A deep-learning-based prediction model for real-time traffic congestion in urban areas

At present, the prediction information of burst traffic flow by various control models is incomplete and inaccurate, which hinders the implementation of fast and effective traffic control measures. A traffic flow information prediction method based on deep theoretical learning is proposed for urban traffic congestion. Through the correlation analysis of the selected parameters, the information flow is analyzed. After that, the classification pre-training strategy deep learning model is proposed, which can quickly identify the traffic information flow caused by congestion while recognizing congestion, so as to carry out effective control. Finally, the traffic information flow information under sudden congestion of Jiuquan city main road is forecasted and analyzed. The results show that the deep learning prediction model proposed in this paper can predict the traffic information flow more accurately than the existing models, and has better practicability.