基于HS-BP算法的股票预测

为了有效预测股票问题,针对股票走势具有随机波动性和非线性的特点以及BP神经网络收敛速度慢、容易陷入局部极小值的缺点,提出利用和声搜索算法优化BP网络权重的方法改进神经网络,并通过对股票预测指标的分析,建立了股票预测模型,将和声搜索算法优化BP网络应用于所建立的模型中求解.实验结果表明,将HS算法和BP神经网络有机结合,加快了网络收敛速度、避免了局部极小值,有效地刻画了股票的随机波动特性,提高了股票预测的准确性.

Stock Forecast Based on HS-BP Algorithm

For the purpose of predicting stock effectively,aimed at the characteristics of stochastic fluctuation and nolinear and the disadvantage that BP network’s convergence speed was slow and it fell into part minimum value easily,the method that harmony search algorithm optimized BP network’s connection weight and threshold value to improve natural network was proposed.At the same time,with the analysis of stock’s forecast,the mathematical model of evaluating stock’s forecast was constructed.At last,the harmony search algorithm improved BP network algorithm was applied to solve the constructed model.The results show that combining HS and BP can overcome flaws of easily getting into minimum and slow convergence,effectively describe the stochastic fluctuation of stock and improve the performance and improve the accuracy of forecasting stock.