GNSS/INS/DTM Integration For Dependable Train Localization

With the development of train control system,train positioning techniques are transforming to the“on-board centralized”continuous positioning method.Aiming at the problems that Global Navigation Satellite System（GNSS）positioning is affected by the environment,train positioning dependability is degraded,this thesis proposes dependable train positioning method based on the integration of GNSS/Inertial Navigation System（INS）/Digital Track Map（DTM）,and the research is carried out from three aspects:positioning accuracy,track occupancy identification,and train positioning dependability,so as to realize the dependable estimation of train positioning status.The main research contents of this thesis are as follows:（1）The GNSS/INS/DTM based multi-sensor fusion train positioning algorithm is studied to enhance train positioning accuracy.The weighted extended Kalman filtering（WEKF）based GNSS/INS tightly coupled positioning model is constructed.Then,the Zero-velocity Update based positioning accuracy optimization method is proposed aiming at solving the problem that the positioning error diverges when the train is in stationary state.Finally,the INS/DTM coupled positioning algorithm is studied to achieve full-scene trains positioning aiming at sloving the problem that GNSS fails in tunnels/stations and other GNSS-denied scenarios.（2）The GNSS/INS/DTM based train track occupancy identification method is studied to achieve accurate train track occupancy identification in the GNSS positioning challenging scenarios.Facing the train track occupancy difficulties in switch and parallel-track scenarios,this thesis proposes the modeling method that correlates GNSS/INS/DTM observations with track occupancy identification.Finally applying the recursive Bayesian estimation for train autonomous track occupancy identification and confidence estimation to enhancing the confidence and reducing distance consumption.（3）The GNSS/INS/DTM based train positioning dependable monitoring method is studied.Fault detection and exclusion are carried out to enhance train positioning dependability.The INS/DTM aided GNSS train positioning integrity monitoring method and the Kalman filtering innovation/residual based horizontal uncertainty level estimation algorithm are studied.Train positioning dependability evaluation method and horizontal dependability index（HDI）are proposed to realize the online evaluation of train positioning dependability and the offline statistical analysis of failure risk.This thesis uses the Beijing-Shenyang high-speed railway experiment data to verify the algorithms.The verification results show that the GNSS/INS/DTM based multi-sensor fusion train positioning algorithm can effectively enhance the train positioning accuracy,and the positioning accuracy can reach 2m when the train runs in the blocks or is in stationary state.The maximum positioning error is 6.36m under the2.2km long tunnel and the positioning accuracy is enhanced by 53.9%compared with INS update only.The train track occupancy identification algorithm based on recursive Bayesian estimation is verified,and the results show that the algorithm can effectively improve the confidence of train track identification,and the maximum distance consumption is 23.36m,which is reduced by 80.6%compared with multi-hypothesis map-matching based train track occupancy identification method.Furthermore,the GNSS/INS/DTM based train positioning dependable monitoring method is verified.The results show that the method can effectively enhance fault detection rate and train positioning dependability.Within 200 epochs of fault injection,the number of epochs with train positioning dependability less than 0.90 is reduced from 18 to 0.Finally,comprehensive experimental verification is carried out.In the 18276 epochs of field collected data,the horizontal positioning error（95%confidence probability）is 2.48m,and the track occupancy identification distance consumption is 3.66m.The number of epochs with train positioning dependability greater than 0.90 accounted for 99.74%.Failure rate of entire experimental process is 5.07×10^-3/h.The verification results show that the dependable train positioning method based on the integration of GNSS/INS/DTM realizes the multi-sensor fusion train positioning,accurate track occupancy identification,and real-time estimation of train positioning dependability.Figure 68,table 31,reference 65.