基于多参数状态时间序列的交通状态预测方法

利用多个参数描述交通状态时，交通流数据表现为多维空间数据。提出了将属于每个状态的多维空间数据转换为一维时间序列的方法，对于此状态时间序列采用BP神经网络进行了下1个时段的交通状态预测。实验结果表明，多参数状态时间序列比单个参数时间序列能更准确地描述交通流状态变化过程，且算法简单，具有较强的预测实时性。

Traffic Prediction Method Based on M ulti-parameter Status Time Series

Prediction of the traffic status plays a very important role in management and guidance of urban traffic. In general, traffic status cannot be fully described by a single parameter. Therefore, the traffic status should be described through a set of parameters, such as flow, speed and density. In the proposed method, the multi-dimensional traffic data are first transformed into one-dimensional time series, and then the traffic status for the next time interval is predicted by using BP neural network. The experimental results show that the multi parameter status time series describes the changes of traffic status more accurately than one parameter time series. The algorithm is not only simple but also practical for predicting traffic status in real time and thus it can be used in future traffic guidance system.