【摘要】 无人机技术的广泛应用对其安全性与可靠性提出了更高要求，基于指纹特征的识别技术成为管控非法无人机的关键手段。为系统综述基于深度学习的无人机指纹识别方法研究现状，重点关注现存神经网络架构、数据集问题及其优化，以及复杂环境中的适应性提升。介绍了深度学习架构如卷积神经网络(Convolutional Neural Network, CNN)、循环神经网络(Recurrent Neural Network, RNN)等在无人机信号特征提取和分类中的应用现状、优势及其改进；研究分析了训练数据集现存问题，总结了样本总量不足和类别间样本比例失衡两方面挑战，针对这些问题的数据扩充技术进行了分析；研究总结了深度学习技术在噪声干扰、硬件缺陷、特定地形等复杂环境下提升无人机指纹识别适应性的优势与做法。清晰指出了现有架构的优势与改进方向、数据集的核心瓶颈及解决路径、复杂环境下的增强策略，为深度学习在无人机指纹识别领域的应用与发展提供了重要参考。更多还原

【Abstract】 The widespread application of UAV technology has imposed higher requirements on its safety and reliability. Recognition technology based on fingerprint features has become a key means for the management and control of illegal UAVs. To systematically review the research status of UAV fingerprint recognition methods based on deep learning, existing neural network architectures, problems in datasets and their optimizations, as well as the improvement of adaptability in complex environments have been focused on. Firstly, the current application status, advantages and improvements of deep learning architectures such as Convolutional Neural Network(CNN) and Recurrent Neural Network(RNN) in feature extraction and classification of UAV signals have been introduced. Secondly, the existing problems of training datasets have been studied and analyzed, the challenges of insufficient total samples and imbalanced sample proportions among categories have been summarized, and the data augmentation techniques for addressing these problems have been analyzed. Finally, the advantages and approaches of deep learning technologies in enhancing the adaptability of UAV fingerprint recognition under complex environments such as noise interference, hardware defects, and specific terrains have been summarized. The advantages and improvement directions of existing architectures, the core bottlenecks of datasets and their solution paths, as well as the enhancement strategies under complex environments have been clearly pointed out. This study can provide important references for the application and development of deep learning in UAV fingerprint recognition.更多还原