分类： 地球科学 >> 地理学 提交时间： 2021-05-19
摘要： 本文提出了一个数学启发算法求解单源设施区位问题(SSCFLP)。该算法从一个初始解开始，迭代地对当前解进行超大邻域搜索改进，直到若干次尝试不能该改进当前解为止。算法中，初始解使用拉格朗日松弛启发算法，超大邻域搜索采用子问题数学模型精确求解。算法设计的要点在于超大邻域的选择，既不能太大造成子问题求解困难，也不能太小造成当前解难以改进。使用5组272个基准测试案例进行算法测试，共发现191个案例的最优解，更新36个案例的已知最好解，表明数学启发算法性能优异。与近年的代表性算法(如割平面、核搜索、超图多交换启发、廊道方法)进行比较，在大规模案例上，本文算法无论求解质量还是计算时间均具有显著的优势。
分类： 地球科学 >> 地理学 提交时间： 2021-05-19
摘要： This article presents a hybrid algorithm for the service area problem. The design of service areas is one of the essential issues in providing efficient services in both the public and private sectors. For a geographical region with a number of small spatial units, the service area problem is to assign the service-demand units to the service-supply units such that each facility has a service area. The basic criteria for the service areas are the highest service accessibility, the contiguous service areas, and that the service demand does not exceed the service supply in each service area. A hybrid algorithm for the service area problem is proposed by extending iterative local search (ILS) algorithm with three schemes: population-based ILS, variable neighborhood descent (VND) search, and set partitioning. The performance of the algorithm was tested using 60 well-designed instances. Experimentation showed that the instances could be solved effectively and efficiently. The solutions found by the hybrid algorithm approximate optimal solutions or the lower bounds with an average gap of 0.15%.
点击量
待评议
点击量
下载量
评论
