正交尺度小波网络及在非线性经济系统预测中的应用

The Application of Wavelet Neural Network Based on Orthogonal Scaling Function in Nonlinear Macroeconomic Forecasting

本文提出一种用正交尺度函数代替RBF网络中的激活函数的小波网络,给出相应小波网络学习算法;并以天津市国内生产总值为样本进行宏观经济模拟预测。

In this paper, a waveletbased neural network is given. The structure of this network is similar to that of the radial basis function(RBF)network, except that here the radial basis functions are replaced by orthogonal scaling functions. Furthermore, the author proposes the algorithms for the wavelet network.Finally,a macroeconomic forecasting based on the sample of GDP in Tianjing is presented, which shows that the forecasting error is less than that of BP network.