基于改进RBF神经网络对股价的演变预测  

Prediction of the Evolution of Stock Prices Based on Improved RBF Neural Network

作　　者：郭兰平[1] 俞建宁[1] 张建刚[1] 张旭东[1] 漆玉娟[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

出　　处：《兰州交通大学学报》2010年第1期141-145,共5页Journal of Lanzhou Jiaotong University

基　　金：甘肃省自然科学基金(0803RJZA012)

摘　　要：对RBF神经网络进行了分析,建立了RBF神经网络模型,并对此模型进行了改进,使其具有更好的预测性能.把一类非线性较强的时间序列(万科A股2009年6月份股票价格)利用该模型进行了非线性逼近.用Matlab软件对网络的学习与训练过程进行了数值仿真.实验结果表明:利用改进后的网络模型对非线性时间序列进行短期预测是可行的,其预测精度高于改进前的预测精度,改进方法有效.RBF neural network is analyzed in this paper,and RBF neural network model is established, too. The model is improved to have better prediction performance. A strong class of nonlinear time series（stock price of Wanke A in June of 2009）is approached by using the model. Matlab software is applied to carry out numerical simulation of the network＇s learning and training process. Simulation results show that the improved network model for short-term prediction of nonlinear the series is feasible, and its prediction accuracy is higher than before. The improved method is effective.

关 键 词：RBF神经网络 预测模型 数值仿真 股价 

分 类 号：F224[经济管理—国民经济]