摘要
本文基于BP神经网络模型 ,进行了 2 0 0 1～ 2 0 10年我国粮食产量的预测。通过对比传统的“平均增长率一阶滞后模型”拟合及预测 1992～ 2 0 0 0年粮食产量与实际产量的误差值大小 ,可明显看出BP神经网络对于处理单输入单输出的时间序列预测问题是一种更具优越性的方法 ,它具有很强的学习与泛化 (推广 )能力 ,具有很好的应用价值。
Based in the BPNN model, forecasts of Chinese grain output are done from 2001 to 2010. According to comparison of the mistake in Chinese grain output of forecasting and practice from 1992 to 2000 between forcecating model, and the BPNN model, the BPNN model is more superiority on the time suite forecasting and assumes the stronger learning and generalization ability. For dealing with the nonlinear time suite regression and forecasting of single input and single output, the BPNN model assumes good application value.
出处
《预测》
CSSCI
2002年第3期79-80,共2页
Forecasting
