基于SR_GA_BP神经网络的河南省油料产量预测研究  

Research on oilseed yield forecasting in Henan Province based on SR_GA_BP neural networks

作　　者：张文慧 杨进进 王哲 Zhang Wenhui;Yang Jinjin;Wang Zhe(School of Information Engineering,North China University of Water Resources and Electric Power,Zhengzhou 450046,China)

机构地区：[1]华北水利水电大学信息工程学院,郑州450046

出　　处：《现代计算机》2024年第10期35-39,共5页Modern Computer

摘　　要：油料产量的高效预测对于制定油料生长期间的精准管理决策具有重要意义。为提高河南省油料产量预测效率和精度,收集河南省油料年度产量、油料种植面积等数据,选取9个指标作为影响因素,使用逐步回归分析法(SR)筛选出影响显著且独立的影响因素作为河南省油料产量主要影响因素。针对BP神经网络模型收敛速度慢且易陷入局部最优解的缺陷,引入遗传算法(GA)对其权值和阈值进行优化,以便更好地拟合河南省油料产量与其影响因素之间的复杂非线性关系。仿真结果表明:相比于单一BP神经网络模型和SR_BP神经网络模型,SR_GA_BP神经网络模型具有更高的预测精度,MAPE仅为1.58%,因此SR_GA_BP预测模型可以对河南省油料产量进行更准确的预测。Efficient prediction of oilseed yield is important for making precise management decisions during the growth period of oilseeds.In order to improve the efficiency and accuracy of oilseed yield prediction in Henan Province,we collected data on an-nual oilseed yield and oilseed planting area in Henan Province,selected nine indicators as influencing factors,and used stepwise regression analysis(SR)to screen out the influencing factors with significant and independent effects as the main influencing fac-tors of oilseed yield in Henan Province.Aiming at the shortcomings of BP neural network model,which is slow to converge and easy to fall into the local optimal solution,genetic algorithm(GA)is introduced to optimize its weights and thresholds,so as to better fit the complex nonlinear relationship between oilseed production and its influencing factors in Henan Province.The simulation re-sults show that compared with the single BP neural network model and the SR_BP neural network model,the SR_GA_BP neural network model has higher prediction accuracy,with a MAPE of only 1.58%,so the SR_GA_BP prediction model can predict the oil-seed production in Henan Province more accurately.

关 键 词：油料产量预测 BP神经网络 遗传算法 逐步回归分析 

分 类 号：TS222[轻工技术与工程—粮食、油脂及植物蛋白工程] TP18[轻工技术与工程—食品科学与工程]