A multi-objective optimization model for student assignments in a school bus routing problem

The aim of this thesis is to investigate the student assignments in a school bus routing problem involving a bus stop selection and the satisfaction score of each student for all bus stops. The optimization model for the focused problem is created in two cases in term of a single-objective optimization and a multi-objective optimization. In case of a multi-objective optimization, the epsilon-constraint method and the weighted-sum method are used for solving this problem. In both cases, two types of our models are investigated. The first type of our model is related to the student assignment in a school bus routing problem without bus stop selection while the second type of our model is related to the same problem with bus stop selection. Two types of our models in both cases are tested with the simulated data. In case of a single-objective optimization, our experiment reports that the second model is suitable to solve the problem whenever the total distance of all buses is interested. In the meantime, both models can be selected if the total of satisfaction score is more important than the total distance of all buses. For the multi-objective optimization, the experiment shows that the total distance of the second model is shorter than the first model while the total of satisfaction scores of both models are slightly different.本论文旨在针对涉及公交站点选择的校车路径规划问题（school bus routing problem）中的学生分配方案，以及所有公交站点下每位学生的满意度评分展开研究。针对该目标问题的优化模型，分为单目标优化与多目标优化两类场景构建。在多目标优化场景中，采用ε-约束法（epsilon-constraint method）与加权和法（weighted-sum method）对该问题进行求解。两类场景下均对两种模型架构展开探究：其一为不含公交站点选择环节的校车路径规划问题学生分配模型，其二为包含公交站点选择环节的同问题模型。两类场景下的两种模型均采用模拟数据开展测试。在单目标优化场景下，实验结果表明，当以所有公交车辆总行驶距离为优化目标时，第二种模型更适配该问题的求解；而当以学生满意度总分作为核心优化目标时，两种模型均可选用。针对多目标优化场景，实验结果显示，第二种模型的总行驶距离短于第一种模型，而两种模型的学生满意度总分差异较小。