基于灰色-BP神经网络模型的多情景交通用地需求预测——以长江中游城市群为例
投稿时间:2019-08-25    点此下载全文
引用本文:史云扬,李牧,付野,王立威,孙敏轩,郝晋珉.基于灰色-BP神经网络模型的多情景交通用地需求预测——以长江中游城市群为例[J].中国农业大学学报,2020,25(6):142-153
摘要点击次数: 70
全文下载次数: 57
作者单位E-mail
史云扬 中国农业大学 土地科学与技术学院, 北京 100193
自然资源部 农用地质量与监控重点实验室, 北京 100193 
yunyangshi@cau.edu.cn 
李牧 中国农业大学 土地科学与技术学院, 北京 100193
自然资源部 农用地质量与监控重点实验室, 北京 100193 
 
付野 北京市规划和自然资源委员会通州分局, 北京 101100  
王立威 自然资源部 国土整治中心, 北京 100035  
孙敏轩 中国农业大学 土地科学与技术学院, 北京 100193
自然资源部 农用地质量与监控重点实验室, 北京 100193 
 
郝晋珉 中国农业大学 土地科学与技术学院, 北京 100193
自然资源部 农用地质量与监控重点实验室, 北京 100193 
jmhao@cau.edu.cn 
基金项目:国家科技支撑计划(2015BAD06B01)
中文摘要:为探索科学预测区域交通用地需求的合理途径,指导国土空间规划编制中交通用地规模的划定,拟通过灰色-BP神经网络模型识别主要社会经济影响因素,以长江中游城市群为例,在预测其远景交通用地需求规模的同时,基于城市群发展的阶段特征选取典型城市群样本设置3类情景,对不同情景下的交通用地需求分别进行预测。结果表明:1)城镇化水平、产业结构高度化程度和劳动力资源禀赋是当前影响长江中游城市群交通用地需求的主要社会经济因素。2)通过系统仿真试验对比不同方法的交通用地需求预测结果,可以发现基于灰色-BP神经网络模型的预测方法精度较高,误差较小,该预测方法对于区域交通用地规模的预测具有一定的适用性。预测得到的长江中游城市群2020和2030年交通用地需求分别为31.22万和49.07万hm2。3)不同情景下长江中游城市群交通用地需求预测结果存在明显差异,底线情景可作为划定交通用地规模的底限,一般情景可作为基准,极限情景可作为红线,长江中游城市群交通用地合理规模应以基准为参考,介于底线和红线之间。
中文关键词:灰色-BP神经网络模型  多情景  交通用地  预测
 
Multi-scenario traffic land demand forecasting based on grey system-BP neural network model: A case study of urban agglomeration in the middle reaches of the Yangtze River
Abstract:In order to explore a reasonable way to scientifically predict the demand for regional traffic land and guide the delimitation of the scale of traffic land in the compilation of territorial spatial planning, a grey-BP neural network model is proposed to identify the main socio-economic factors. Taking the urban agglomeration in the middle reaches of the Yangtze River as study case, the typical cities are selected based on the stage characteristics of urban agglomeration development while forecasting their future demand scales for traffic land are forecasted. Three scenarios were set up to forecast the demand of traffic land. The results show that: 1)The level of urbanization, degree of industrial structure and endowment of labor resources are the main social and economic factors affecting the demand of traffic land in the middle reaches of the Yangtze River. 2)By comparing the results of different methods of traffic land demand prediction through system simulation experiments, it is found that the prediction method based on grey-BP neural network model has higher accuracy and less error, and this prediction method has certain applicability to the prediction of regional traffic land scale. The forecasted demand for traffic land in the middle reaches of the Yangtze River in 2020 and 2030 are 312 200 and 490 700 hectares, respectively. 3)Under different scenarios, the results of traffic land demand forecasting in the middle reaches of the Yangtze River are obviously different. The baseline scenario can be used as the bottom line for delineating the scale of traffic land, the general scenario can be used as the benchmark, and the ultimate scenario can be used as the red line. The reasonable scale of traffic land in the middle reaches of the Yangtze River should be based on the benchmark, which is between the bottom line and the red line.
keywords:grey system-BP neural network model  multi-scenario  traffic land use  prediction
查看全文  查看/发表评论  下载PDF阅读器