基于分布鲁棒优化的危化品运输事故应急救援站选址-分配问题研究  

Location Allocation Problem of Emergency Response Stations for Hazardous Materials Transportation Accidents Based on Distributionally Robust Optimization

作　　者：王伟[1,2] 葛颂[1] 张文思 侯小雨 WANG Wei;GE Song;ZHANG Wen-si;HOU Xiao-yu(School of Economics,Ocean University of China,Qingdao 266100,Shandong,China;Hazardous Materials Logistics Research Center of Northeast Asia,Ocean University of China,Qingdao 266100,Shandong,China)

机构地区：[1]中国海洋大学经济学院,山东青岛266100 [2]中国海洋大学东北亚危险品物流研究中心,山东青岛266100

出　　处：《中国公路学报》2025年第3期210-223,共14页China Journal of Highway and Transport

基　　金：国家自然科学基金项目(71701189,72202223);山东省自然科学基金项目(ZR2021QG019);北京市社会科学基金规划项目(22GLC060)。

摘　　要：针对危化品运输事故应急救援站的选址与分配问题展开研究:首先,为准确描述救援站对道路的全覆盖且不出现重复,提出了非增连续形式的道路全覆盖度函数,建立了确定性情况下的改进广义最大弧覆盖选址-分配(Improved Generalized Maximal Arc-covering Location-allocation,IGMACLA)模型;其次,考虑到应急救援时间具有不确定性,以及新提出的道路全覆盖度函数在处理不确定救援时间方面的适用性,构建了基于分布鲁棒优化方法的带有模糊机会约束的IGMACLA模型;然后,运用易处理的近似方法,分别在零均值有界扰动和高斯扰动模糊集下将原始分布鲁棒优化模型转换为整数二阶锥规划模型,并进一步使用分支切割算法求解;最后,借助数值算例验证了上述模型的有效性和可靠性,并分析了分布鲁棒优化方法相较于传统鲁棒优化方法和随机规划方法的优势所在。研究结果表明:基于分布鲁棒优化的IGMACLA模型的计算结果较之于确定性IGMACLA模型而言相对保守,但是具有较强的鲁棒性;随着容许度水平的增大,基于分布鲁棒优化的IGMACLA模型的最优目标值,即应急救援站总覆盖效果的下界值逐渐增大(或减小);通过结合部分的概率分布信息,分布式鲁棒优化方法显著优于传统鲁棒优化方法;与随机规划方法相比,分布鲁棒优化方法可以通过付出较小的代价,来抵抗未知完整概率分布信息所带来的不确定性;模糊集中部分概率分布信息被利用得越多,分布鲁棒优化方法受容许度水平和分布非精确性变化的影响越小。研究成果可作为危运事故应急救援站选址与任务分配的决策依据。This study investigates the location allocation problem of emergency response stations for hazardous material transportation accidents.First,to accurately describe the full coverage of rescue stations on roads without redundant coverage,a nonincreasing continuous form of the road full coverage degree function was proposed,and an improved generalized maximal arc-covering location allocation(IGMACLA)model was established under deterministic conditions.Second,considering the uncertainty of the emergency response time and the feasibility of the newly proposed full-coverage road degree function in addressing uncertain response times,an IGMACLA model was constructed based on a distributionally robust optimization(DRO)approach with fuzzy chance constraints.Third,an approximation method was employed to formulate the original DRO model as an integer second-order cone programming model under both zero-mean bounded perturbation and Gaussian perturbation ambiguous sets,which were further solved using the branch-and-cut algorithm.Finally,the effectiveness and reliability of the proposed models were verified using numerical examples,and the advantages of the distributional robust optimization method over traditional robust optimization and stochastic programming methods were analyzed.The computational results of the IGMACLA model based on DRO are relatively conservative compared with those of the deterministic IGMACLA model but present strong robustness.As the tolerance level increases,the optimal objective value of the IGMACLA model based on DRO,which is called the lower bound value of the total coverage effect of emergency response stations,increases(or decreases).By incorporating partial probability distribution information,the DRO approach significantly outperforms the traditional RO approach.Compared with the stochastic programming method,the DRO method pays a small price to resist the uncertainty associated with unknown complete probability distribution information.As more information about the partial probability distri

关 键 词：交通工程 选址-分配 分布鲁棒优化 危化品运输事故 应急救援 道路全覆盖 

分 类 号：U491[交通运输工程—交通运输规划与管理]