基于神经网络的我国猪肉年度消费需求量预测研究

PREDICTION STUDY ON PORK ANNUAL CONSUMPTION DEMAND IN CHINA BASED ON NEURAL NETWORKS

主要运用神经网络对中国猪肉年度消费需求量的预测问题进行了研究.首先对畜产品预测方法进行了综述,对神经网络预测原理进行了介绍,然后分别建立GM(1,1)、ARIMA模型、GRNN神经网络模型,GA-GRNN神经网络模型模型对猪肉消费量进行了预测,最后对模型进行了集成.结果表明,优化后的GA-GRNN模型预测误差率显著降低,集成模型在稳定性和精度方面均比较理想.

In this paper,the issues on prediction of pork annual consumption demand in China mainly using neural networks are studied.Firstly,the prediction methods of animal products are reviewed,and the principles of neural networks are introduced,then we establish GM（1,1） model,ARIMA model,GRNN neural network model,GA-GRNN neural network model respectively to predict the pork consumptions. At last,we integrate these models.The results show that prediction error of the optimized GA-GRNN model is significantly reduced,and integrated model is ideal both in stability and accuracy.