ID No: 459 Accuracy Performance Test Methodology for Satellite Locators on Board of Trains Developments and results from the EU Project APOLO Author: Dipl. Ing. G.Barbu, Project Manager European Rail Research Institute (ERRI) Arthur van Schendelsraat 754, NL-3511 MK Utrecht Tel: +31 30 232 42 19 Fax: +31 30 236 89 14 e-mail: gbarbu@erri.nl The Problem Train autonomous localisation using satellite navigation technology in combination with inertial navigation and on-board odometry is an important and challenging alternative to the ground based train positioning systems (track circuits, axle counters, loops, balises). The advanced location positioning for trains developed within the EU project APOLO has applied the intelligent fusion of on-board sensors (satellite receiver, gyrometer, odometer) to create the train autonomous localisation system capable to function in all points of a railway route, thus also there where the satellites are not visible. The APOLO targeted positioning accuracy performance was: - In the range of 2 = 20 m at 95 % confidence when using receivers in the standard service of the GPS (USA NAVSTAR Global Positioning System). - In the range of 2 = 4 m at 95% confidence when implementing GPS receivers with differential corrections augmentation. Determination of localisation accuracy performance achievable in real conditions (on board of running trains) is essential: - For verification and validation of the accuracy performance of the realised devices, - For research and development, to study the function of different locator s components in a wide range of regimes and utilisation scenarios Classical methods to compare the real position on track obtained through marking in reference points with the position determined by the locator was not satisfactory because: - Marking has poor accuracy and can not provide repetitive results because of the inherent marking errors - The number of affordable reference points is limited, statistic base is poor - Could not use trains in commercial service (speed too high) The requirements for an accuracy determination method and analysis tool were: - To create reference points at each determined locator s fix, e.g. once a second, with the accuracy of one magnitude order better than the expected best accuracy of the tested device - To enable automatic registration of reference and locator s test data with market standard equipment - To enable tests on board of running trains, in normal service - To enable creation and implementation of test scenarios that should correspond to all real conditions needed for analysis, performance estimation and validation. - To be independent of the train speed The principle solution The solution selected for the test methodology of APOLO prototypes was based on the original principle: 1
- To create and to automatically register on board of the train a high precision reference database that calculates each second a reference point. The reference is based on data of on-board independent receiver augmented with differential corrections using Long Range Kinematic (LRK) post-processing procedure or Real Time Kinematic (RTK) technique. - To synchronise and to align this database with the data registered from the APOLO locator - The reference and the APOLO receiver use the same antenna or, when different antennas are mounted, the offset is measured and controlled in the evaluation process. The figure 1 Test and evaluation method illustrates the measuring and evaluation principle. After post-processing of on-board and of fix point raw reference data in a LRK procedure, a precision reference data base is obtained. The typical accuracy of the LRK determinations is in decimetric precision range. ***<01. Test and evaluation method>*** The position determinations of the reference and of the test data are triggered by the UTC (Universal Time Coordinated) clock marks of the satellite receivers. So, in the same second, a reference point and a point determined by APOLO are registered. A dedicated software formally validates each registration and aligns in synchronism the reference and the APOLO data. Another software application processes the synchronised data in order to evaluate differences to the reference and to calculate statistic parameters. The Application The figure 2 Integration in trains for test shows the equipment mounted in train for tests in Spain, on the line Madrid Escorial. The reference is obtained by post-procession of data registered on a PCMCIA memory card of the onboard reference receiver (DSNP SCORPIO 6002 SK receiver). The data of the APOLO prototype are registered on a standard portable computer. The APOLO and SCORPIO receiver have each one separate antennas, mounted on the roof of the train. The mounting offset is measured and evaluated within data procession procedure. ***<02.Integration on trains for test>*** The trains on which the on-board equipment is mounted are the passenger trains, in normal service, on the test line in Spain (figure 3: Test train in Spain). ***<03.Test train in Spain>*** The location determinations of the fix reference point necessary for computing of differential corrections in the LRK post-processing procedure are registered each second, during the test duration, on the PCMCIA memory card of a second SCORPIO receiver mounted in one railway stations placed about at the middle of the test area. Figure 4 Fix point principle layout shows the equipment layout for registration of fix point raw data. Figure 5 Antenna and receiver in Vilalaba shows the fix point receiver s antenna and the fix point position location equipment, a second SCORPIO receiver as mounted in the station Vilalba on the test line. ***<04.Fix point principle layout>*** ***<05.Antenna and receiver in Vilalba>*** One of the major advantages of the application in Spain was that no broadcasting of the differential corrections from the fix point station was necessary, hence no licence for frequency was requested. The tests on the lines of the Czech Railways have implemented the same principle. The local conditions enabled to implement more complex equipment that could use broadcasting of augmentation signals (differential corrections) from the fix point reference station. CD installed Ashtech s 2
DGPS/DGLONASS L1 24 channel reference receiver in Pardubice railway station. The equipment generates and broadcasts local RTCM-104 differential corrections needed for the Real Time Kinematic reference for tests. The CSMA (Carrier Sense Multiple Access) 150 MHz/ 19kbps radio network was installed along CD trial tracks in Pardubice area in order to perform APOLO tests in LADGPS mode and generate RTK reference trajectory. The figure 6 "Test equipment layout on CD line" shows the principle schematic of the test equipment installed in the Pardubice area and on the test railway vehicle. ***<07.On-board equipment on CD test rail vehicle>*** ***<08.Test locomotive>*** A detailed presentation of the on-board equipment for tests on the CD line is presented in figure 7 "Onboard equipment on CD test rail vehicle". The test vehicle is a locomotive in normal traction service of the Czech railways (Figure 8: "Test locomotive") Results Accuracy performance using the GPS standard service As available from 1 st May 2000 when the selective code was suppressed the GPS ranging offers without differential corrections an accuracy performance which is 2 =20m (95% confidence). In a real railway environment with hybridisation -odometer and gyrometer using the APOLO technology the accuracy is in the range of 2 = 4,8 m at 95% confidence. The determined accuracy corresponds to typical condition of a railway route, with alternating zones and portions where the SIS is not available. The APOLO faults derived from inherent experimentation problems and adjustments at prototype level have been not considered and were filtered out. In such a condition, the determined accuracy corresponds to the average performance of an expected industrial equipment that would implement the corrections and improvements as identified during the tests and verifications of APOLO prototype. Determination of improvements and corrections were substantially supported by the synchronised databases that enabled a detailed analysis of components' and fusion software. These improvements, as suggested to the Thales-Navigation, are: a. Improvement of gyrometer quality, e.g. elimination of gyrometer's temperature drift effects b. Improvement of hybridisation strategy (Kalman filter, reciprocal calibration of sensors) when taking into account the train's (rail vehicle's) characteristic dynamics (maximum acceleration, deceleration, angular velocity in curves depending on linear velocity, etc.) c. Improvement of hybridisation strategy when considering the train's trip characteristic sequences (start after -long- stay, hold of cap after start, evaluation of GPS fix utilisation in relation with the fix quality and the trip sequence). Accuracy performance with GPS + augmentation The augmentation is available through EGNOS (European Global Navigation Overlaid System) WAAS (Wide Area Augmentation System) or through LAAS (Local Area Augmentation sent by differential correction broadcast station). The route is characterised by a typical railway environment. Positioning uses APOLO technology for hybridisation. The determined accuracy is in the range of 2 ~ 1m at 95% confidence. This accuracy corresponds to the average achievable performance in conditions of industrial equipment hat implements the suggested improvements. 3
Conclusion The presented methodology for test of accuracy performance of autonomous train location systems using the satellite navigation equipment corresponds to the requirements. The method has also the advantages: - To use simple and available test equipment, based on standard devices - Does not mandatory need broadcasting equipment for generating a Local Area Augmentation Signal - Enables a high degree of automation of tests - Generates reliable database with large "statistic population" hence, enables application of consistent statistic processing to derive performance estimates with high confidence degree - Enables detailed analysis of test data synchronised with reference data, thus intimate behaviour of sensors and of fusion software can be evaluated for research and development purposes - Enables generation of test scenarios in laboratory condition, when combining synchronised sequences registered in field, to complement the tests in field - Through availability of detailed registration of values of components' outputs synchronised with reference data, provides further research support to analyse and verify different "sensors' fusion strategies" and optimisation of these to the train environment conditions. Figures TEST FLOW & EVALUATION TEST TEST DOCUMENTATION / / TEST TEST PROTOCOL - - EVENTS APOLO APOLO TEST TEST FILE FILE - - N.M.E.A. N.M.E.A. PROTOCOL PROTOCOL ON-BOARD ON-BOARD FIX FIX POINT POINT RAW RAW RAW RAW (SCORPIO-PCMIA) (SCORPIO-PCMIA) (SCORPIO-PCMIA) (SCORPIO-PCMIA) POST PROCESSING REFERENCE REFERENCE LRK LRK PRECISION PRECISION REFERENCE REFERENCE ANALYSIS AND EVALUATION ALIGNEMENT ALIGNEMENT SYNCHRONISATION SYNCHRONISATION BUILDING BUILDING OF OF SCENARIOS SCENARIOS COMPUTATION COMPUTATION OF OF ERORS ERORS EVALUATION EVALUATION STATISITC STATISITC ANALYSIS ANALYSIS CORRELATION CORRELATION INTERPRETATION INTERPRETATION <01. Test and evaluation method> 4
<02.Integration on trains for test> <03.Test train in Spain> 5
<04.Fix point principle layout> <05.Antenna and Receiver in Vilalba> 6
<06.Test equipment layout on CD line> <07.On-board equipment on the CD test rail vehicle> 7
<08.Test locomotive> 8