无检测器交叉口交通流量预测的灰色神经网络模型

为解决一般预测方法要求原始数据量较大,而无检测器交叉口获得的交通流量数据又非常有限的矛盾,提出了一种基于灰色神经网络的无检测器交叉口交通流量预测方法。通过选择不同长度的历史数据构建不同的灰色预测模型,对于不同灰色预测模型得到的预测结果再使用神经网络进行组合,该方法综合了GM预测所需原始数据少、方法简单,而神经网络具有非线性拟合能力的特点。以107国道新市站2002年观测的交通流量作为原始数据,采用灰色神经网络进行时交通流量预测,结果表明了该方法是有效可行的。

A Gray Neural Network Model for Traffic Flow Prediction at Non-detector Intersections

Conventional prediction methods require large number of samples while traffic flow data at non-detector intersection is very limited. To solver this problem a new gray neural network model to predict the traffic flow at non-detector intersection is proposed. After many gray forecasting models are established using different length traffic flow history data, the different results of gray forecasting were combined using neural network. The presented methods synthesizes the advantages of GM forecasting method, which is simple and needs less original data, and of neural network method which possesses the characteristics of nonlinear fitting. The traffic flow pre hour is predicted using the history data of XingShi Station of National highway NO107 in 2002, and the results show the presented method is feasible and effective.