改进的BP神经网络模型预测四川GDP

Improved BP Neural Network Models to Predict GDP in Sichuan Province

运用TRAINGDX训练函数对标准BP神经网络进行改进.根据2012年《四川省统计年鉴》相关数据,利用影响国内生产总值(GDP)的6个主要因素,借助Matlab软件平台,建立了6∶5∶1的三层BP神经网络GDP预测模型,实现四川省GDP值的预测.改进后BP神经网络预测相对误差在1%以内,仿真结果同模型外推法比较,预测精度较高.

An improvement of traditional BP Neural network was made with TRAINGDX training function. According to the data in 2012 Statistical Yearbook of Sichuan Province, by using the six main factors of the gross domestic product （GDP）, with the tools of blatlab software platform, the research established a 6：5：1 three - layer BP neural network prediction model of GDP in or- der to predict the GDP value of Siehuan Province. The BP neural network prediction relative error is within 1%, and the simula-tion results, compared with model extrapolation results, indicated that the model has high prediction precision, and has certain feasibility and practicability.