摘要
为预测江苏省粮食产量,通过分析相关文献,选择农业机械总动力、有效灌溉面积、农用化肥施用折算纯量、除涝面积、农用柴油使用量、农药使用量、粮食作物播种面积和受灾面积这8个对粮食产量具有较大影响的指标,构建反向传播神经网络模型,并利用该模型预测2017年、2018年和2019年的粮食产量。模型以江苏省1993—2019年的数据为基础,剔除缺失指标的数据,将剩余的数据分为2组,一组数据作为训练集用于训练模型参数,包含21个样本;另一组数据作为测试集检验拟合模型的预测精度,包含5个数据样本。最终预测结果除了2019年预测结果的相对误差达到了5.74%,其他年份的预测结果相对误差基本控制在5%以内。结果表明,该模型具有较好的预测精度,能够有效预测粮食产量,为将来粮食产量的预测提供了一种新的思路。
In order to predict the grain yield of Jiangsu province,in this paper we analyze the relevant literature,and select the total power of agricultural machinery,effective irrigation area,converted pure amount of agricultural chemical fertilizer application,waterlogging area,agricultural diesel oil usage,pesticide usage,grain crops,and disaster area.The eight indicators that have greater impact on grain yield are used to construct a back propagation(BP)neural network model.Based on the data of Jiangsu province from 1993 to 2019,1999 with missing index data is eliminated.The remaining data are divided into two groups.One group contains 21 samples as a training set for training model parameters.A set of five data sets is used as test sets to test the prediction accuracy of the fitting model.The relative error of the final prediction results is basically controlled within 5%,except that the prediction data in 2019 reached 5.74%.These show that the model has a good accuracy in effective prediction predict grain output,and provides a new idea for future grain output prediction.
作者
于涧
洪欣
于泽翔
马涛
YU Jian;HONG Xin;YU Zexiang;MA Tao(College of Mathematics and Systems Science,Shenyang Normal University,Shenyang 110034,China;Sydney Smart Technology College,Northeast University,Shenyang 110819,China)
出处
《沈阳师范大学学报（自然科学版）》
CAS
2023年第4期316-320,共5页
Journal of Shenyang Normal University:Natural Science Edition
基金
教育部产学合作协同育人项目(22060022827110)。
关键词
反向传播神经网络
正向传播
梯度下降法
粮食产量影响因子
back propagation neural network
positive communication
gradient descent method
impact factors of grain yield
