Abstract: The land transformation model (LTM), which couples geographic systems (GIS) with artificial neural networks (ANN) to simulate and forecast urban spatial growth is presented. Using ANN with its highly intelligent function, LTM can make a quantitative analysis on the relationship between social, political, environmental factors and urban growth, and then give a prediction of urban spatial expanding. The main approach of LTM includes three steps: first, select some appropriate impact factors and pretreat them; second, set up Artificial Neural Network and train; third, input all factors data and gain the prediction result. A map of urban growth probability was calculated and used to predict future urban patterns. Then, Beijing which underwent a rapid urbanization in the past years is selected as the research area, and it shows that this LTM has a good efficiency and it is a useful tool for simulating urban growth.

Key words: urban spatial expanding, land transformation model (LTM), GIS, artificial neural network(ANN)

摘要： 结合GIS强大的空间分析功能与人工神经网络（ANN）处理非线性适应性信息的独特能力建立一种土地转化模型（land transformation model，LTM），用以定量分析城市土地扩展与社会、政策、环境等因子之间关系，并基于此对城市空间扩展的动态进行模拟与预测。LTM模型的运行主要分为3步：因子选取与数据预处理；建立人工神经网络并输入数据对其进行训练与仿真；应用PID法对人工神经网络的输出进行分析，同时在GIS平台上模拟出城市扩展的动态分布。选取相应的影响因子并运用该模型对北京市的城市扩展进行实证模拟检验与预测，结果表明此LTM模型确实提供了一种定量分析和预测城市空间扩展的方法，能够为城市规划与城市发展政策的制定提供重要的科学参考。

关键词: 城市空间扩展, 土地转化模型(LTM), GIS, 人工神经网络(ANN)