两类车辆路径问题的遗传算法-软件工程专业论文.docxVIP

万方数据 万方数据 两类车辆路径问题的遗传算法 摘 要 物流和科研当中经常涉及到车辆路径问题，它是一个典型的组合优化问题. 对该问题的研究不仅有理论意义，还具有重要的实际应用价值和经济效益. 此类 问题的研究种类非常多，对于不同种类车辆路径问题，采用的数学模型和算法也 大不相同. 考虑到问题的复杂性，智能算法已成为求解车辆路径问题的主要方法 之一，如遗传算法、粒子群算法、模拟退火算法等. 本文采用带局部搜索技术的 遗传算法求解两类车辆路径问题，具体工作如下： 1. 对于带容量限制的车辆路径问题，我们设计了一个基于局部搜索技术的 遗传算法. 该局部搜索过程通过概率选择要改进的客户组合，通过优化该组合 而改进当前路径. 实验结果表明，该局部搜索方法能有效提高算法的搜索效率. 另外，为了更好地产生新的杂交后代，本文给出了一个基于两个父代个体求和 的杂交算子. 该算子的特点是即使两个父代个体一样，也能产生不同于父代的 一个后代个体. 这使得算法在运行后期能有效产生新的后代，保持种群的多样 性. 2. 有时间窗的车辆路径问题，现有的大多数方法采用“硬”的时间窗约束， 即必须严格满足时间窗约束. 本文软化了时间窗约束，利用罚函数方法给出了 一个新的双目标优化模型. 该方法将时间约束作为罚函数项放在目标函数中， 通过改变罚因子的值来控制时间约束违反度. 为了有效求解该模型，设计了一 个结合局部搜索技术的双目标进化算法. 数值实验表明该算法是可行有效的. 关键词：车辆路径问题，遗传算法，时间窗、多目标优化 I Genetic Algorithms for Two Classes of Vehicle Routing Problems Abstract Both in the logistics field and scientific research, the vehicle routing problem is often taken into account. This problem is a typical combinatorial optimization problem, and not only is it very important in theoretical study, but has also important practical value and economic benefit. There exist various vehicle routing problems, and different mathematical models and algorithmic approaches are adopted for dealing with different type of problems. Considering the complexity of the problem, as main solution approaches, some intelligent algorithms, such as genetic algorithms, particle swarm optimization algorithms, and simulated annealing algorithms, etc, have been applied to solve the corresponding models. In this thesis, two genetic algorithms with local search techniques are developed for dealing two types of the vehicle routing problems, respectively. The detailed work is as follows: For the vehicle routing problem with capacity limitation, a genetic algorithm based on local searching technology is designed. In the proposed local search procedure, the sub-sequence of the customers is first chosen by using a probability selection scheme. Then the sub-sequence is optimized which makes current path improved. The experimental results show that the loc