针对危化品运输事故辅助救援无人机起降点选址与分配问题展开研究，首先，建立了确定性情况下的无人机起降点选址-分配模型；其次，考虑路段潜在风险的不确定性，构建了基于分布鲁棒优化方法的无人机起降点选址-分配模型；然后，运用易处理的近似方法，在零均值有界扰动模糊集下将原始分布鲁棒优化模型转化为整数规划模型，并使用基于Benders分解的分支切割算法求解；最后，利用数值算例验证了上述模型和算法的有效性。研究发现：分布鲁棒优化模型的计算结果虽然相对保守，但具有较强的鲁棒性，可较好地提升应急救援系统的可靠性；无人机起降点的覆盖效果随着拟建起降点数量的增加而增大，但其边际效益是递减的；在缺乏不确定参数的概率分布信息时，分布鲁棒优化方法优于随机规划方法；与传统鲁棒优化方法相比，分布鲁棒优化方法通过融合部分概率分布信息规避了过于保守的选址-分配方案。   

This study investigates the location and allocation problem of auxiliary rescue drone landing points for hazardous material transportation accidents. First, a drone landing points location and allocation model was established under deterministic conditions. Second, considering the uncertainty of potential road risk, a drone landing points location and allocation model was constructed based on distributionally robust optimization method. Third, an approximation method was employed to formulate the original distributionally robust optimization model as an integer programming model under the zero-mean bounded perturbation ambiguous set, which was further solved using the branch and cut algorithm based on Benders decomposition. Finally, the effectiveness of the proposed model and algorithm were verified using numerical examples. It is found that the results of the distributional robust optimization model are relatively conservative but present strong robustness, which can significantly enhance the reliability of emergency rescue systems. The coverage effect of drone landing points increases with the number of planned landing points, but its marginal benefit is diminishing. When there is a lack of information about the probability distribution of uncertain parameters, the distributionally robust optimization method outperforms the stochastic programming method. Compared to the traditional robust optimization method, the distributionally robust optimization method can avoid overly conservative location-allocation solutions by incorporating partial probability distribution information.