摘要
针对新疆北部地区在玉米生产过程中存在的水肥利用率低下问题,提出一种基于麻雀搜索算法和广义回归神经网络模型的玉米产量预测方法,以探索水肥因素与玉米产量间的非线性关系,可为优化新疆北部地区春播玉米灌溉制度及水肥配饰方略提供理论指导.该方法采用36组既有玉米生产数据,通过广义回归神经网络技术建立模型,使用麻雀搜索算法对关键参数进行优化,降低了人为因素对模型超参数选择的影响.该模型用于新疆灌溉中心试验站的玉米产量预测仿真实验,证明该模型预测性能良好,相较于传统模型,在学习速度、预测精度及鲁棒性上具有优势,可有效预测新疆北部地区春播玉米产量.
In view of the low utilization rate of water and fertilizer in the process of maize production in Northern Xinjiang,a maize yield prediction method based on.Sparrow Search Algorithm and General Regression Neural Network model is proposed to explore the nonlinear relationship between water and fertilizer factors and maize yield,which can provide theoretical guidance for optimizing the irrigation system and water and fertilizer accessories strategy of spring maize in Northern Xinjiang.Using 36 sets of existing production data,the model is established by General Regression Neural Network technology,and the key parameters are optimized by Sparrow Search Algorithm to reduce the influence of human factors on the selection of super parameters.The model was applied to the simulation experiment of maize yield prediction in the experimental station of Xinjiang irrigation center.It is proved that the model has good prediction performance.Compared with the traditional model,it has advantages in learning speed,prediction accuracy and robustness,and can effectively predict the spring maize yield in Northern Xinjiang.
作者
崔兴华
靳晟
姚芷馨
马云鹏
CUI Xing-hua;JIN Sheng;YAO Zhi-xin;MA Yun-peng(School of Computer and Information Engineering,Xinjiang Agricultural University,Urumqi 830052,China)
出处
《数学的实践与认识》
2022年第7期88-96,共9页
Mathematics in Practice and Theory
基金
新疆高技术研究发展计划“农村信息数据标准与资源整合”(2015X0102)
新疆农业信息化工程技术研究中心基金项目“玉米微灌水肥一体化自动控制技术示范与推广”(2520HXKT2)。
关键词
玉米
水肥因素
产量预测
麻雀搜索算法
广义回归神经网络
maize
water and fertilizer factors
yield forecast
sparrow search algorithm
general regression neural network
作者简介
通信作者:靳晟。
