交通运输工程与信息学报

【背景】传统路网监测方法易受时空因素影响，而无人机的快速发展为交通路网监测提供了新的视角。【目标】在无人机路网巡检过程中，环境状态不确定会影响无人机执行任务效率且监测需求不确定对系统效率构成主要挑战。为实现经济高效的路网监测，需在无人机配置与调度之间达成多目标协同优化，实现经济高效的总目标。【方法】本文基于分布式鲁棒优化（DRO）框架，通过状态模糊集刻画监测需求的不确定性，反映需求随环境状态变化的分布特征，构建两阶段状态描述模型。运用嵌套Benders分解算法拆分模型，进行对偶简化，求得在最坏情况下第二阶段的主问题与子问题最优解。【数据】采用上海市杨浦区20个主要路网节点及2001年至2020年的对应节点降水量数据，构建场景模型与状态模糊集，并对模拟算例与真实算例（杨浦区重要路网）进行数据模拟与分析。【结果】数值结果表明，所提出的DRO模型相较于传统SAA模型、OR模型能有效降低系统惩罚成本，在不确定性条件下保持解的稳健性，实际应用中能够应对各种不确定性带来的风险。

[Background] Traditional road network monitoring methods are often susceptible to spatiotemporal influences， while the rapid development of unmanned aerial vehicles (UAVs) offers a new perspective for traffic network surveillance. [Objective]During drone road network inspections， uncertain environmental conditions impair mission efficiency， while unpredictable monitoring demands pose the primary challenge to system effectiveness. To achieve cost-effective road network monitoring， multi-objective coordination optimization between drone configuration and scheduling is required to realize the overarching goal of economic efficiency. [Method] This paper employs a distributionally robust optimization (DRO) framework. It characterizes monitoring demand uncertainty using ambiguity sets to reflect demand distribution patterns across varying environmental states， constructing a two-stage state description model. The nested Benders decomposition algorithm is applied to decompose the model， enabling dual simplification to derive optimal solutions for both the main problem and subproblems in the second stage under worst-case scenarios. [Data] Precipitation data from 2001 to 2020 for 20 major road network nodes in Shanghai’s Yangpu District are used to construct scenario models and state ambiguity sets. Both simulated and real-world cases (key road networks in Yangpu District) are employed for data simulation and analysis. [Result] Numerical results demonstrate that the proposed DRO model effectively reduces system penalty costs compared to traditional SAA and OR models， maintains solution robustness under uncertainty， and can effectively address risks arising from various uncertainties in practical applications.