摘要
提出一种基于RBF神经网络的美元兑人民币汇率预测模型,该模型通过对近两年美元兑人民币汇率的历史数据分析,采用了改进的K-均值聚类算法,动态地确定RBF神经网络中心,并采用最小二乘法进行RBF神经网络的权值调整,通过美元兑人民币汇率的预测,结果表明该模型有较好的预测和泛化能力,可以取得好的预测结果。
A forecast model of US dollar/RMB exchange rate based on RBF neural network is presented.The historical data of US dollar/RMB exchange rate near two years is analyzed by the model,the model is developed by combining the k-clustering algorithms to search for the network centers,and the network weight is adjusted by OLS(orthogonal least square) algorithms.The prediction result of US dollar/RMB exchange rate shows that the model has a good forecast and pan-ability and can obtain a good forecasting result.
出处
《信息技术》
2009年第12期24-26,共3页
Information Technology
基金
甘肃省教育厅研究生导师科研项目计划资助(0704-11)
兰州交通大学"青蓝"人才工程基金资助计划(QL-06-10B)
关键词
人民币汇率
RBF神经网络
K-均值
预测
RMB exchange rate
RBF neural network
K-means algorithms
prediction