基于BP神经网络的大豆变量施肥模型研究与建立

Study and Establish of Variable Fertilizing Modeling Based on BP Neural Network

通过对比国内外处方图生成方法,研究了基于BP神经网络的变量处方施肥关系模型。应用Matlab7.0人工神经网络工具箱,设计了BP人工神经训练网络,实现了以土壤中施氮肥量、磷肥量、钾肥量和种植密度为输入参数,以大豆的产量为预测目标的模拟关系,预测在一定范围内的氮磷钾投入配比可获得的最大产量。该方法能够实现各种营养元素的全面平衡供应,使大豆栽培肥料投入更为科学合理,达到减肥增效的目的。

This paper had compared the generation methods of the prescription map at home and abroad, and established the variable fertilizing model based on BP artificial neural network. This model was designed through Matlab7.0 neural network to predict the maximum yield based on the mount of soil＇s NPK. The method could achieve the balance of a variety of nutrition elements, and could make more reasonable fertilizer inputs and increase efficiency.