In order to overcome the inaccuracy of current research results of traffic flow prediction, this paper proposes a prediction method for traffic flow with small time granularity at intersection based on probability network. This method takes one minute as time granularity, collects traffic data suchas cross-section flow, section traffic flow velocity data, traffic density, road occupancy, section delay and steering ratio by using RFID technology, and analyzes and processes the data. By introducing Bayesian network in probabilistic network and combining K-nearest neighbor method, historical data and predicted traffic flow state are classified to realize the prediction of traffic flow with small time granularity at intersections. The experimental results show that this method has high prediction accuracy and reliability, and is a feasible traffic flow prediction method.

中文翻译：

---

基于概率网络的交叉口小粒度交通流量预测

为了克服当前交通流量预测研究结果的不准确性，提出了一种基于概率网络的交叉口小时间交通流量预测方法。该方法以一分钟为时间粒度，使用RFID技术收集横断面流量，路段交通流速数据，交通密度，道路占用率，路段延误和转向比等交通数据，并对数据进行分析和处理。通过在概率网络中引入贝叶斯网络，并结合K最近邻法，对历史数据和预测的交通流状态进行分类，以实现对交叉口时间粒度小的交通流的预测。实验结果表明，该方法具有较高的预测精度和可靠性，