Research On Real-time Vehicle Location System In Tunnel Based On DL-TDOA

With the rapid development of information processing technology,control technology,transportation conditions and other technologies,as well as the demand for new technologies in national construction,the construction of smart roads has been expedited.The so-called smart highway takes safety,green,convenience and efficiency as the construction goal,forms an open and shared basic vehicle platform,including people vehicle mode,vehicle-vehicle mode,etc.to provide it with safe and reliable network traffic and open and mutual aid communication pipeline services,so that it can provide real-time response emergency services in case of emergency events,and provide accurate and independent travel activities for travelers.The tunnel is generally used to shorten the transportation distance through mountainous areas.Its natural nature leads to its shielding effect on radio signals.Beidou satellite positioning cannot reach the inside of the tunnel,resulting in the loss of relevant vehicle information and data in the tunnel.In order to realize the real-time and accurate positioning of vehicles in the tunnel,the following work has been completed in this paper:1)Selection of vehicle positioning method in the tunnel based on the principle of downlink arrival time difference(DL-TDOA): after comparing the indoor positioning methods at home and abroad,the advantages of large capacity,low power consumption and strong stability based on ultra wideband(UWB)are selected for improvement.Based on the comprehensive consideration of the possible error factors caused by positioning in the tunnel,TDOA positioning method is selected.2)In view of the complex environment in the tunnel and the signal receiving error caused by visible distance obstacles,in the TDOA based positioning algorithm,the performance of the traditional positioning Chan algorithm and Taylor algorithm are compared,and a Chan Taylor hybrid algorithm based on the adaptive Kalman filter algorithm is proposed to improve the positioning accuracy.3)On the basis of adaptive Kalman filter for vehicle positioning,RNN—LSTM algorithm is further adopted to predict the distance within 5 seconds after the vehicle,and route fitting is performed with the above algorithm to further improve the positioning accuracy of TDOA vehicles on the basis of deep learning algorithm.Vehicles in the tunnel may have emergency stop,turn,avoid danger and other actions,which leads to the positioning and prediction algorithm can not only rely on the satellite data outside the tunnel,and the error is relatively large.The algorithm further improves the accuracy of vehicle location.4)In order to verify the vehicle accurate positioning algorithm proposed in this paper,experiments will be carried out in the actual tunnel scene.In view of the fact that the wireless electromagnetic wave and other characteristics cannot be obtained in the tunnel,and the navigation data such as Beidou can not be used as a reference frame to judge the DL-TDOA real-time positioning accuracy,it is introduced to set up anchor nodes and bridge nodes in the open highway sections to verify the accuracy of the algorithm.Then the coordinate data actually collected by the vehicle in the tunnel is used as the data set for algorithm training.The DL-TDOA positioning algorithm was compared in the open field of the expressway.After comparing the same positioning algorithm,the RNN-LSTM algorithm with the highest positioning accuracy was applied to the static test,30km/h,60km/h and 80km/h dynamic test in the tunnel.The average error was less than one meter,meeting the RTLS requirements.5)A multi vision and multi data application system is designed for the practical application problems in the tunnel.The system is divided into: sensor device layer,network device layer,network management platform,application data platform,data analysis engine and large screen application display platform.After the system is completed,the real-time vehicle positioning information based on DL-TDOA is uploaded to the network management platform,and then the vehicle information is managed through the application system,which has strong scalability.