Name: Chengming Jin Supervisor: Allison Kealy GNSS-based Positioning Scheme & Application in Safety-critical Systems of Rail Transport
CONTENT 1 Introduction 2 Challenges 3 Solutions
Introduction How Modern Railway Signal Works? 1
Introduction Signalling System: Track Circuit 2
Introduction History of Signalling Systems Diversity of European ATP systems ETRMS/ETCS Cockpit 3
Introduction Train Control Systems: Positioning Scheme Signalman Token one engine in steam Driver A token being offered by a signalman on the Keighley and Worth Valley Railway (from Wikipedia) 4
Introduction Train Control Systems: Positioning Scheme speed, location speed, location Accumulated error Calibration 5
Introduction Train control systems: Balise 2.5 km More in station Expensive difficult to maintain 6
Introduction Signalling System: Fixed Block & Moving Block A B C D E Mainstream Signalling System Signalling System in the Future *THE DEVELOPMENT AND PRINCIPLES OF UK SIGNALLING 7
Introduction Next-Generation Train Control System No track circuit Ability to determine train integrity on board No or less balise Trains find their position themselves Full radio-based train spacing Moving Block 8
GNSS Introduction Location info. Time info. Short messages(bds) with high accuracy in all weather conditions anywhere on or near the Earth Cost-efficient available 24/7/365 9
Introduction GNSS-based train control systems EC and European Railway Agency (ERA) launched many projects to promote the progress of GNSSbased railway applications 2014 Shift2Rail 2012 3InSat 2010 2005 2004 2001 GPS-based PTC (Positive Train Control) had been equipped in the US and China (Qinghai-Tibet Line) GLONASS SDCM WAAS SATLOC GRailⅡ ATLAS 400, an European GPSbased train control system GALILEO EGNOS GPS WAAS BDS MSAS QZSS GAGAN GRail GEORAIL ECORAIL Locoprol InteGRail Gaderos Non-safety applications RUNE Europe GNSS-based railway applications projects GNSS is a worldwide, cost-efficient approach to locate the target, which makes GNSS-based positioning become one of the most promising positioning solutions for the next-generation train control system. 10
CONTENT 1 Introduction 2 Challenges 3 Solutions
Challenges GNSS was refused by railway: Policy issue ETCS (European Train Control System) CTCS (Chinese Train Control System) have been standardized in the last two decades. Balise and STM (Specific Transmission Module) are necessary in ETCS-1,2. *SUBSET-026 ERTMS/ETCS System Requirements ERTMS/ETCS reference architecture* 11 Specification issue:3.0.0
Challenges GNSS was refused by railway: Accuracy Accuracy of distances measured on-board: ± (5m+ 5% S) Accuracy of distinguishing parallel tracks: 1.5m 34% Masked sky & multipath Masked sky & multipath *SUBSET-026 ERTMS/ETCS System Requirements Specification issue:3.0.0 3-5m ERTMS/ETCS reference architecture* 12
Challenges GNSS was refused by railway: There is a wall!! RAMS Railway applications must meet the requirements for Reliability, Availability, Maintainability, and Safety GNSS performance parameters, which are derived from aviation, are SIS Availability, Integrity, Continuity Safety: According to CCS TSI 2012/88/EU, for the hazard `exceeding speed and/or distance limits advised to ERTMS/ETCS' the tolerable rate (THR) is 10-9 /h for random failure, for on-board ERTMS/ETCS and for track-side, and positioning unit is just one of many subsystems. *Debiao Lu, GNSS for Train Localisation Performance Evaluation and Verification, Dissertation, 2014. Relation between GNSS and Railway Signalling QoS Properties* <10-9? 13
CONTENT 1 Introduction 2 Challenges 3 Solutions
Solutions Solutions: Pseudorange-based GNSS Carrier-phase-based GNSS SPS DGNSS RTK PPP Potentially High Accuracy High Availability High Safety SPS: Standard Positioning Service DGNSS: Differential GNSS RTK: Real Time Kinematic PPP: Precise Point Positioning 14
Solutions Solutions: Why PPP? Differential solutions ϕ = ρ + ε i i i = ϕ ρ ε k j j j PPP solutions ϕ = ρ + ε k k Station movements that result from geophysical phenomena such as tectonic plate motion, Earth tides and ocean loading enter the PPP solution in full, as do observation errors resulting from the troposphere and ionosphere. Relevant satellite specific errors are satellite clocks, satellite antenna phase center offset, group delay differential, relativity and satellite antenna phase wind-up error. Receiver specific errors are receiver antenna phase center offset and receiver antenna phase wind-up. In comparison with DGNSS, PPP has higher accuracy(centimetre to decimetre level*) Compared with RTK, PPP requires fewer reference stations globally distributed. PPP gives a highly redundant and robust position solution * M.D. Laínez Samper et al, Multisystem real time precise-point-positioning, Coordinates, Volume VII, Issue 2, February 2011 15
Solutions Solutions: PPP-based multi-sensor fusion ba, b b n g δv, δ p IMU ODO PPP b f ib b ω ib n v O δψ nb Navigation Processor + δ K p + p n I n G eb, v, v n I n G b EKF Integrity Monitoring Corrected Position, Velocity, Attitude Integrity Information IMU: Inertial Measurement Unit EKF: Extended Kalman Filter ODO: odometer 16
Solutions Scenarios GNSS/PPP IMU ODO Scenario 1 available not converged available available Scenario 2 available converged available available Scenario 3 unavailable available available 16
Solutions On-site test 2.8956 6GNSS/INS 10 Kalman Filter compares with GNSS position 2.8954 2.8952 2.895 ECEF y axis (unit:m) 2.8948 2.8946 2.8944 2.8942 2.894 2.8938 Kalman Filter Solution GNSS Position Info. 2.8936-4.1299-4.1298-4.1297-4.1296-4.1295-4.1294-4.1293-4.1292-4.1291 ECEF x axis (unit:m) 10 6 Trajectory of On-site Test Position Error 17
Solutions Simulation test 4.3882 10 6 INS Navi. Solution Compares with Real Trajectory 4.38815 INS Navi. Solution Real Trajectory 4.3881 ECEF y axis (unit:m) 4.38805 4.388 4.38795 4.3879 4.38785-2.1723-2.1722-2.1721-2.172-2.1719 ECEF x axis (unit:m) 10 6 SPIRENT Simulator Navigation Trajectory 17
Solutions Solutions: PPP-based multi-sensor fusion 1 INS/ODO Kalman Filter Position Error 15 INS Navigation Error 0.5 0 10 North error East error Down error INS/ODO Navigtaion error(unit:m) -0.5-1 -1.5-2 North Error Navigation Error(unit:m) 5 0-5 -2.5 East Error Down Error -3 0 1000 2000 3000 4000 5000 6000 Num of Navigation Solution -10 0 1000 2000 3000 4000 5000 6000 Num of Navigation Solution INS/ODO Kalman Filter Navigation Error *GNSS position error ~ N(0,1); GNSS velocity error ~ N(0,0.01); ODO velocity error ~ N(0,0.01) INS Navigation Error 17
Solutions Solutions: PPP-based multi-sensor fusion 1.2 GNSS/INS Kalman Filter Position Error 0.6 GNSS/INS/ODO Kalman Filter position error 0.8 1 North error East error Down error 0.4 North error East error Down error GNSS/INS Kalman Filter Navigation Error(unit:m) 0.6 0.4 0.2 0-0.2-0.4-0.6-0.8 0 1000 2000 3000 4000 5000 6000 Num of Navigation Solution (Sample frequency: 100 Hz) GNSS/INS/ODO Navigation Error(unit:m) 0.2 0-0.2-0.4-0.6-0.8 0 1000 2000 3000 4000 5000 6000 Num of Navigation Solution (Sample frequency: 100 Hz) GNSS/INS Kalman Filter Navigation Error GNSS/INS/ODO Kalman Filter Navigation Error *GNSS position error ~ N(0,1); GNSS velocity error ~ N(0,0.01); ODO velocity error ~ N(0,0.01) 18
Solutions Quality Control: Detection, Identification and Adaptation(DIA) Based on consistency check of innovations *Quality control and integrity, Delft school t k = vq v T 1 k v k m k k 19
Solutions Quality Control: Detection, Identification and Adaptation(DIA) Bias (unit: m/s) Detected Missed Detection Success Rate 0 1000 20 98.04% 0.1 134 886 13.13% 0.5 1000 20 98.04% 1 1020 0 100% Bias (unit: degree) Detected Missed Detection Success Rate 0 1000 20 98.04% 0.0000001 77 943 7.54% 0.000001 1000 20 98.04% 0.1 1000 20 98.04% 0.5 1000 20 98.04% 10 1020 0 100% 20
Solutions Threshold THR <= 10-9 /h 0.12 0.1 Chi-square Noncentral Chi-square Threshold 0.08 0.06 0.04 0.02 0 0 5 10 15 20 25 30 35 40 45 50 20
Solutions Further Research DIA global test Track maps aided PPP integrity monitoring scheme 21
Lhasa Thank You