The prediction of traffic flow can be greatly useful for the work of traffic management departments and the travel planning of drivers. How to make accurate and efficient traffic flow prediction is a very important issue.Traditional traffic flow prediction data sources are usually vehicle speed and driving trajectory which are obtained by arranging traffic sensors on the highway at regular intervals. Although the existing method applied to suburban areas and highways have achieved good results, it can not be used to make the predictions on dense and complicated urban roads for the inconvenience of large-scale deployment of sensors to obtain the required data. This paper proposed a forecasting method by using traffic flow data of urban road checkpoints. We first got the characteristics of cyclic changes in traffic flow by analyzing existing traffic data.Then we extracted corresponding features based on these cyclic changes. Finally we trained traffic flow prediction models suitable for urban checkpoints based on these features. A large number of experiments have been carried out according to real traffic data sets, and the results show that our traffic flow prediction model has a good prediction effect. With RMSE (15.3)and MAPE(7.3) of the predicted values, the accuracy can reach 92.7%.