基于并行多智能体人工免疫算法的土地利用空间优化模型研究

Land use spatial allocation plays a very important role in promoting land use efficiency. It is usually defined as a constrained multi-objective spatial optimization problem. The intelligent optimization algorithms, which including Genetic Algorithms, Artificial Immune Systems and etc. , have been proved to be effective tools for solving these optimization problems in the previous studies. However, these algorithms are originally designed for solving non-spatial problems, such as numerical or combinatorial optimization problems. Therefore, these algorithms still have some limitations in solving spatial optimization problems: a) the encoding models of them are difficult to meet the demand for representing the complicated spatial characteristics of the optimization problems, b) the stochastic search process cannot explain the principles and mechanisms of practical land-use optimization, c) Most of these models were developed under serial computing environment, which were incapable of handling large scale problems. In order to overcome the above limitations of the existing models, this project is aims to advance the application of intelligent algorithms in the optimization of land-use. To achieve this goal, a parallel land-use spatial allocation optimization model, which based on multi-objective artificial immune algorithm coupled with multi-agent systems, will be proposed in this study. Three key elements of this model will be studied, including: a) a multi-level and multi-dimensional encoding scheme for representing the spatial characteristics of the problems, b) improved immune searching operators based on the competition, collaboration and learning mechanisms of multi-agent systems, c) the parallel computing technology will be introduced to enhance the ability of the model to solve large-scale problems. Furthermore, a case study will be carried out to validate and assess the performance and accuracy of the model. The model proposed in this study are expected to improve the accuracy and efficiency of the intelligent land-use spatial allocation optimization models and provide a novel method for solving other spatial optimization problems. Finally, the ultimate goal of the study is to provide an effective decision-making method for supporting practical land use planning.

土地利用空间优化配置是实现土地资源合理利用的必要途径，也是当前地理科学和土地科学研究领域的热点问题。以遗传算法和人工免疫系统为代表的智能优化算法为土地利用空间优化问题的求解提供了有效工具。然而，在高精度、高效率优化模型构建方面，现有模型存在一定局限性，如难以满足优化问题的多层次空间特征表达需求、寻优过程随机化、大多基于串行环境实现等。针对上述问题，本项目旨在解决智能优化算法在土地利用空间优化领域面临的“空间化”和“机理化”难题，突破大规模优化问题的“并行化”求解技术。提出了包括多层次多维度空间化抗体编码方案、基于多智能体的免疫进化策略和多层次混合并行等关键技术在内的并行多智能体人工免疫优化模型构建方案。研究结果预期将提高智能优化模型的寻优精度和效率，增强优化方案的科学性与可行性，丰富土地利用空间优化模型研究的理论与方法体系，为土地利用规划实践工作的开展提供科学依据和决策支持。

土地利用空间配置受到土地利用利益相关方的多重影响，例如，政府、部门和土地使用者。旨在解决不同土地利益相关方之间潜在的利益冲突的协同式土地利用规划，在近年来越来越得到关注。尽管在已有研究中，元启发式搜索算法被证明为解决土地利用空间优化配置问题的有效工具。然而，在高精度、高效率优化模型构建方面，现有模型依然存在一定局限性，例如难以满足优化问题的多层次空间特征表达需求、寻优过程随机化、大多基于串行环境实现等。针对上述问题，本项目针对智能优化算法在土地利用空间优化领域存在的问题，本项目分别从以下几个方面展开了研究：1）设计了多维度空间显式编码的抗体模型，实现优化模型的空间化；2）提出了面向土地利用空间优化配置的多智能体系统，在此基础上设计了基于智能体的土地利用空间优化配置算法，实现了优化模型的空间化；3）引入了并行计算计算，实现了算法的并行化。为验证模型的有效性，研究分别选取了四川安岳县和湖北武汉市为实验区展开了应用试验。试验结果表明，本研究设计的土地利用空间优化配置模型可以为土地利用规划提供灵活的辅助决策工具。研究结果预期可为我国开展的“多规合一”等相关工作提供重要的技术手段。项目共发表SCI/SSCI收录论文4篇，获得发明专利授权4项，申请发明专利3项。