面向农村小学广域布局优化的进化算法研究

我国农村小学学校规划以区县统筹、布局政策包含多重内容和学校布局涉及多方关联利益，使布局优化问题具有广域、多目标、多约束的特征。进化算法是求解多目标多约束优化问题的主要技术。本项目围绕实际场景下农村小学学校布局优化问题的建模技术及其规划优化方法开展研究，内容包括：（1）农村小学广域布局优化问题的多目标多约束特征刻画和量化建模；（2）研究目标函数和场景地理空间属性的可分解性，建立子问题的关联行为模式，提出高效的基于协同优化的问题分解分析方法；（3）构建适当的问题域特征和启发式策略的知识集，发展基于选择的启发式算法框架，设计多目标多约束组合优化问题的自选择式的高效启发式优化算法。研究成果将获得契合实际场景的农村小学广域布局问题建模技术和优化技术，为省、市、县级教育部门提供包含“撤销、合并、新建、改造、交通补贴”等内容的学校布局优化方案。

The distribution adjustment of the rural primary schools in china has some characters:taking a county as a whole, many contents contained in the layout policy and involving multiple related interests. The optimization of the Wide-Area Layout of Rural Primary Schools (WALRPS) is a multi-restrictions multi-objectives optimization problem and evolutionary computing constitutes major tools to solve this kind problem. This project focuses on Modeling techniques and planning optimization methods of WALRPS in real scenarios. Firstly, we will investigate the characteristics of WALRPS and establish the multi-restrictions multi-objectives problem model. Secondly, we will analyze the decomposable features of the attributes of objective functions and geographic elements, explore the association behavior between sub-problems and propose co-evolutionary based decomposition methods.Finally, based on specific features of actual problems, we will construct the knowledge set of appropriate problem domain features and heuristic strategies, develop a selection based heuristic algorithm framework and design efficient and self-adaptive heuristic algorithms for multi-restrictions multi-objectives problem. The research results will obtain the modeling technology and optimization technology of WALRPS in real scenarios and provide WALRPS optimization schemes with contents including "cancellation, merger, new construction, transformation, transportation subsidies" for the provincial, municipal and county-level education departments.