基于智能体与元胞自动机的滇池流域不透水表面扩张模拟  

Simulation of Impervious Surfaces Expansion of Dianchi Lake Sub-watershed based on Cellular Automaton and Agent-based Modeling

作　　者：苏飞[1,2] 杨昆[3,2] 许泉立[1,2] 王桂林[1,2] 刘涛[1,2] 

机构地区：[1]云南师范大学旅游与地理科学学院,昆明650500 [2]西部资源环境地理信息技术教育部工程研究中心,昆明650500 [3]云南师范大学信息学院,昆明650500

出　　处：《湖北农业科学》2015年第15期3763-3768,共6页Hubei Agricultural Sciences

基　　金：国家科技支撑计划项目(2013BAJ07B00);云南社会发展科技计划项目(2012CA024);高等学校博士学科点专项科研基金项目(20115303110002)

摘　　要：本文所建立模型分ANN-CA层与多智能体层,其中ANN-CA层运用神经网络代替元胞自动机原本复杂的转换规则,模拟自然环境对不透水表面变化的影响;多智能体层通过构建政府、居民、农民等3种智能体模拟人类对不透水表面的影响。利用Repast模拟平台构建扩张模型,预测其未来扩张趋势,并验证其精度,点对点精度为71.71%,Kappa系数达到80%以上。结果表明,滇池流域不透水表面2002-2014年间发展速度较快,而在未来不透水表面增速会趋于平缓,且持续扩张,并且将围绕滇池流域形成环滇池流域不透水表面圈。In this paper, the established model was divided into ANN-CA layer and Muhi-agents layer. The ANN-CA layer replaced the original complex cellular automata transform rules by using neural network, simulating the effect of natural envi- ronment of the impervious surface changes. Multi-agents layer simulated the effect of human of impervious surface by building the government agents, resident agents, farmer agents. Using Repast simulation platform the expansion model was built,which predicted the future trend of expansion, and verified its accuracy. Its point-to-point accuracy was 71.71% and the Kappa coefficient was above 80%. Impervious surface results showed that the Dianchi Lake watershed developed faster from 2002 to 2014, but in the future growth of impervious surface will flatten out. Impervious surface will continue to expand and form a ring of impervious surface around Dianchi Lake watershed.

关 键 词：滇池流域 智能体模型 元胞自动机 地理信息系统 不透水表面 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术] P237[自动化与计算机技术—计算机科学与技术]