基于深度学习技术的水稻环境因素产量预测  

Prediction of rice yield with environmental factors based on deep learning technology

作　　者：张春磊 李颜娥 丁煜 罗煦钦 Zhang Chunlei;Li Yan′e;Ding Yu;Luo Xuqin(College of Mathematics and Computer Science,Zhejiang A&F University,Hangzhou 311300,China;Key Laboratory of Forestry Intelligent Monitoring and Information Technology of Zhejiang Province,Hangzhou 311300,China;China Key Laboratory of State Forestry and Grassland Administration on Forestry Sensing Technology and Intelligent Equipment,Hangzhou 311300,China;Hangzhou Lin'an District Agricultural and Rural Information Service Center,Hangzhou 310000,China)

机构地区：[1]浙江农林大学数学与计算机学院,浙江杭州311300 [2]浙江省林业智能监测与信息技术实验室,浙江杭州311300 [3]林业感知技术与智能装备国家林业局重点实验室,浙江杭州311300 [4]杭州市临安区农业农村信息服务中心,浙江杭州310000

出　　处：《电子技术应用》2024年第4期81-86,共6页Application of Electronic Technique

基　　金：浙江省自然科学基金(Q21H180001);浙江农林大学科研发展基金(2019RF065);浙江农林大学创新训练计划项目(2023KX122)。

摘　　要：水稻作为全球重要的粮食作物,准确预测水稻产量在农业发展中起着重要作用。由于水稻在环境因子与其生长机理的作用下往往呈现出非线性的特点,难以对其做出较为准确的预测,因此,提出CE-CGRU水稻产量预测模型,对非线性环境因子Copula熵(CE)方法进行提取特征并与CNN和GRU技术结合在一起。其目的是在水稻品种确定的条件下,识别产量预测的重要特征。根据使用浙江省临安区真实数据分析和比较所提出的模型的性能,构建了其他5个产量预测模型进行对比,分别是MLR、RF、LSTM、GRU和CNN-LSTM。结果显示,CE-CGRU模型的MAE、MSE和MAPE分别为0.677、0.87和5.029%,表明CE-CGRU模型具有更好的能力来捕捉水稻产量与环境因素之间的复杂非线性关系。此外,还对不同的特征选择方法以及不同时间步长进行了比较和分析。Rice is a globally important staple crop,and the accurate prediction of rice yield plays a significant role in agricultural development.Due to the influence of external environmental factors and the growth mechanisms of rice,rice yield often exhibits nonlinear characteristics,making it challenging to make precise predictions.Therefore,the CE-CGRU rice yield prediction model is proposed,which extracts features using the Copula Entropy(CE)method for nonlinear environmental factors and combines them with CNN and GRU technologies.The aim is to identify crucial features for yield prediction under specific rice varieties.Based on the analysis and performance comparison using real data from Lin'an District of Zhejiang Province,the proposed model is compared to five other yield prediction models:MLR,RF,LSTM,GRU,and CNN-LSTM.The results indicate that the CECGRU model achieves a MAE of 0.677,a MSE of 0.87,and a MAPE of 5.029%,demonstrating its superior capability in capturing the complex nonlinear relationship between rice yield and environmental factors.Furthermore,a comparison and analysis of different feature selection methods and time steps are conducted.

关 键 词：水稻产量预测 Copula熵 深度学习 CE-CGRU 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]