摘要
对1978-2000年影响我国粮食生产的7个因子分别建立了多元回归分析预测模型和BP神经网络多变量输入预测模型。实证研究结果表明,与回归预测模型相比,用BP网络建立的模型经过训练后,可得到影响粮食产出的主要因子及其之间的非线性关系,网络模型新颖,具有很高的预测精度及较好的预测效果,可广泛应用于各种预测研究,有较高的推广价值。
Aiming at the influencing variables and data from the year of 1978 to 2000 in China, this paper sets up multi-regression analysis model and BP artificial neural network model. Take prediction of total corn production as an example, two methods of modeling of corn production prediction based on multi-regression analysis and BP artificial neural network model are introduced in this paper. The results show that after being exercised, the network can provide nonlinear mapping relation between independent variables and dependent variable of corn production in China. The model is novelty, has higher precision and good effect. It can be widely applied in modeling of many forecast area, and also has high generalizing value.
出处
《华东师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2003年第4期73-79,共7页
Journal of East China Normal University(Natural Science)
基金
国家社会科学基金项目(00BJL051)
关键词
多元回归分析模型
BP神经网络模型
粮食生产
预测
比较
BP artificial neural network
multi-regression analysis
BP artificial neural network model
corn production
prediction
comparison
