中国苹果产量预测模型比较分析

A comparative study on forecast models of apple output in China

提出了一种灰色神经网络模型。该模型结合了灰色GM(1,1)预测模型和BP神经网络2种预测模型的优点,并用此模型对我国的苹果产量进行预测。将GM(1,1)预测模型的预测值作为BP神经网络的输入变量,实际值作为输出变量进行神经网络训练。灰色神经网络模型对2004、2005年苹果产量的预测精度分别为98.11%、98.45%,高于单一的灰色GM(1.1)预测模型或BP神经网络。

In this paper the author uses grey prediction combined with BP neural network method to establish a new model named grey neural network, and the apple output in China is estimated. The forecast data of grey prediction GM （1, 1） were used as the input variable of BP neural network, the actual output were as output variable. Grey neural network forecasted the apple output in China in 2004 ＆ 2005, its precision were 98.11% ＆ 98.45%, higher than any single grey prediction or BP neural network method.