基于GRNN模型的城市道路短期交通流预测研究

采用剔出了城市道路短期交通流的周周期性特征的周差分数据作为广义回归神经网络（GRNN）模型的预测对象,这样既能避免合理选择交通流影响因素作为神经网络输入变量的困难,又能迅速获得实时短期交通流预测结果。研究结果表明,构建的神经网络模型能够很精确地实时预测城市道路短期交通流。

Study on Forecasting of Urban Short-term Traffic Flow Based on a GRNN Model

Difference data of urban short-term traffic flow is emploied weekly to do prediction target of a generalized regression neural network (GRNN) model. This treatment not only avoids to correctly choosing influencing factors as input variables of a neural network model, but also quickly obtains forecast values of real-time short-term traffic flow. The results show that the GRNN model constructed in this way can precisely forecast urban short-term traffic flow.