Research On Early Warning Of Urban Road Traffic Conditions Based On Improved CUSUM Control Chart

Traffic is an important link between cities and has a decisive impact on the development of cities.With the rapid development of economy and urbanization,the following traffic congestion and road traffic safety problems are becoming increasingly prominent.If traffic events on the road cannot be handled in time,It is easy to pose a greater threat to people’s property losses and casualties.The traditional urban road traffic event detection algorithms have various problems,such as low detection performance,long detection time and poor applicability.Therefore,how to quickly and effectively screen traffic events is a very important problem in the field of urban traffic management and control,.In the field of intelligent transportation,CUSUM control charts are widely used in traffic event detection by a large number of scholars at home and abroad.The CUSUM control charts used here are univariate or composite of multiple cusums,and the correlation between variables is not used.This paper establishes an urban road traffic event detection model based on the improved CUSUM control chart to early warning the urban traffic situation.The model has the excellent performance of monitoring multiple variables and adaptive adjustment at the same time,The realization of real-time monitoring of road conditions and abnormal state alarm has theoretical value and reference significance for urban traffic planning and management and the progress of intelligent transportation system.Firstly,this paper introduces the definition and classification of traffic events,expounds some classical traffic event detection algorithms and the advantages and disadvantages of these algorithms,and introduces the theoretical knowledge of traffic flow parameters and traffic flow propagation model that can reflect the road conditions.Because CUSUM control chart has excellent performance in monitoring small drift,a traffic event detection model based on CUSUM control chart is selected,By analyzing the vehicle flow wave transmission process after traffic events,the variation trend of traffic flow parameters in the upstream and downstream of the road section is studied,and the detection parameters of the model are determined in combination with the correlation between density and speed in the green hilch model.Secondly,considering the possible diversity and uncertainty of traffic events,the AMCUSUM control chart which can monitor multiple variables and adaptively adjust the current data drift is finally selected to establish the traffic event detection model.The control chart can spontaneously adjust the parameter values with the size of the sample drift in the monitoring process,design the AMCUSUM control chart and determine the relevant parameters,A road traffic incident detection model based on AMCUSUM control chart is established.Finally,according to several factors affecting the expansion degree of traffic event congestion analyzed by traffic flow propagation theory,the simulation experiment is designed through VISSIM software,and the AMCUSUM control chart and the classical detection methods based on detector data,namely California algorithm and SND algorithm,are used to detect traffic events in the simulation experiment.Combined with the evaluation indexes of traffic detection model,According to the simulation results,the factor analysis and comparative analysis are carried out respectively.The experiment shows that the traffic event detection method based on AMCUSUM control chart has stable and excellent performance in various traffic event states.Finally,the model is applied to the actual road traffic event detection,and it is found that the detection model also has good detection performance.