杂环境中显著提高目标检测精度和鲁棒性,有效减少误判和漏判,为航空器安全飞行提供技术保障。

关键词: 机载视觉感知, 鲁棒性设计, 深度集成学习, K折交叉验证方法, 深度可分离卷积

Abstract:

The visual perception function based on machine learning is crucial for enhancing situational awareness or autonomous flight capabilities of aircraft in complex environments, and its performance has significant impact on flight safety. However, the inherent probabilistic nature of machine learning techniques poses substantial challenges to meeting airworthiness safety objectives, thereby hindering their application in airborne systems. To address this issue, a robustness-oriented design method for airborne visual perception based on deep ensemble learning is established. First, a highly representative dataset is generated based on the operational design domain, and a K-fold cross-validation method based on CW-SSIM is proposed to improve the independence between the training and validation sets with limited data. Second, based on the YOLO architecture, depthwise separable convolution is introduced, and three optimized base learners are designed to address different detection needs through multi-scale feature fusion, enhanced focus on small object detection, and fine-grained feature extraction. Finally, an ensemble learning method is designed using a weighted adaptive fusion strategy to dynamically adjust the weights of base learners, thereby improving the model accuracy and robustness. Experimental results show that the ensemble learning model outperforms detection box fusion algorithms such as NMS and WBF. When the IoU is not less than 0.7, the ensemble model improves the average P-value, R-value, and F1 score by at least 11.36%, 2.06%, and 6.78%, respectively, compared to a single model. When the IoU is no less than 0.75, the AP value increases by at least approximately 3%. These results indicate that proposed method significantly enhances target detection accuracy and robustness in complex environments, effectively reducing false positives and missed detections, and provides technical assurance for the safe flight of aircraft.

Key words: airborne visual perception, robustness design, deep ensemble learning, K-fold cross-validation method, depthwise separable convolutions

中图分类号: