一种“效率-公平-运力”多维权衡的需求可拆分配送方法

针对自然灾害及重大社会公共事件等各类突发事件的配送问题，本文以公路运输为研究场景，将配送时间最短、加权时间攀比值最小和使用车辆数最少为多维目标，在引入需求可拆分这一限制条件的基础上，构建“效率-公平-运力”多维权衡的需求可拆分应急物资配送模型。针对该问题设计改进的蚁群算法求解模型。从选择拆分点、信息素更新和引入变邻域搜索算子这3个方面改进了算法，并实现当解持续不变时，初始化信息素，以增加随机性。结果表明，与传统求解算法相比，改进算法的稳定性更高(平均偏差率降低7.00%)，寻优性更好(优化率提高7.41%)。 通过分析考虑三目标、双目标和决策者具有明显偏好的多重场景下的求解结果得知：效率、公平、 运力这3个子目标相互悖反，增加运力投入可以显著提高配送方案的效率与公平；当运力不变时，效率与公平之间近似呈同比例反比关系。研究结论可为救灾目标不确定条件下多因素考量的应急物资配送决策生成与优化问题提供方法改进与可量化决策支撑。

A Multi-dimensional "Efficiency-Fairness-Capacity" Balancing Method for Demand Splitting Distribution

This paper focuses on the emergency material distribution in special cases such as in natural disasters and major social and public events. A multi-dimensional method of "efficiency-fairness-capacity" is proposed for the distribution of emergency materials using highway transportation based on the restriction that demand can be split, and the objectives of the shortest distribution time, the smallest weighted time climbing value and the least number of distribution vehicles. An improved Ant Colony algorithm is designed to solve the model. The algorithm is improved in three aspects: choosing splitting points, pheromone updating and introducing variable neighborhood search operators, and implementing the initialization of pheromones when the solution is continuously constant to increase the randomness. The results show that the improved algorithm has higher stability (7.00% lower average deviation rate) and better optimality finding (7.41% higher optimization rate) compared to the traditional algorithm. From the results under multiple scenarios considering tri-objective, bi-objective, and decision makers with obvious preferences, it was found that the three sub-objectives of efficiency, equity, and capacity are paradoxical to each other. Increasing the distribution capacity can significantly improve the efficiency and equity. When the capacity is constant, the efficiency and equity are negatively correlated in the same proportion. The study results provide methodological references and support for the emergency material distribution with multi-factors and uncertain disaster relief objectives.