基于改进BP网络的经济发展预测模型及应用

分析了标准BP神经网络用于复杂经济系统预测的原理以及其中存在的弊端,通过综合运用附加动量法、改变作用函数的方法、以及将预测对象从生产总值改为生产总值增长率的方法,改进BP神经网络用于经济预测时的精度、收敛速度及易陷入局部极值等问题,并将改进后的新方法运用于江苏省第三产业生产总值的预测,通过实验证明改进的BP神经网络在经济系统预测的应用中具有良好的效果。

Improvement of BP ANN Application in Economic Forecasting

Analyze the principle of actual application of ANN(artificial neural network) in economic forecasting and the major drawback of it.Including theoretical knowledge of additional momentum,alteration of action function,replace gross domestic product with growth rate of it for forecasting object.Unify these four means to improve the precision of forecasting,convergence rate,and solve the problem of plunge local extremum.The emphasis of the thesis is on application of the new means to the economic data of tertiary occupation of Jiangsu province.The new means has obvious comparative advantage to classical means that BP ANN application in economic forecasting.