基于LSTM增强的无人机低空物流远程识别差分隐私框架
冀星宇 钟乐轩 冯玉琴 肖邦媛 莫德涛 韦慧欣

桂林电子科技大学人工智能学院，广西桂林，541004；

摘要：本文提出一种基于长短期神经网络（LSTM）的动态隐私预算ε控制方法，可根据无人机电量、位置误差及充电状态自适应调整隐私强度，并引入时间感知拉普拉斯机制以保证轨迹平滑性。仿真结果表明，本方法在保证100%充电成功率的前提下，平均位置误差仅为3.89 米，显著优于静态ε策略。

关键词：无人机远程识别；差分隐私；长短期神经网络；低空经济；物流；轨迹扰动；电量感知隐私

参考文献

[1] Liu, Y.; Liu, J.; Shi, X.; Tang, C. Multi-Station UAV–UGV Cooperative Delivery Scheduling Problem with Temporally Discontinuous Service Availability Under Diverse Urban Scenarios. Drones 2026, 10, 269.

[2] Bu, D.; Hernandez, M.; Haruna, F.; Abasi, P.M.; Kremer, P. Improving health access through the distribution of COVID-19 vaccines using drones in Ghana. SSRN Electron. J. 2023.

[3] Aniculaesei, A. Regulatory Frameworks and Development Standards for Civilian Unmanned Aircraft Systems: From Regulatory Safety Intent to Development Lifecycles. Drones 2026, 10, 271.

[4] European Union. Commission Implementing Regulation (EU) 2019/947 of 24 May 2019 on the rules and procedures for the operation of unmanned aircraft. Off. J. Eur. Union 2019, 152, 45–71.

[5] RTCA-DO-344:2013; Operational and Functional Requirements and Safety Objectives for Unmanned Aircraft Systems Standards. RTCA: Washington, DC, USA, 2013.