中期交通量的多项式和Fourier级数组合预测模型

Combined Forecast Model for Medium-term Traffic Flow Based on Polynomial and Fourier Series

为提高中期交通流量预测的精度,利用交通流存在平峰、高峰这一先验信息对交通量进行多项式趋势回归分析,采用傅里叶级数修复随机因素对交通流造成的影响,从而建立道路交通流量多项式傅里叶级数组合预测模型,最后通过哈尔滨市瓦盆窑路段的交通流数据检验了方法的有效性和稳健性。结果表明,该方法具有较高的预测精度,总体上优于指数平滑、ARIMA、BP神经网络以及RBF神经网络等典型的短时交通流预测方法。

In order to improve the accuracy of medium-term traffic flow prediction,a combined forecast model for traffic volume was established based on polynomials and Fourier series.In the model,the prior information of traffic flow in flat hour and peak hour was used through polynomial trend regression analysis,and Fourier series was adopted to repair the impact of random factors on traffic flow.Finally the stability and effectiveness of the method were tested by a case study using the traffic flow data of road Wapenyao in Harbin.The results show that the method is of high prediction accuracy,and is generally superior to all kinds of exponential smoothing models such as ARIMA,BP neural network,RBF neural networks and other typical short-term traffic flow prediction methods.