基于遗传LMBP神经网络模型的股票预测  

Research on Stock Prediction Based on Ga-LMBP Neural Network

作　　者：齐晓娜[1] 程维刚[1] 

机构地区：[1]河北金融学院,河北保定071051

出　　处：《无线互联科技》2016年第3期86-87,共2页Wireless Internet Technology

基　　金：河北省高等学校科学技术研究项目;项目编号:Z2015116

摘　　要：股票市场是一个高度非线性的系统,通过传统的方法建立较为精确的预测模型比较困难。文章建立了基于遗传算法的LMBP神经网络组合预测模型。利用遗传算法优化BP神经网络的连接权和阈值。采用LMBP算法改进模型的收敛速度。实例验证表明,在建模样本和预测因子相同的条件下,该模型比传统BP网络的预测结果稳定且精度高。Stock market is a highly nonlinear system, and it is difficult to establish a more accurate prediction model by the traditional method. A combined forecasting model of LMBP neural network based on genetic algorithm is established. Using genetic algorithm to optimize the connection weights and thresholds of BP neural network. The convergence rate of the improved model is improved by using LMBP algorithm. Under the same conditions as the modeling samples and the prediction factors, feasibility and validity of this model has been verified by concrete examples.

关 键 词：遗传算法 LMBP算法 组合算法 股票预测 

分 类 号：F830.91[经济管理—金融学] TP183[自动化与计算机技术—控制理论与控制工程]