深层灌水条件下基于BP人工神经网络方法的冬小麦根系分布预测模型  

BP Artificial Neural Network Prediction Model for Winter Wheat Root Distribution Under Deep Irrigation Conditions

作　　者：盖志远 孙西欢[1,3] 马娟娟[1] 郭向红[1] GAI Zhi yuan;SUN Xi huan;MA Juan juan;GUO Xiang hong(College of Water Resources Science and Engineering, Taiyuan University of Technology, Taiyuan 030024, China;Shanxi Provincial Flood and Drought Prevention Center, Taiyuan 030002, China;Jinzhong University, Jinzhong 030619, Shanxi, China)

机构地区：[1]太原理工大学水利科学与工程学院,太原030024 [2]山西省水旱灾害防御中心,太原030002 [3]晋中学院,山西晋中030619

出　　处：《节水灌溉》2019年第7期41-44,55,共5页Water Saving Irrigation

基　　金：国家自然科学基金项目(51579168);山西省研究生教育创新项目(2018SY028)

摘　　要：对冬小麦实施深层灌水是一种高效的节水灌溉方式,为了简化深层灌水条件下冬小麦根系研究工作,建立了以土层深度、发育时间、土层根系日均吸水量、土层日均温度、地上部干重、株高为输入因子,土层根系密度为输出因子的BP人工神经网络预测模型。研究结果表明:在所建立的预测模型下,训练样本各土层根系密度预测值与实测值之间平均相对误差为5.92%,检验样本的平均相对误差为7.30%,训练样本和检验样本都具有较高的精度,因此以该模型对深层灌水条件下冬小麦根系分布情况进行预测是可行的,可为深层灌水条件下冬小麦的根系研究提供新方法。Deep irrigation of winter wheat is an efficient water-saving irrigation method. In order to simplify the research of winter wheat root system under deep irrigation conditions, this article builds a BP artificial neural network prediction model, which uses soil depth, development time, daily water absorption of soil roots, daily average temperature of soil layer, dry weight of shoots and plant height as the input factors, and uses root density as the output factor. The experiment results show that under the established prediction model, the average relative error between the predicted and measured values of root densities of the training samples is 5.92%, and the average relative error of the test samples is 7.30%. Both the training samples and the test samples have higher precision. So, this model is feasible to predict the distribution of winter wheat roots under deep irrigation conditions. It can provide a new method for root research of winter wheat under deep irrigation conditions.

关 键 词：深层灌水 冬小麦 根系分布 BP人工神经网络 

分 类 号：S126[农业科学—农业基础科学]