区域耕地数量变化预测方法的对比研究  

Comparison on the Methods for Predicting Regional Cultivated Land Quantity Changes

作　　者：车明亮[1] 聂宜民[1] 刘登民[1] 张建华[1] 陈红艳[1] 王硕 

机构地区：[1]山东农业大学资源与环境学院,山东泰安271018 [2]山东省费县国土局,山东费县273400

出　　处：《中国土地科学》2010年第5期13-18,共6页China Land Science

摘　　要：研究目的:对区域耕地数量变化预测方法进行对比分析,科学合理确定区域耕地数量。研究方法:通过Pearson相关分析法筛选影响耕地变化的关键因子,利用改进的BP神经网络算法、灰色模型和多元线性回归模型法对费县耕地数量变化进行预测。研究结果:改进的BP神经网络算法相对传统的灰色模型和多元线性回归模型等预测方法具备较高的预测精度。研究结论:改进的BP神经网络算法是进行耕地数量变化预测的较好方法,具有实际指导意义,其预测结果可以为当地相关部门合理地确定耕地保有量、推进耕地保护提供科学依据。The purpose of this paper is to compare and analyze the methods for predicting the quantity change of regional cultivated land in order to determine the regionalfarmland quantity scientifically. The pivotal factors affecting cultivated land changes were selected by Pearson Correlation Analysis. The methods of Enhanced BP Neural Network Algorithm, Gray Model and Multiple Linear Regression Model were used to predict the changes of cultivated land quantity in Feixian County. The result indicates that the Enhanced BP Neural Network Algorithm has higher prediction accuracy compared with the traditional Gray Model and Multiple Linear Regression Model. It is concluded that the Enhanced BP Neural Network Algorithm is a better way in predicting the changes of cultivated land quantity, and is meaningful for its practical implications. The predicting results can provide scientific references for determining a reasonable quantity of protected farmland and effectively ensuring the cultivated land conservation at the local level.

关 键 词：耕地变化预测 改进的BP神经网络 灰色模型 多元线性回归模型 

分 类 号：F301.2[经济管理—产业经济]