Research And Application Of The Forecast Of Urban Regional Roads’ Traffic Situation

As the car ownership of urban residents increases year by year,the traffic pressure has gradually increased,and the traffic congestion has become more and more serious.In order to alleviate traffic congestion,it is an important research direction at present that how to accurately predict the traffic situation of urban regional roads.In order to improve the accuracy of prediction and the reliability of traffic situation assessment,the main work of this thesis could be stated as follows:（1）A model for predicting traffic parameters is proposed.The model uses Conv-LSTM as the extraction module of spatial features of traffic data,and combines the attention mechanism with Bi-LSTM model as the extraction module of temporal features of traffic flow data to obtain deeper temporal features.Compared with the traditional prediction model,the model further improves the prediction accuracy due to better utilization of the spatio-temporal correlation of the data.（2）A method to classify traffic situation levels by using multiple parameters is proposed.In this method,the traffic volume and traffic speed predicted by the previous model and the real-time traffic situation level of the third-party platform are taken as input to calculate their corresponding situation levels respectively.The weighted information method is used to determine the weight coefficient of each situation level,and the final traffic situation level based on the forecasted result is obtained by weighting calculation,to realize the short-term prediction of urban regional roads’ traffic situation.Compared with the single parameter evaluation method,multiple parameter information combined in this method makes the evaluation result more reliable.（3）The urban regional road traffic situation prediction system is designed and implemented,which integrated and embedded the traffic prediction model and the traffic situation assessment algorithm.The server of the system processes,trains and predicts the collected data,after the prediction,the future traffic situation is calculated according to the evaluation algorithm,and the location information of possible congestion is counted.The system visualizes the monitoring points on the map through web pages,showing the changing trend of traffic data and the forecasted results on each point.The managers can make corresponding strategies according to the displayed information,to achieve the purpose of early warning.