基于免疫智能体的多目标湿地景观变化模拟与格局优化模型

项目名称： 基于免疫智能体的多目标湿地景观变化模拟与格局优化模型

项目编号： No.41271196

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 天文学、地球科学

项目作者： 张树清

作者单位： 中国科学院东北地理与农业生态研究所

项目金额： 75万元

中文摘要： 在进行湿地景观模拟与格局优化过程中，必须要面对：农业生产与环境保护的矛盾；多目标优化与有限可利用总体资源和空间的矛盾。传统方法通常难以应对这些矛盾并提出行之有效的模拟与优化方案。 因此本研究拟：(1)建立基于免疫智能体的多目标湿地景观变化模拟与格局优化模型；(2)基于多智能体与元胞自动机相结合模拟技术，建立兼顾景观驱动力与多维度地理空间信息的模拟优化环境；(3)在多智能体的控制管理部分引入免疫机制，实现多种目标的整合与优化，提高多维复杂问题求解能力，构建复杂生态系统景观模拟与优化理论与方法体系。 以三江平原湿地为试验区，有效模拟湿地景观在不同自然与人为作用下的形成与演化过程，预测其发展趋势；辨识其变化的关键驱动因子及作用，获得更加可行的、兼顾各方利益的湿地-农田景观格局优化方案，为区域可持续发展提供决策支持，研究内容和目标将具有重要科学实用价值。

中文关键词： 景观变化模拟；多目标优化；多智能体；人工智能；湿地

英文摘要： Contradiction between farming and the environment protection as well as that between multi-objective optimization and limited available resources and spaces are the questions which we have to face in the process of wetland landscape simulating and pattern optimizing. Traditional methods generally can't deal with these problems efficiently. Considering the previous issues, this project intends to achieve three goals: first, set up a model for multi-objective wetland landscape simulation and pattern optimization based on artificial immune inspired multi-agent system; second, create a simulation optimizing environment which combines with multi-agent and cellular automata, as well as taking into account driving forces and multi-dimensional geospatial information; third, introduce immune mechanisms to control and management of multi-agent system, in order to not only achieve multi-objective integration and optimization but also improve the ability of settling multi-dimensional questions. Based on these researches, a theory and method system for complex ecosystem landscape simulation and optimization will be constructed. Selecting Sanjiang Plain as our study area, we intend to effectively simulate the wetland landscape formation and evolution processes which are caused by different natural and human activities. Furth

英文关键词： process modeling of landscape dynamics；multi-objective optimization；multi-agent；artificial intelligence；wetland