灰色误差神经网络模型在预测中的应用研究

Application research of grey error term and neural network model on prediction

在分析GM(1,1)模型的建模机理的基础上,指出了传统建模方法的不足,即发现了预测数据序列中的第一点的值并不能用原始数据序列中第一点的值来代替,因为存在误差,同时给出了误差项的一般表达式,然后基于BP神经网络对误差项进行优化模型。结果表明,该模型拟合误差小,预测精度高。

According to the building mechanism of GM （1,1）,the existent weakness is pointed out to of traditional method build grey model,that is,the first point of original data is different with 1st point of predictive value that both exist an error term μ. Furthermore,by the error term μ,this paper formulates other error term of training data.The optimum model is given to error term on the basis of BP neural network.The results of experiment show that the model is valid,feasible and high precision.