项目名称: 多目标复杂车辆路径问题中的模因优化方法研究
项目编号: No.61301298
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 骆剑平
作者单位: 深圳大学
项目金额: 23万元
中文摘要: 多目标复杂车辆路径问题(MOCVRP)是现代智能物流业需要研究的一项重要内容,传统智能算法求解该问题面临多相变点难解、难于收敛到真实Pareto前沿等问题。 模因进化是一种通过模因传播,模拟人或动物思想传递,处理动态复杂问题的新型智能计算方式。项目探索面向MOCVRP 的模因优化新方法:以混合蛙跳算法为原型,改进并拓展模因扩散机制,利用基于自组织临界性理论的极值动力学优化设计模因挖掘机制,研究并建立完整且可动态扩展的模因进化计算模型;从数学角度对模型的收敛性及参数设置进行理论分析和改进;进而提出基于Monte Carlo 采样及ASF的超体积近似估计多目标模因计算框架,利用模因扩散、模因挖掘进化机制以及超体积指标快速估算方式实现对MOCVRP的快速有效求解。 项目探索求解MOCVRP 这一复杂问题的新思路,首次将模因进化框架应用于多目标优化领域,为多目标复杂组合优化问题提供新的优化方法。
中文关键词: 智能计算;生物计算;多目标优化;路径优化;
英文摘要: Multi-Objective Complicated Vehicle Routing Problem (MOCVRP) is an important research content of the modern intelligent logistics. The traditional intelligent algorithms are facing many problems to solve it. Memetic computing(MC)is a novel intelligent computational method, which simulates ideas transmission ofhuman and/or animal through memes propagation to handle various complex and dynamic problems. This project aims at solving MOCVRP by memetic computing. A complete and expandable MC model, which derived from shuffle frog leaping algorithm (SFLA) through improving and expanding its memes diffusion mechanism and adding the memes mining mechanism achieved by extremal optimization, is proposed in this project. The convergence of the proposed MC model is analyzed theoretically. Furthermore, the multi-objective MC model is researched to solve the MOCVRP using the hyper volume measure based on approximation pattern as well. This project is to explore a new method, new framework, and new algorithm to solve MOCVRP, and to lay a solid foundation in the further research and development of MC.
英文关键词: intelligent computing;biology computing;multi-objective optimization;routing scheduling;