基于非参数样条拟合的交通流预测方法研究

准确的交通流短时预测是智能交通系统,尤其是其先进的交通管理系统与先进的出行者信息系统研究的关键内容之一。随着预测时间跨度的缩短,交通流量的变化显示出越来越强的不确定性,使得一般预测方法的预测精度大大降低。针对智能交通系统的开发,论文将样条拟合的思想应用到交通流预测领域,利用贝努利多项式求解核函数,进而利用非参数回归理论进行交通流预测。经过实测数据仿真试验表明,样条拟合能较好地兼顾最优拟合与曲线光滑度的选择,算法的预测效果良好。

Study on Traffic Flow Forecasting Method Based on Non-parameter Regression Spline Fitting

Accurate short-term traffic flow forecasting is becoming a crucial step in ITS research,especially,for its Advanced Traffic Management System and Advanced Traveler Information System research.With the shortening of the forecasting term,the uncertainty of traffic flow becomes more and more seriously,so that the forecasting effect of general approaches is decreasing.Aimming to the development of ITS,this paper applies the theory of spline fitting to traffic flow forecasting and solves the kernel function with Bernoulli polynomial.And then,traffic flow condition can be forecasted using the theory of non-parameter regression.Results of the test using field data show that the spline fitting can make a better compromise of best fitting and smoothness,and the algorithm performs better in short-term traffic flow forecasting.