基于粒子群优化小波神经网络的行程时间预测

Travel Time Prediction Based on Particle Swarm Optimization Wavelet Neural Network

为使道路使用者在出发前获得具有高实时性和可靠性的行程时间预测信息,提高出行效率,需提升高速公路行程时间的预测精度。鉴于此,将生物学中粒子群优化算法(Particle Swarm Optimization Algorithm,PSO)引入小波神经网络(Wavelet Neural Network,WNN)中,构建基于粒子群优化小波神经网络(Particle Swarm Optimization Wavelet Neural Network,PSO-WNN)的高速公路行程时间预测模型。首先将高速公路原始收费数据规整化,截取其中有效字段,获取研究路段一个月的行程时间数据并对其进行数据处理。然后分别基于PSO-WNN模型和WNN模型,利用Matlab进行实验。实验结果显示,PSO-WNN模型预测结果的平均绝对误差、平均相对误差和均方误差较WNN模型分别降低了83.36%,82.20%和98.15%。PSO-WNN行程时间预测模型不仅预测精度高,而且能较准确地预测出行程时间的走向及波动情况,在收敛速度方面也呈现出一定的优势,具有较好的适应能力。

In order to make road users obtain highly real-time and reliable travel time prediction information before departure and improve the travel efficiency,the prediction accuracy of the expressway travel time should be improved.Given this,Particle Swarm Optimization(PSO)Algorithm in biology was introduced to Wavelet Neural Network(WNN),and a highway travel time prediction model based on Particle Swarm Optimization Wavelet Neural Network(PSO-WNN)was constructed.First,the original toll data of expressway was normalized,and the valid fields were intercepted.The travel time data covering the research section for one month was acquired and processed.Then,experiments based on PSO-WNN model and WNN model were carried out respectively by using Matlab.The simulation results showed that the mean absolute error,mean relative error and mean square error of the prediction results of PSO-WNN decreased by 83.36%,82.20%and 98.15%respectively,compared with these of WNN.PSO-WNN travel time prediction model not only has high prediction accuracy,but also can accurately predict the travel time trend and fluctuation.It also has certain advantages in convergence speed and adaptability.