基于多目标粒子群的土地整理项目选址模型

土地整理项目选址实质上是一个多目标空间优化问题,将传统选址方法用于解决多目标土地整理项目选址存在明显不足。该文提出了一种基于多目标粒子群的土地整理项目选址模型,归纳了土地整理项目选址的选址规则,提取3个能够体现土地整理项目选址目的及意义的目标函数,即新增耕地潜力最高、空间分区集中连片、土地适宜性评价指数最高,同时考虑了土地整理项目最小新增耕地率、实施规模2类约束条件,并详细阐述了算法的核心思想、矢量编码策略、状态更新机制等内容,最后选取湖北省嘉鱼县为试验区,验证了模型的可行性和有效性,结果表明通过对目标的权重调控可以得到不同目标偏好的土地整理项目选址方案,该文所构建模型在土地整理实践中具有可操作性,提供的选址方案科学合理,为土地利用规划和土地利用调控管理提供支持。

Site selection model of land consolidation projects based on multi-objective optimization PSO

Abstract: Land consolidation, as a significant way to optimize the allocation of land resources in relatively small regions, is one of the important contents of land use plan. The only way in China to implement the work of land consolidation is project, so the site selection of land consolidation projects, demarcating the border in the two-dimensional space, is the first step of land consolidation projects. Its scientificity and rationality will decide the achievement of maximum benefit of land consolidation. But from the current actual situation of land consolidation, the theory and the practice of the site selection of land consolidation projects are both weak, so on the occasion of new round of land use planning, the research of scientific theories and methods is necessary and urgent. Site location of land consolidation project was discussed in this paper based on multi-objective optimization. The site location rules of land consolidation project were generalized, and 3 objective functions were comprised, which were the maximum potential of newly-increased cultivated land, the higher connectivity among zones and the best land suitability. Two types of constraint conditions that included the minimum newly-increased cultivated land ratio and the area limitation of land consolidation project were also considered. Site selection model of land consolidation projects based on the multi-objective particle swarm optimization algorithm was proposed to solve the multi-objective spatial optimization problem with the assist of GIS (geographic information system). The mapping relationships between different concepts in the site selection of land consolidation projects and intelligence algorithms were analyzed. Each vector parcel was considered to be a decision-making unit, and the value of decision variable was suggested to be 1 when the corresponding parcel was chosen to the project areas, otherwise it was zero. Each particle in the particle swarm optimization (PSO) algorithm represented a site selection scheme of land consolidation project, which comprised all the multi-dimensional decision-making units. The particle was encoded based on the identification of the parcels and the values of decision variable. The structures of velocity calculation operator and position updating operator were designed based on the spatial coding scheme of individual. The mechanism of particle status update and the procedure of evolution should be improved because of the discrete solution space of the model of the site selection of land consolidation projects. At last, take Jiayu county, Hubei province as a case study, different weight schemes were chosen. Jiayu county is an important base for producing the agricultural by-products and aquatic products, which made it our choice of study area to test the model. The model was expected to reasonably select spatial units in accordance with multiple objectives and constraints, and to optimize the newly-increased cultivated land ratio and the spatial pattern. Different weight schemes were chosen to generate several different solutions, and the results showed that 3 objective functions in the model were conflicting with each other, and the increase of any objective function value came at the cost of the decrease of others. In the running process of the model of the site selection of land consolidation projects, the particle tended to select the land use parcels of garden, grassland and ditch into the project areas with the guidance of the objective of newly-increased cultivated land potential, because these land could be converted into cultivated land after land consolidation, and the index of land consolidation suitability decreased to some degree. The neighborhood identity index was improved effectively and the degree of fragmentation reduced across spatial units, because the intelligent particle chose the parcels adjacent to the selected ones or some isolated parcels into the project areas whenever possible, which was leaded by the objective of compact spatial pattern. The decision-makers can get alternative solutions of site selection for land consolidation project which satisfy their preference by setting corresponding weight coefficients of the 3 objective functions.