摘要
基于简单遗传算法的神经网络训练速度慢、易陷入局部极值,用具有较好的全局搜索能力自适应遗传算法来优化神经网络权值和阈值,设计了基于自适应遗传算法的BP神经网络的股票预测系统.该系统根据对股票历史数据分析,预测股价未来几天时间的走势.结果表明,改进算法具有很强的可行性和高效性.
Neural network based on simple genetic algorithm has very slow training speed and easily gets into the local extremum. The weights and thresholds of BP neural network were optimized by using the adaptive genetic algorithm with good global searching ability. A stock prediction system based on GA- BP history data of the stock was designed. The system could predict the trend of stock market the next several days. Results show that the improved algorithm is effective and feasible in real application.
出处
《合肥学院学报（自然科学版）》
2009年第3期34-36,共3页
Journal of Hefei University :Natural Sciences
基金
安徽省教育厅自然科学基金项目(KJ2008B034)资助
关键词
遗传算法
BP神经网络
金融时间序列
预测
genetic algorithm
BP neural network
financial time series
prediction
