改进PSO优化的BP神经网络短时交通流预测

BP Neural Network Short-term Traffic Flow Prediction Based on Improved Particle Swarm Optimization

针对目前城市短时交通流预测精度不高的问题,提出一种基于改进粒子群算法优化的BP神经网络短时交通流预测方法。在粒子群算法迭代过程中,当判断算法未成熟收敛时,除最优值对应的个体外,其它部分个体执行遗传算法交叉与变异操作,另外部分个体随机初始化。仿真结果表明,新算法有效提高了收敛精度与稳定性,将其应用于BP神经网络权阈值优化,有效提高了预测精度。在此基础上,开发了基于BP神经网络城市短时交通流预测软件,形象直观,简洁高效,可应用于城市短时交通流预测。

Aiming at the low accuracy,the paper proposed a urban short-term traffic flow prediction based on a BP neural network which uses an improved particle swarm optimization algorithm.In the iterative process of the novel algorithm,when the algorithm was immature convergent,except for the individual corresponding to the optimal value,other parts of the individual performed the genetic algorithm crossover and mutation operation,and some other individuals randomly initialized.The simulation results show that the convergence accuracy and stability can be effectively improved,and the new algorithm was applied to BP neural network weight threshold optimization,which effectively improves the prediction accuracy.On this basis,the short-term traffic flow prediction software based on BP neural network was developed,which is intuitive,simple and efficient,and can be applied to urban short-term traffic flow prediction.