📄 中文摘要
该研究针对高速公路智能交通系统中的多目标无人机（UAV）布置问题，旨在平衡覆盖范围、链路质量和无人机数量，同时满足几何约束。通过高D高速公路录音构建可重复的基准数据集，并利用非支配排序遗传算法（NSGA-II）生成帕累托最优标签。采用偏好规则生成可部署目标，同时保持多目标评估。研究中训练了快速代理模型，将无序的车辆位置映射到无人机数量和连续布置，使用了感知排列的损失函数和约束正则化训练，涵盖基于集合和序列的架构。评估协议结合了帕累托质量指标和成功率曲线，以全面评估模型性能。
高速公路智能交通系统中多目标无人机布置的快速代理学习
📄 English Summary
Fast Surrogate Learning for Multi-Objective UAV Placement in Motorway Intelligent Transportation System
This research addresses the multi-objective placement of unmanned aerial vehicles (UAVs) within motorway intelligent transportation systems, focusing on balancing coverage, link quality, and UAV count under geometric constraints. A reproducible benchmark is constructed from highD motorway recordings, with recording-level splits, and Pareto-optimal labels are generated using the Non-dominated Sorting Genetic Algorithm II (NSGA-II). A preference rule is applied to yield deployable targets while maintaining multi-objective evaluation. Fast surrogate models are trained to map unordered vehicle positions to UAV count and continuous placements, employing permutation-aware losses and constraint-regularized training across set-based and sequence-based architectures. The evaluation protocol integrates Pareto quality metrics and success-rate curves to comprehensively assess model performance.