Stochastic optimization methods for transportation and network design problems

在一个越来越多的公司开展国际活动的经济体中，物流扮演着至关重要的角色。在物流领域，运输和网络设计问题是一些更重要的问题的核心，必须解决这些问题才能确保竞争力。由于这些问题要考虑到总是存在不确定性的计划环境，因此作为该提案的一部分的研究计划致力于开发与该领域相关的随机问题的优化模型和求解算法。这里将涵盖三个主要的研究领域。研究领域一将考虑随机车辆路径问题，因为它们是在考虑复杂的运营问题时定义的(例如，“丰富的”规划环境)。研究领域二将涵盖当不确定性影响所要考虑的参数时的供应链设计。在前两个研究领域，将研究实际问题和应用问题；因此，将承担三大任务。第一个将致力于对所考虑的问题进行建模。将特别强调阐述所观察到的不确定性，并说明面临这些问题的公司可采取的实际补救行动。第二项任务将包括为所制定的问题开发解决算法。最后，第三项任务将包括验证所获得的解决方案。在这项任务中，将使用基于模拟的模型来衡量所提出的随机方法的有效性。本提案中定义的第三个也是最后一个研究领域将用于发展经典随机问题的一般求解方法。在这方面，将考虑使用混合战略和并行战略。

In an economy where more and more companies have international activities, logistics plays a crucial role. Within the logistics field, transportation and network design issues are at the heart of some of the more important problems that must be solved to ensure competitiveness. Since these issues are to be addressed considering a planning environment where uncertainty is always present, the research program that is part of this proposal is dedicated to the development of optimization models and solution algorithms for stochastic problems related to this field. Three main research areas will be covered here. Research area one will consider stochastic vehicle routing problems as they are defined when complex operational issues are to be taken into account (e.g., "rich" planning environment). Research area two will cover the design of the supply chain when uncertainty influences the parameters to be considered. Within these first two research areas, practical and applied problems will be studied; therefore, three major tasks will be undertaken. The first will be dedicated to the modelling of the problems considered. A particular emphasis will be placed on formulating the uncertainty observed and on expressing the real recourse actions available to companies that face these issues. The second task will include the development of solution algorithms for the problems formulated. Finally, the third task will cover the validation of the solutions obtained. Within this task, simulation based models will be used to measure the effectiveness of the stochastic approach that is proposed. The third and last research area defined in this proposal will be used for the development of general solution methodologies for classical stochastic problems. Within this area, the use of both hybrid and parallel strategies will be considered.