基于BP神经网络的三峡库区开县耕地面积预测研究

开县是三峡库区的一个农业大县,选择开县为研究点,一方面探讨BP神经网络在县域内应用的可行性;另一方面探讨开县耕地变化的过程及其相关影响因子,并预测开县耕地的数量,为开县的耕地资源保护和农业的可持续发展提供保证。利用二元定距变量相关分析的方法分析耕地数量和相关指标的相关性系数,同时利用BP神经网络的方法,训练得出耕地面积与相关指标的关系,然后利用训练网络建立预测模型。通过二元定距变量相关分析得出年末总人口等14个指标与耕地的数量相关性明显。然后通过BP神经网络训练出指标与耕地面积的关系,其训练精度达0.00001,利用该网络预测出了开县2008年、2010年、2015年、2020年的耕地面积。

STUDIES ON FORECASTING THE ARABLE LAND AREAS OF KAI COUNTY IN RESERVOIR REGION OF THREE GORGES BASED ON BP NERVE NETWORK

Kai County is a big agricultural county is the Three Gorges Region.To choose Kai County as the research focus is for one hand to probe into the feasibility of using BP nerve network inside of Kai County and for another hand to discuss the changing procedure of arable land in Kai County and the related influencing factors and to forecast the exact areas of arable land in Kai County,so as to provide guarantee for protecting arable land resources in Kai County and ensuring the sustainable development of agriculture.The duality analytical method with fixed distance and variable correlation is adopted to analyze the number of arable land and correlated index for related targets.At the same time,Bp nerve network is used to get the relation between arable land areas with related targets.Then,the training network is used to set up a forecasting model.By this method,the total population at yearend and other 14 targets show remarkable correlation with arable land number.Then,the relation between targets and arable land areas is trained by BP nerve network.Its accuracy has reached 0.00001.This network has forecasted the arable land areas of Kai County in 2008,2010,2015 and 2020.