Vehicle Operation Status Sensing And Risk Assessment At The Entrance And Exit Of The Long Tunnel

With the rapid development of highways,the road network density is increasing,and the public travel demand is growing rapidly.As an important node of expressway,long tunnels bear increasing traffic flow,which leads to traffic congestion and frequent traffic accidents in long tunnels,especially the entrance/exit section of long tunnels is easy to become a section with accumulation and sudden change of operation risk.Therefore,this paper addresses the traffic safety needs of long tunnels,based on the video data of Banlongshan Tunnel of Qinglan Expressway,extracts the vehicle operation status characteristics of tunnel entrances and exits,analyzes and assesses the risk using safety alternative indicators based on risk field theory,and proposes a traffic control strategy for tunnel entrances and exits,which is especially necessary to reduce the traffic risk of long tunnel entrances and exits.Firstly,theoretically,we analyze in depth the operating state of vehicles at the entrances and exits of long tunnels,and analyze the influence of drivers,vehicles,roads and environment on the operating state of vehicles in long tunnels;analyze the risk of traffic operation in long tunnels,and summarize the risk characterization indexes based on space and time,based on distance,and based on acceleration.Secondly,for the video data of Panglong Mountain Tunnel,one of the real-time detection algorithms with high accuracy,YOLOv5,is used,and the algorithm is fine-tuned and trained with self-built datasets to improve the accuracy of the tunnel entrance/exit target recognition;in the target tracking part,the classical multi-objective tracking algorithm Deep SORT is used to achieve the internal tracking of the target,and the spatial of four sign points are obtained using the lane width After that,the focal distance is used as the optimization parameter,and the optimal focal distance value is estimated based on the root mean square of the reprojection error of the marker points to obtain the internal and external parameters of the camera.Then,the data are extracted to derive the velocity characteristics and acceleration characteristics of the vehicles at the entrance and exit of the tunnel.Again,we analyze the similarity between traffic risk and "field",define the concept of "traffic safety field" from the risk field theory,and propose the traffic safety proxy risk-deceleration(RD)based on the risk field theory for tunnel entrances and exits.The results show that RD has higher risk prediction accuracy,timeliness and other advantages.Finally,it introduces the traffic operation risk warning process at the entrances and exits of grown-up tunnels,clusters the values of the safety alternative RD,classifies the traffic operation risk level,and proposes traffic control strategies such as variable information boards at tunnel entrances and exits,vehicle mobile terminal information dissemination,tunnel entrance and exit shade shelters,colorful anti-skid road surfaces at tunnel entrances and exits,and traffic guidance facilities at tunnel entrances and exits from the perspective of engineering practice.This paper extracts the vehicle operation status from the measured data of Panglong Mountain Tunnel by algorithm and evaluates the traffic risk,and then proposes the control strategy for the entrance and exit of the grown-up tunnel,which has significant application value to improve the traffic safety of the tunnel entrance and exit.