基于PSO-BP神经网络的股市预测模型

为解决股市短期预测中BP神经网络法对初始权值敏感、易陷入局部极小值和收敛速度慢等问题,基于全局随机优化思想的粒子群优化(PSO)算法,对BP神经网络的初始权值进行了优化,建立了PSO-BP神经网络股市预测模型.并通过实例分析与传统BP神经网络预测法进行对比,结果表明该方法收敛速度明显提高,有效降低了训练误差,避免了陷入局部极小值,达到了较高的预测精度,在股市短期预测中具一定的实用价值.

Stock market forecast model based on PSO-BP neural network

In order to overcome the problems of the over-fitting,local minima and low convergence speed in Back Propagation(BP) neural network method,the PSO-BP prediction model concerning stock price forecast was developed,which was based on the Particle Swarm Optimization(PSO) algorithm with the global stochastic optimization idea.The experimental results show that compared with the traditional BP method,this one increases the convergence speed,effectively reduces the training error,avoids falling into local minima ...