Optical-Inertial System for Railway Track Diagnostics E. D. Bokhman 2, A. M. Boronachin 2, Yu. V. Filatov 2, D. Yu. Larionov 2, L. N. Podgornaya 2, R. V. Shalymov 2, G. N. Zuzev 1 1 ZG Optique SA Fin-de-Praz 24, CH-2024 St-Aubin, Switzerland and 2 St. Petersburg State Electrotechnical University Department of Laser Measurement and Navigation Systems (LINS), Pr. Popova str., 5 197376 St. Petersburg RUSSIA Inertial Sensors and Systems 2014 Karlsruhe, Germany 978-1-4799-4663-1/14/$31.00 2014 IEEE P16
Abstract The paper presents the results of development of the Optical-Inertial System for Railway Track Diagnostics. It is demonstrated that in order to implement the solution at a speed of up to 430 kmph (used for example in South Korean high-speed train HEMU-430X, standing for High-Speed Electric Multiple Unit 430 km/h experimental) while satisfying the accuracy of 0.1 0.5 mm during measurement of longitudinal level, cross level, twist, curvature, rail profile, etc., it is needed to combine the optical scanners of the inner profile of the rail line with the strapdown inertial navigation system (SINS) in a single block. Supplying of odometer and Global navigation satellite system receiver (GNSS) into the system structure allows to determine measurement point position. Thanks to our a priori knowledge of the semipermanent nature of the railway track, and also to the fusion of the odometer data and satellite navigation system reception equipment data, it is possible to use fiber-optic gyros as the sensitive units of the SINS (both open-loop and closed-loop configurations of FOG can be used). The distinctive feature of the system s algorithm is that it solves both the navigation/orientation task (i.e. it fuses odometer data, satellite navigation system data and inertial navigation system data), and the task of measuring the inner surface profile of the rail line. The use of a sole odometer to localize the found rail flaws does not provide satisfactory results because of its errors. Integration of the odometer, SINS and GNSS receiver data offers highly accurate referencing of diagnostic results to the traversed track coordinate. Odometer readings are updated using the navigation system data. The system provides measuring of the track geometry and accurate localization of the measurement point using the geographical coordinates (latitude and longitude) and orientation parameters (roll, pitch and course angle). The possibility of using SINS based on fiber-optic gyros (FOG) for railway applications is considered in the article. Some practical results are given. 1. Introduction Ensuring a high level of safety and comfort of rail transport requires regular precise track diagnostics through a large number of parameters. The following parameters are monitored on the railways of the world: longitudinal level, cross level, twist, curvature, rail profile, etc (demanded for example by [1]). One can recall such world leaders in the field of designing specialized track geometry cars as OMWE (Germany) [2] and Burlington Northern Santa Fe (USA) [3]. In the field of self-
propelled cars those are EM SAT of Plasser & Theurer (Austria) production [4], Matisa (Switzerland) [5] and MM Roger 1000 of MerMec (Italy) production [6], while in the field of systems on truck and off-road car basis there are the TrackSTAR of Holland (USA) [7] and Track-Inspector of Nordo (USA) [8] productis respectively. One has to mention the significant European advance in the high-speed railway trains development by Network Rail (Great Britain) [9] and the French national logistic enterprise (ELOG) IRIS 320 [10,11], which implement all the world s experience in this field. There is also an ongoing development of track measuring systems on the base of plain passenger cars in order to provide rail diagnostics on day-to-day basis by means of regularly cruising trains [9-15]. Such approach is widely used on Russian railroads. PIK PRO- GRESS (Moscow) [16], RPC INFOTRANS (Samara) [17] and TVEMA (Moscow) [18] succeeded in production of unified means, which can work at train speed of up to 160 km/h and provide measurement of railroad track geometry (with the use of ground-penetrating radars and rail flaw detectors), of contact wire position and of the overall infrastructure, including bridges, tunnels etc. In recent years, Russian Railways intends to develop high-speed rail transportation, which requires the emergence of new track diagnostic systems. These systems must be able to monitor track condition at high speed (e.g., speed of South Korean high-speed train HEMU-430X reaches 430 kmph). Development of new diagnostic systems should be made using the international experience in this field. The article provides a perspective of the optimal specifications of railway track diagnostic system, as well as the results of the development and testing of Optical-Inertial System for Railway Track Diagnostics, consisting of Track Defects Diagnostics System (TDDS) [19] and Track Geometry Measurement System (TGMS). The possibility of compensating the FOG error caused by their sensibility to the magnetic field (which is substantial in railway infrastructure conditions) will be considered. Optical-Inertial System for Railway Track Diagnostics algorithms will be given. TGMS prototype hardware and software will be shown. 2. Inertial technologies in measurement of geometrical parameters of a railway Fig. 1 shows the routine measurement of the main geometrical parameters. Track gauge is measured using laser scanners of the rails profile (a). Level is measured as the difference in height of the adjacent running tables computed from the angle between the running surface and the horizontal reference plane using INS (b). Sagging is measured using a displacement sensor axle box/bogie as the displacement of the rail relative to the chord joining two neighboring wheel pairs (c).
a) b) c) Figure 1. The routine measurement of the main geometrical parameters A number of measured track parameters, as well as the accuracy of their measurements and the operating speed are different for different laboratory cars and surely meet the State standards (see table 1). CNII-4MD PIC-PROGRESS Table 1 ERA IN- FOTRANS INTEGRAL TVEMA European standard Cross level, mm Range ±160 ±155 ±160 ±225 Error ±3,0 ±0,8 ±1,0 ±0,5 Parameter Gauge, mm Range 1520-10/+40 1520-10/+40 1520-10/+40 nominal -15/+50 Error ±1,0 ±0,8 ±0,8 ±0,5 Sagging, mm Range ±50 ±50 ±50 Error ±2,0 ±0,8 ±1,0 Maximum car speed, km/h 160 160 160 200/300 Table 1 shows that the new diagnostic system for Russian Railways must perform measurements at significantly higher speeds and with higher accuracy. A track parameters measurement accuracy is preset primarily by the characteristics of INS, which forms the base coordinate system.
Fusion with GNSS and odometer usually made to improve SINS accuracy is the essence of implementation of integrated orientation and navigation systems (IONS). Let's consider some of the features of IONS realization on a railway-carriage. Fig. 2 shows the possible SINS locations implemented in Russian laboratory cars (speed up to 160 km/h). Option (a) provides the most efficient use of SINS for rail track geometric parameters diagnostics (see. table 1), and it is the most optimal engineering solutions (laboratory car INTEGRAL, TVEMA). SINS and profilers (laser scanners) are mounted on the solid construction installed to the sprung part of the bogie. Figure 2. Possible SINS locations implemented in Russian laboratory cars The disadvantage of option (b) (laboratory car ERA, INFOTRANS SPC) is the complexity of providing stiffness of additional beam, where profilers are mounted. However, the advantage of this option is more favorable conditions for SINS operation, because thanks to the extra suspension, fluctuations on the platform are in a low-frequency region in comparison with the vibrations of the bogie frame (option a). The disadvantage of option (c) (CNII-4MD, PIC-PROGRESS) is that contact point of "wheel-rail" high-frequency motion component should be defined to measure sagging. Unreliable sensors system "axlebox/carriage" performs these measurements. It constitutes a rope stretched across the roller, change of the rope length is controlled (see Fig. 3). Furthermore, in case of BINS installation on the carriage body it identifies the movement of 17-meter chord (carriage base) along the track. Thus, not a true track position is measured, but the trajectory of the car base. Figure 3. "Axlebox/carriage" sensor The same argumentation should be noted in the context of high-speed (300 km/h) highprecision measurements of the basic geometric parameters in accordance with the stand-
ard [1]. Rigidity in this case is even more important. In this case, it is necessary to combine SINS and optical scanners of rail head inner profiles in a monoblock unit. With this configuration SINS as reference coordinate system builder could provide information about the track centerline and monoblock rotation angles, and hence the scanners angles with respect the rails. 3. Concept of Optical-Inertial System for Railway Track An analysis of the requirements for modern track diagnostic systems was formed concept of Optical-Inertial System for Railway Track (see. Fig. 4). Figure 4. Optical-Inertial System for Railway Track Optical-Inertial System for Railway Track consists of Track Defects Diagnostics System (TDDS), described in [19], and Track Geometry Measurement System (TGMS), which will be discussed below. TGMS is a monoblock construction containing a SINS and a profilometer based on two laser scanners. It is installed on the bogie. One of the major problems in the field of track monitoring is determining the position of diagnostic results on the railway. There should be two more subsystems: GNSS (GLONASS, GPS, etc.) receiver and odometer, in addition to SINS for solving of this problem.
To ensure the dynamic characteristics of SINS fiber optic gyroscopes are used as sensors. Required FOG range is determined by the minimum speed (5 km/h for Russian Railways) when driving on the curve of the minimum radius: 2,83 180 R min Rmin 170m 13 /sec. Random drift is determined by the minimum speed and measuring chords b and L definition accuracy (see. Fig. 5). Alignment of rail (similar to the sagging in plan) is measured with respect to these chords. Figure 5. Bogie yawing in a curve Several SINS based on FOG of ZG Optique SA production (see Fig. 6) meet these requirements [20]. These systems have been tested on turntables (see fig. 7а). Allan variances were figured out (see. Fig. 7b), and confirmed competence in random drift. There are two basic random components in the output signal of this model FOG: angle random walk and bias instability. Angular rate white noise determined from the Allan variance (see fig. 7b) comes to 2, 28 10 /s/ Hz 0,015 / h 4. Speed of integrated signal (angle) increasing over time depends on angle random walk. That is, for a given FOG model an error in the angle determination will be 0.002 for 1 minute of work, 0.01 for 30 minutes. Figure 6. SINS of ZG Optique SA production
a) b) Figure 7. a) Turntable, b) Allan variance Monitoring of magnetic environment both during testing and during operation of the FOG has been proposed since there are significant variations of the magnetic field in the rail infrastructure. FOG triad with the magnetometer board was mounted on a turntable so that the Earth's rotation angular velocity projection on the axis of the triad had the following form: x 0; (cos cos sin sin ); (sin cos cos sin ). y z Then, when substituted '' 59 56 and 30 it is x 0, y 0 and z 15 /h. So at different angular positions around the axis z the y- and x- gyroscopes bias variations may be caused solely by a change of magnetic field vector projections on the axis of the triad. Let's associate a coordinate system magnetometer (see fig. 8). оx with the measurement axes of the m ymzm Magnetometer board FOG triad z, y m x, x m Turntable o y, z m Figure 8. FOG triad and magnetometer position on the turntable
The following model has been proposed to assess the additional offset due to the magnetic field influence: where s i FOG bias; s R i 0i i 0i Ai M jk, (1) 0i constant component; R i gyro bias caused by magnetic field characterized by the first harmonic of the Fourier series, A ( i x, y, z) expansion coefficient; M ( jxyzk,, ; xyzi,, ; j k) the projection of the magnetic induc- jk tion on the plane orthogonal to the i-gyro measurement axis. FOG signals have harmonic component. There is a correlation between FOG and magnetometers: x-fog with y-magnetometer, and y-fog with x-magnetometer. Expansion coef- i ficients (Table 2, where i x y, g, g - magnetometer bias, j x m, ym ) are obtained by M s j fitting FOG experimental data (fig. 9а x-fog, fig. 9b y-fog) and magnetometer readings according to expression (1). a) b) Figure 9. a) x-fog bias; b) y-fog bias FOG sensitivity to the magnetic field was about 1.5 104 2,5 104 º/h/T. Table 2 x g y g X m y m s i, º/h -238.5 185.3 (1) A i, º/h 0.534-0.186 (2) A i, º/h -0.350-0.705 M s j, 10-4 T -0.029-0.001 ( 1) A j, 10-4 T 0.032 0.205 ( 2) A j, 10-4 T 0.205-0.024 Thus, FOG error model was refined sensitivity to the magnetic field was considered. Magnetometers were introduced into SINS structure for analytical elimination of this de-
pendence. Such an approach would correct SINS readings, as well as perform preliminary determination of the course in the initial alignment mode, which can be very important in the case of lack of information from GNSS (the carriage starts from the shaded area of the track). 4. Navigation algorithm The general scheme of the Optical-Inertial System for Railway Track is shown on fig. 10. It is noteworthy that the development of an integrated navigation system for use on railways requires translation of GNSS data to the SINS installation location. A system of bogie/body position encoders is used for this purpose. Figure 10. Optical-Inertial System for Railway Track structure Since all controlled parameters of the railway track condition are measured against the distance covered by the track measurement car, accuracy requirements for said distance are extremely high. During the measurement of the railway track s geometrical parameters found flaws must be located with high precision. The primary device for metering the covered distance is an odometer. Odometer generates N pulses per each turn of the wheel. Its error model can be represented in the form m3 0 1 2 c od, S S mmv m V SS R (2)
where S0 is the setup error; m 0,005...0, 05 is the slippage coefficient; m 1,m2 and m 3 are speed, acceleration and curvilinear motion dependency ratios; Sc is the error caused by the track s roughness and the car s oscillations inside the track; od is the random instrumental error. Using nothing but the odometer makes it impossible to accurately locate the found flaws because of its errors, which depend of various factors such as speed, wheel slippage, meteorological conditions, etc. and can be as high as 5 meters per kilometer. Manual and automatic methods of odometer correction exist. Manual method requires pressing the synchronization button in a timely manner, when the train passes by distance posts (ranging points). This is difficult to accomplish on high speed. Automatic correction requires equipping the track with navigational beacons (such as InduSI). The railway car must be fitted with a corresponding readout device. Such method implies additional financial investments in the railway infrastructure. The most effective method of increasing the accuracy of covered distance measurement is fusion of the odometer data, SINS data, GNSS data, and data from the navigational beacons. Algorithms of geoinformation database building using microelectromechanical modules as navigational beacons are described in [19]. The distinctive feature of the IONS design is odometer error estimation. To achieve that, an algorithm shown on fig. 11 is used. This algorithm allows to estimate the slippage coefficient m: m k 2 2 2 k k k kk kk kk Depending on the quality of GNSS signal, two designs of IONS can be used: correcting odometer with coordinates if GNSS signal is valid and with speeds if it s not. In case of the former, the first step of attacking the problem is translation of the data from these devices to a consistent form. For this purpose odometer data is translated to increments of the geographical coordinates using the heading and pitch values, which are obtained from the SINS. Increments of the geocentric coordinates obtained from GNSS are also translated to increments of the geographical coordinates. One has to take into account the difference between wheelpair and GNSS antenna trajectories. Measurement vector takes this form: X s r s r s r 1 dl, dl, dk, dk, d, d, d', d', d', d ', d ', d ' т
where dk s, d s and dk r, d r are systematic and random components of track s heading and pitch angle measurement errors; dl s and dl r are systematic and random components of displacement vector l length error dl; d r d r, d r are errors of measuring the increments of the geographical coordinates with GNSS., Figure 11. IONS algorithm In existing track-measuring devices the heading angle and the falling gradient are calculated with special expressions and are not equal to the bogie s respective parameters. This leads to the necessity of using bogie/body relative displacement encoders to get the bogie s heading and pitch angles. Switching to velocity correction allows to get rid of this necessity. The scheme on fig. 12 shows that if SINS is installed near the pivot pin, average values of SINS and bogie s velocity vectors have roughly the same values and directions, although the bogie/body relative angular orientation changes during the Figure 12. Bogie movement car s movement. Speed information can be effectively obtained from SINS or SINS integrated with GNSS. Comparing the odometer speed components with corresponding SINS values, one can estimate the latter. In this case, measurement vector looks like this: т X ds, dr, υod s, υod r, υins, υ h s INS, υ h r INS, υ v s INSv r where υins and υ h INS are horizontal and vertical components of INS speed error. v
Corrected value of the distance increment is determined from the formula 2 2 h v od S l υˆ υˆ / υ. IONS testing allowed to correct the nominal value of the odometer s step width depending on the movement speed (fig. 13, a). It is concluded that the odometer s step width is prone to errors. The reason for this is that the measurement wheel s radius decreases during operation. Also, the wheel itself systematically slips against the rail. Altitude variations obtained from the GNSS (h GNSS ), the SINS (h INS ) and the correction algorithm ( h A ), are shown on fig. 13, b. a) b) Figure 13. a) Odometer correction b) Altitude variations The dependency diagrams of h GNSS and h A are intentionally spaced by a constant value (~17 m) for illustrative purposes. As the behavior of h INS clearly shows, the SINS inclination angle ψ can t be used as the rail pitch angle. That s because (1) SINS has systematic errors of angular orientation measurement, and (2) SINS is installed inside the car s body which can, for a variety of reasons, have a constant angular misalignment with the measurement bogie (whose angular orientation describes the orientation of the railway). Experiments allowed to rectify the analytical model of the odometer error, which defines its dependencies from calibration errors, speed, acceleration, and radius of curvature: m3 * S S0mmυ 1 m2υ 1m sign υod SSt od, R where S0 is the initial setup error; m 0,005...0, 05 is the slippage coefficient; m 1, m2 and m 3 are speed, acceleration and curvilinear motion (with R being the radius of curvature)
dependency ratios; υ od is movement speed as measured by the odometer; * m is the coefficient of dependence of the slippage value from the measurement wheelpair position during movement; St is the error caused by the track s roughness and the car s oscillations inside the track; od is the random instrumental error. 5. Some practical results At the first stage of the Optical-Inertial System for Railway Track development, a prototype was made in order to test the system and demonstrate its capabilities to eventual buyers. The prototype s picture is shown on fig. 14. OISRTD prototype consists of following parts (see fig. 14, a): SINS based on fiber-optic gyros; Two laser scanners; Odometer; Lead-acid batteries ; Laptop. Folding handle Folding shelf (for laptop) SINS GNSS aided Power supply Laser scanners in protective cases Odometer Figure 14 а)
b) c) Figure 14. Optical-Inertial System for Railway Track prototype The SINS used in this prototype has common performance and delivers data at 100Hz frequency. The data is transmitted by 3 RS-422 channels and 1 RS-232 channel. The RS- 422 channel by which the inertial data is transmitted is connected to the laptop via an RS422/USB converter. The power cord is connected to the battery. The laser scanners are mounted in two metal housings. Data is transmitted to the laptop by FireWire. The odometer is an angular sensor mounted on one of the wheels. It is able to measure the distance travelled by the bogie with accuracy of about 0.5 mm. It is connected to the laptop with USB. The lead-acid batteries are two standard 12V 7Ah accumulators, yielding total voltage of 24V DC. 14.0 14.0 Figure 15. Optical-Inertial System for Railway Track prototype interface
Fig. 14, b and fig. 14, c show the prototype s testing in the Republic of Korea. On fig. 14, b a rail head with laser surface analyzer beam can be seen. A screenshot of the prototype s software is shown on fig. 15. 6. Conclusion This article deals with various measurement schemes of track measurement cars. Worthwhileness of installing a single housing with SINS and surface analyzer directly on the bogie frame was shown. Possibility of fiber-optical gyros application for forming SINS for rail track diagnostics was demonstrated. However, for effective use of FOGs as part of SINS in railway conditions, their magnetic sensitivity should be corrected. Further research should be carried out to elaborate a rail profile recognition algorithm with the use of SINS data. Authors acknowledge funding through Russian Science Foundation under the contract 14-19-00693. References [1] British standard. Railway applications Track Track geometry quality. Part 1: Characterisation of track geometry. BS EN 13848-1:2003+A1:2008. [2] C. Hemmrich, J. Schmeister, H.-G. Thies, H. Zuck, ETR Eisenbahntechnische Rundschau, 2007, Nr.6, р. 359-363. [3] D. Li. Railway Track & Structures, 2005, Nr 9, p. 19 23. [4] G. Staccone. Railway Technical Review, 2006, Nr 3, р. 36 39. [5] www.matisa.ch [6] www.mermec.com [7] T. Judge. Railway Age, 2004, Nr 9, р. 77 80. [8] W. Weart. Progressive Railroading, 2009, Nr 3, p. 47 53; [9] K. Cordner. Modern Railways, 2004, Nr 666, р. 66, 68 72, 74. [10] M. Barberon. La Vie du Rail, 2006, Nr 3055, р. 4 6. [11] J.-M. Descusses. Le Rail, 2006, Nr 132, р. 34 38. [12] J. Braband. Signal und Draht, 2007, Nr 4, p. 34 37. [13] F. Auer. ZEVrail Glasers Annalen, Special Edition: Strategy of Track Maintenance, 2005, p. 38 45. [14] W. Hanreich. Glasers Annalen, 2005, special edition, p. 17 26.
[15] B. Robert, J. Derocher. Progressive Railroading. 2004, Nr 6, p. 50 52. [16] www.pikprogress.ru [17] www.infotrans-logistic.ru [18] www.tvema.ru [19] A. Boronahin, D. Larionov, Yu. Filatov, L. Podgornaya, E. Bokhman, R.Shalymov. Inertial System for Railway Track Diagnostics/ Proc. on Symp. Inertial Sensors and Systems, Germany, 18 19 Sept, 2012, Karlsruhe, P. 17.1 17.20. [20] http://www.zgoptique.ch/
本文献由 学霸图书馆 - 文献云下载 收集自网络, 仅供学习交流使用 学霸图书馆 (www.xuebalib.com) 是一个 整合众多图书馆数据库资源, 提供一站式文献检索和下载服务 的 24 小时在线不限 IP 图书馆 图书馆致力于便利 促进学习与科研, 提供最强文献下载服务 图书馆导航 : 图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具