基于改进小波神经网络的汇率预测研究  

Exchange Rate Prediction Based on the Improved Wavelet Neural Network

作　　者：漆玉娟[1] 漆玉虎[2] 李根全[1] 

机构地区：[1]兰州交通大学数理与软件工程学院,甘肃兰州730070 [2]上海理工大学管理学院,上海200093

出　　处：《云南民族大学学报（自然科学版）》2012年第1期40-43,56,共5页Journal of Yunnan Minzu University:Natural Sciences Edition

基　　金：甘肃省自然科学基金(3ZS051-A25-030;3ZS-042-B25-049);兰州交通大学科研基金(DXS2010-021)

摘　　要：选取2010-01-01—2010-10-31期间内,美元兑人民币的汇率基准价,以Morlet为母小波基函数,采用紧密结合的的小波神经网络对汇率基准价作非线性逼近,并在此小波神经网络基础上进行改进,并通过Matlab软件对原网络与改进网络的训练过程进行了数值仿真.仿真结果表明,改进网络模型对汇率基准价的预测是可行的,其预测精度更高.Based on the benchmark exchange rates between U. S dollars and RMB from January 1,2010 to October 31, 2010 with the Morlet as the mother wavelet basis function and the application of the closely combined wavelet neural network to the benchmark exchange rate for nonlinear approximation, it improved this wavelet neural network and used the software Matlab to perform numerical simulations. The results showed that the improved network model for the prediction of currency benchmark is feasible, and the prediction is more accurate.

关 键 词：汇率 小波神经网络 改进 基准价 预测 

分 类 号：F224[经济管理—国民经济]