期货价格预测的仿真研究

摘要

The problem of futures prices is discussed. Because the affecting factors between future price and the period are nonlinear, the traditional BP network is easy to get into local optimization and the global search capability is not strong. In order to improve the prediction precision, a future price prediction method is put forward based the PSO and BP neural network. PSO-BPNN first uses the improved PSO instead of BPNN initial optimization, then uses the BP algorithm to optimize network parameters optimization, and finally through training establishes particle swarm optimization of BP neural network. The model is applied to the future price prediction research. The simulation results show that PSO-BPNN model has better global optimization ability and can effectively prevent the BP neural network into local optimization, thus effectively improve the prediction precision of the neural network.%研究期货价格准确预测问题,针对期货价格是一种复杂的非线性和突变性系统,传统神经网络在期货价格预测中易陷入局部极小值,预测精度受到影响.为了提高期货的预测精度,提出一种粒子群算法(PSO)优化 BP 神经网络模型的期货价格预测模型.利用 PSO 算法优异的寻优能力对 BP 神经网络参数进行优化,加快 BP 神经网络学习速度,最后将模型应用到期货价格预测研究中,从而提高 BP 期货价格的预测精度.仿真结果表明,经过 PSO 优化的 BP 神经网络模型有效地提高了速度和期货价格的预测精度,为设计提供了参考.