With the increasing demand of urban traffic management, intelligent transportation system (ITS) has gradually become an important means to solve the problem of urban traffic congestion. The combination of ETC gantries and Internet of Things (IoT) monitoring technology provides data support for accurate prediction of realtime traffic flow and congestion warning. In this paper, an attention-weighted SG-LSTM-based traffic flow prediction model is proposed and applied to the traffic flow prediction and congestion warning of ETC gantry data in M city A area. Through data preprocessing, the introduction of Savitzky-Golay filter, and the training of LSTM neural network, this model can effectively improve the accuracy of traffic flow prediction. The experimental results show that the model has higher prediction accuracy compared to the traditional LSTM, CNN, GCN and other classical methods. Specifically, the model reduces 27.43% and 43.07% in RMSE and MAE metrics, respectively. While the accuracy of traffic flow prediction is improved, this study also designs a congestion warning model based on support vector machine, which predicts the traffic flow and speed through real-time data and accurately warns the traffic congestion condition, which verifies the effectiveness and high accuracy of this model.
