土壤含水率与像素颜色之间关系的BP人工神经网络模型  

BP Artificial Neural Network Model of the Relationship between Soil Water Content and Pixel Color

作　　者：李盼盼[1] 缴锡云[1] 马海燕 朱霞[1] 

机构地区：[1]河海大学水文水资源与水利工程科学国家重点实验室 [2]山东水利科学研究院,济南250013

出　　处：《中国农学通报》2008年第1期463-466,共4页Chinese Agricultural Science Bulletin

基　　金："十一五"国家科技支撑计划"灌区田间节水改造技术集成模式研究"(2006BAD11B02);水文水资源与水利工程科学国家重点实验室开放研究基金"膜沟灌溉入渗规律试验研究"(2005405811)

摘　　要：建立像素颜色RGB值与土壤含水率之间的数学关系,是染色入渗法的应用基础。结合沟灌染色入渗试验,研究了染色入渗过程中土壤含水率与像素颜色分量之间关系的BP人工神经网络模型。分析土壤含水率与像素颜色分量之间的关系,确定BP人工神经网络的拓扑结构,以像素颜色分量的相对值作为输入因子,土壤含水率作为输出因子,建立了包含1个隐层的BP人工神经网络。结果表明,该模型具有较高的拟合精度和验证精度,优于二次多项式模型。It is necessary to quantify the for the application of dye infiltration. A relationship between pixel color RGB value and soil water content BP artificial neural network model of the relationship between soil water content and pixel color RGB value was researched based on dye infiltration experiment. The topologi- cal structure of a BP Artificial Neural Network was determined based on the relationship between soil water content and pixel color component. Using the relative value of pixel color component as input and soil water content as output, a one - hidden-layer BP Neural Network model was developed. Results show that the proposed model has higher fitting precision and predicting precision, and is superior to quadratic polynomial model.

关 键 词：染色入渗法 土壤含水率 BP人工神经网络 

分 类 号：S152.7[农业科学—土壤学]