基于遗传算法优化BP神经网络下马铃薯产量预测模型  

Optimizing Potato Yield Forecast Model Based on Genetic Algorithm under BP Neural Network

作　　者：孙炬仁[1] Sun Juren(Shuozhou Vocational and Technical College,Shuozhou 036002,China)

机构地区：[1]朔州职业技术学院,山西朔州036002

出　　处：《农机化研究》2023年第6期53-57,共5页Journal of Agricultural Mechanization Research

基　　金：山西省重点研发计划项目(201803D221008-4);山西省高等学校科技创新基金项目(2015145);晋中市重点研发计划项目(Y192011)。

摘　　要：马铃薯产量的高效预测对于制定马铃薯生长期间的精准管理决策具有重要意义。为此,针对传统BP神经网络在产量预测中存在的精度差、准确度低等问题,选择遗传算法对单一BP神经网络模型开展网格优化。基于朔州市朔城区沙楞河村2010-2019年田间物联网获取的田间环境数据(土壤含水率和土壤温度)、气象环境数据(大气湿度、大气温度、降雨量)和马铃薯产量,采用BP神经网络及GA-BP神经网络模型对所选地区马铃薯产量进行预测分析。研究结果表明:GA-BP神经网络模型下,马铃薯产量的预测精度明显高于BP神经网络模型,R 2达到0.99327,平均相对误差仅为0.88%。试验证明,GA-BP神经网络模型能够更加科学、合理地进行马铃薯产量预测,说明利用遗传算法优化BP神经网络在马铃薯产量预测中是可行且有效的。Efficient prediction of potato yield is of great significance for making precise management decisions during potato growth.In view of the poor precision and low accuracy of traditional BP neural network in yield prediction,this paper selects genetic algorithm to develop a single BP neural network model.Grid optimization,based on field environmental data(soil moisture content and soil temperature),meteorological environmental data(atmospheric humidity,atmospheric temperature,rainfall)and potato production obtained from the field of Internet of Things in Shalenhe Village,Shuocheng District,Shuozhou City from 2010 to 2019,Using BP neural network and GA-BP neural network model to predict and analyze potato output in selected areas.The research results show that under the GA-BP neural network model,the prediction accuracy of potato yield is significantly higher than that of the BP neural network model,R 2 reaches 0.99327,and the average relative error is only 0.88%.Experiments have proved that the GA-BP forecasting can be used to predict potato yield more scientifically and reasonably,indicating that the use of genetic algorithms to optimize the BP neural network is feasible and effective in predicting potato yield.

关 键 词：马铃薯 产量 预测模型 神经网络 GA-BP 

分 类 号：S11.7[农业科学—农业基础科学]