一种线性ARIMA基础上的非线性BP神经网络修正组合方法在粮食产量预测中的运用

Forecast of Grain Output Using a Method of Nonlinear Bp Neural Network Based on Linear ARIMA Modified

对我国粮食产量预测工作作出了重大贡献.通过线性规律和非线性规律的区分,构建了自回归移动平均(ARIMA)和BP人工神经网络的组合预测模型,结论显示:ARIMA和BP算法能够各自对粮食产量序列中的线性和非线性规律实现充分挖掘,拟合精度几乎达到完美的程度,克服了以往单一预测和组合预测中信息挖掘能力不足的缺陷.

In this paper, we made a major contribution to the prediction of grain yield in China. Distinguished by linear rule and the nonlinear, law, constructs the autoregressive moving average （ARIMA） and the combination of BP artificial neural network prediction model, the conclusion shows： ARIMA and BP algorithm can separate the linear and nonlinear rules of grain yield in the sequence of fully, fitting accuracy almost reached the degree of perfection, overcomes the shortcomings of the previous the information of single and combined forecast in mining ability.