基于BP神经网络的城市增长边界预测——以北京市为例  

The Prediction of Urban Growth Boundary based on BP Artifi cial Neural Networks: An Application to Beijing

作　　者：付玲[1] 胡业翠[1] 郑新奇[1] 

机构地区：[1]中国地质大学(北京)土地科学技术学院,北京100083

出　　处：《中国土地科学》2016年第2期22-30,共9页China Land Science

基　　金：国家自然科学基金资助(41171440);中央高校基本科研业务费专项资金资助(2652015175)

摘　　要：研究目的:构建城市增长边界预测模型,以北京市为例,研究该模型的可行性。研究方法:尝试采用BP人工神经网络方法,结合GIS和RS技术,并选定绿地、建筑物、行政中心、主要道路、次要道路、坡度、坡向和海拔8个对城市边界扩张影响较大的因子,建立城市增长边界模型(UGBM),并应用该模型预测了北京市2020年城市增长边界,同时用面积匹配值法评估了模型的精度。研究结果:使用UGBM模型预测城市增长边界,总的面积匹配值为106%,稍微高估了城市扩张面积。研究结论:基于BP神经网络的UGB划定方法对确定城市未来扩张方向有指导作用,可为城市规划和发展政策的制定提供指导。The purpose of this study is to establish urban growth boundary model and apply it to Beijing to study the feasibility of the model. In this paper, we tried to use BP artificial neural network combined with geographic information systems（GIS） and remote sensing（RS） technology to establish urban growth boundary model（UGBM）. We selected eight factors that might lead to urban boundary expansion for the model i.e. green areas,buildings,administration centers, main roads, minor roads, slope, aspect and altitude. We used this model to predict Beijing urban growth boundary in 2020, and evaluated the accuracy of this model via a percent area match metric. The results showed that the area match value of this model was 106% when predicting Beijing urban growth boundary. Although this model overestimated urban area slightly, it predicted urban growth boundary quite well in general. It concludes that the model can predict urban growth boundary and it can provide certain guidance for urban planning and urban development policy.

关 键 词：土地管理 城市增长边界预测 BP人工神经网络 训练 

分 类 号：F301.2[经济管理—产业经济]