Research On The Train Positioning Method For Subway Based On Visible Light Communication And Binocular Vision

As the preferred means of transportation to relieve the urban traffic pressure,subway has a high demand for the safety of train operation.Accurate information of train position is of great significance to the safe operation of trains,and it is the necessary information to realize the interval of train control in the Communications Based Train Control(CBTC)systems.Although the existing train positioning technology can meet the needs of train operation control,it suffers from discontinuous positioning,poor anti-interference ability and shortage of spectrum resources.In recent years,Visible Light Communication(VLC)stands out among many emerging communication technologies,and has broad application space in indoor and outdoor communication and positioning,traffic signal management,Internet of things and other fields.VLC technology can integrate lighting and communication,and s avoid spectrum shortage and wireless interference.Moreover,the positioning technology based on VLC can realize centimeter-level positioning,which is expected to be used for subway train positioning in tunnel environment.In this dissertation,a new train location method based on VLC and binocular vision is proposed by using the existing lighting equipment in the subway tunnel to achieve high precision train location,which is a supplement to the existing train location technology.First,the train receives the corresponding unique identifier in real time according to the light signal from the LED light source mounted on the tunnel wall.Gray gravity center method and least square method were used to extract the spot center coordinates of LED light source image,and the train static positioning was realized by binocular vision principle.Secondly,Wiener filter and Inertial Measurement Unit(IMU)are used to compensate the impact of motion blur of the train image and mechanical vibration of the receiving end on the dynamic positioning error.Then,the train running process is divided into moving section and zero-speed section.In the moving section,the unscented Kalman filter algorithm is used to integrate the dynamic location results of the train and the train motion model to obtain the predicted state estimation results,and the optimal state estimation results are obtained.Finally,the zero velocity update algorithm is introduced to detect whether the train meets the zero-velocity condition.The continuous zero-velocity point is taken as the zero-velocity section,and the unscented Kalman filter is applied to update the zero velocity The cumulative error divergence of IMU is overcome,and the influence of cumulative error of IMU on the continuous high precision positioning of the proposed method is eliminated.Combining the line data and equipment information of Chengdu Metro Line 1,a subway train positioning model based on VLC and binocular vision was established to verify the feasibility and effectiveness of the proposed method.The results show that the maximum static positioning error of this method is 29.93 cm,and when the train runs at different speeds,the maximum positioning error is 36.11 cm,and the maximum positioning time is 51.32 ms.After optimization,the maximum positioning error of the train is 19.57 cm under uniform speed running in a straight line,and 18.48 cm and 18.04 cm under uniform acceleration and deceleration respectively.The average positioning error of VLC after train stop is 6.05 cm,and the maximum positioning error of train is less than ±5 m ～ ±10 m required by IEEE1474.1 standard.The proposed method can meet the requirements of train positioning in tunnel environment,and can provide a certain reference for train positioning in CBTC systems with vehicle-to-vehicle communication.