Estimation device and procedure for determining estimated movement values
Patent Information
- Authority / Receiving Office
- ES · ES
- Patent Type
- Patents
- Current Assignee / Owner
- SIEMENS MOBILITY GMBH AT
- Filing Date
- 2023-06-12
- Publication Date
- 2026-07-10
AI Technical Summary
Existing methods for determining the speed of rail vehicles lack accuracy and are costly, particularly in estimating motion parameters with high precision.
An estimation device utilizing a Kalman filter, slip module, and evaluation module to determine vehicle speed, incorporating torque, vertical force, and wheel speed values, with a slip module adjusting wheel circumferential speed based on axle slippage, and an evaluation module refining estimates based on Kalman-estimated vehicle speed and slip values.
Achieves highly accurate motion estimates with low hardware and time costs, enabling precise slip regulation and improved traction and braking performance.
Smart Images
Figure 00000011_0000 
Figure 00000012_0000 
Figure 00000013_0000
Abstract
Description
[0001] In the field of railway engineering, a relatively accurate determination of the speed of rail vehicles is of great importance in order to meet the requirements that are common on the part of operators today.
[0002] From DE 10 2019 211934 A1 a method for determining the speed of a train is known.
[0003] The invention is based on the objective of providing an estimation device with which motion estimates describing the movement of a rail vehicle can be determined with relatively high accuracy.
[0004] This problem is solved according to the invention by an estimating device with the features according to claim 1. Advantageous embodiments of the estimating device according to the invention are specified in the dependent claims.
[0005] The following is provided: an estimation device for determining a new motion estimate value that indicates the speed of a rail vehicle having at least two axles, wherein the estimation device comprises a Kalman filter, which is inputted with at least torque values, vertical force values, wheel size values and wheel speed values and determines a Kalman-estimated vehicle speed value based on these values and on the basis of a Kalman filter; a slip module, which is inputted with the torque values, wheel size values, wheel speed values and the respective previous motion estimate of the estimating device and determines a wheel circumferential speed value and a slip value indicating any slippage of the respective axle; and an evaluation module, which determines the new motion estimate of the estimating device based on the Kalman-estimated vehicle speed value, the wheel circumferential speed value and the slip values.
[0006] A significant advantage of the estimation device according to the invention is that, with the Kalman filter provided according to the invention in combination with the slip module according to the invention, very accurate motion estimates can be determined with low hardware costs and low time costs, which can be used for the operation of the rail vehicle.
[0007] Regarding the slip module, it is considered advantageous if, during braking operation of the rail vehicle and without any slippage of axles, the slip module outputs the wheel circumferential speed value of the axle with the highest wheel circumferential speed value to the evaluation module, during drive operation and without any slippage of axles, it outputs the wheel circumferential speed value of the axle with the lowest wheel circumferential speed value to the evaluation module, and when all axles are slipping, it outputs a general slip value to the evaluation module.
[0008] Regarding the evaluation module, it is considered advantageous if, during braking operation of the rail vehicle and without any slippage of axles, the evaluation module outputs the wheel circumferential speed value of the axle with the highest wheel circumferential speed value as the new motion estimate; during driving operation and without any slippage of axles, the wheel circumferential speed value of the axle with the lowest wheel circumferential speed value is output as the new motion estimate; and when all axles are slipping, the new motion estimate is determined based on the Kalman-estimated vehicle speed value of the Kalman filter.
[0009] It is advantageous if the estimating device has a vertical force calculation module that generates a vertical force value per axle, at least on the basis of the force moment values also used by the Kalman filter, the wheel size values and the respective previous motion estimate of the estimating device, wherein each of the vertical force values of the vertical force calculation module forms one of the vertical force values used by the Kalman filter.
[0010] It is also advantageous if the estimating device has a friction braking torque calculation module that generates a friction braking torque value per axle, based at least on a brake pressure value per axle, the wheel size values and the respective previous motion estimate of the estimating device, with each of the friction braking torque values forming one of the force torque values used by the Kalman filter.
[0011] The Kalman filter preferably uses axle-related engine torque values as force torque values, or at least takes them into account as well.
[0012] The Kalman filter is preferably an extended Kalman filter.
[0013] The estimating device preferably outputs a vehicle acceleration value indicating the acceleration of the rail vehicle as a further motion estimation value.
[0014] Furthermore, it is considered advantageous if the Kalman filter also determines a force-lock value per axis relating to the force-rail contact and the estimating device also outputs the force-lock values.
[0015] The estimation device preferably comprises a computing device and a memory in which the following are stored: a Kalman filter software module, which, when executed by the computing device, forms the Kalman filter; a slack software module, which, when executed by the computing device, forms the slack module; and an evaluation software module, which, when executed by the computing device, forms the evaluation module.
[0016] Preferably, a vertical force calculation software module is also stored in the memory, which, when executed by the computing device, forms the vertical force calculation module.
[0017] Preferably, a friction braking torque calculation software module is also stored in the memory, which, when executed by the computing device, forms the friction braking torque calculation module.
[0018] The invention also relates to a rail vehicle equipped with an estimating device as described above.
[0019] Regarding the advantages of the rail vehicle according to the invention and its advantageous embodiments, reference is made to the above statements in connection with the estimating device according to the invention and its advantageous embodiments.
[0020] The estimating device is preferably connected to a slip controller which, using the estimated friction values per axle, the wheel circumferential speed values and the respective motion estimate of the estimating device, regulates the slip of the respective axle.
[0021] It is also advantageous if the estimating device is linked to a plausibility check device that performs a plausibility check on the movement estimates of the estimating device with regard to movement data from other sources.
[0022] The invention further relates to a method for determining a new motion estimate that indicates the speed of a rail vehicle having at least two axles, wherein in the method at least torque values, vertical force values, wheel size values and wheel speed values are fed into a Kalman filter on the input side and a Kalman-estimated vehicle speed value is determined on the basis of these and on the basis of a Kalman filter, a wheel circumferential speed value and a slip value, which indicates any slippage of the respective axle, are determined with the wheel size values, the wheel speed values and the motion estimate determined previously for each axle, and the new motion estimate is determined on the basis of the Kalman-estimated vehicle speed value, the wheel circumferential speed value and the slip values.
[0023] Regarding the advantages of the method according to the invention and its advantageous embodiments, reference is made to the above statements in connection with the estimating device according to the invention and its advantageous embodiments.
[0024] The invention is explained in more detail below with reference to exemplary embodiments; these show, by way of example, Figure 1 shows an embodiment of an estimating device according to the invention in the form of a block diagram, Figure 2 shows a computer system in which the estimating device is implemented according to Figure 1 is implemented and accordingly the valuation facility according to Figure 1 forms, and Figure 3 a rail vehicle equipped with the estimating device according to the Figuren 1 and 2 is equipped.
[0025] For the sake of clarity, the same reference symbols are used in the figures for identical or comparable components.
[0026] The Figure 1Figure 1 shows an embodiment of an estimating device 1 according to the invention in the form of a block diagram. The estimating device 1 comprises a friction braking torque calculation module 10, a vertical force calculation module 20, an extended Kalman filter 30, a slip module 40, and an evaluation module 50.
[0027] The estimating facility 1 has entrances for Brake pressure values in the form of brake cylinder pressures p_i, electric motor torque values m_e_i (hereinafter also referred to as motor torque values), wheel size values, for example in the form of dynamic wheel radii r_i and wheel speed values ω _i, which are also referred to below as axle speeds.
[0028] The index i designates the respective axis. In the embodiment according to Figure 1 For example, it is assumed that the rail vehicle for which the estimating device is used has four axles, so i can take on values between 1 and 4.
[0029] The estimating facility 1 has outputs for Force-lock values in the form of adhesion values µ_i, which describe the wheel / rail contact, a new motion estimate in the form of a velocity estimate v_virt and a new motion estimate in the form of an acceleration estimate a_virt.
[0030] In the friction braking torque calculation module 10, friction braking torque values m_p_i per axle are calculated from the measured brake cylinder pressures p_i, the dynamic wheel radii r_i, geometric constants (such as the friction radius) and a speed-dependent coefficient of friction between wheel and brake disc.
[0031] In the vertical force calculation module 20, the friction braking torque values m_p_i, the electric motor torques m_e_i, the dynamic wheel radii r_i, and the axle speeds are used to calculate the following: ω_i known driving resistances, the known mass of the car body and known geometric factors dynamic wheel contact forces Q and calculated.
[0032] The driving resistances F in can be approximately determined using empirically derived constants and from the respective vehicle speed.
[0033] The Kalman filter 30 is implemented using a control-engineered observer. The mathematical model preferably uses the equations of motion for the wheel and the car body according to the following equations (1) - (5): M s ¨ = F w + Q 1 μ 1 + Q 2 μ 2 + Q 3 μ 3 + Q 4 μ 4 + F Kupl I 1 ω 1 . = m _ p 1 + Q 1 μ 1 r 1 + m _ e 1 I 2 ω 2 . = m _ p 2 + Q 2 μ 2 r 2 + m _ e 2 I 3 ω 3 . = m _ p 3 + Q 3 μ 3 r 3 + m _ e 3 I 4 ω 4 . = m _ p 4 + Q 4 μ 4 r 4 + m _ e 4
[0034] In the translational direction according to equation (1), the following can be derived from Newton's laws: the sum of the driving resistances ( F in ), products from the wheel contact forces Q and (calculated in the vertical force calculation module 20) and the adhesion values µ i , coupling forces F Kupl between car bodies, whereby the coupling forces, if not measured, can alternatively be neglected, and the product of mass M and acceleration s̈ .
[0035] The mass M consists of a static component, which is known through engineering, and an operational component, which can be measured by spring forces of the car body.
[0036] In the rotational direction, the following can be derived using equations (2) to (5): the resulting axle-related engine torque values m_e _i, which result from the electric motor drive and brake torques, where the motor torque values m_e _i, for example, can be calculated from the electrical parameters of the motor, the pneumatic braking torques m_p _i, which generates the friction braking torque calculation module 10, a torque based on the wheel contact force Q and(calculated in the vertical force calculation module 20), the adhesion value µ i and the wheel radius r i , and the product of the moments of inertia And and and the wheelset accelerations ω̇ and , where the moments of inertia And and the axles are known through engineering and the wheelset accelerations can be determined by deriving the axle rotation speeds.
[0037] The equations of motion are transformed into the state-space representation and made available to the Kalman filter 30 as a mathematical boundary condition: x → ˙ = f x → u → with u → = m _ p 1 , m _ p 2 , m _ p 3 , m _ p 4 Q 1 , Q 2 , Q 3 , Q 4 , m _ e 1 , m _ e 2 , m _ e 3 , m _ e 4 , r 1 , r 2 , r 3 , r 4 and x → = ω 1 ω 2 ω 3 ω 4 μ 1 μ 2 μ 3 μ 4 s ˙ follows x → ˙ = 1 I 1 u 1 + u 5 x 5 u 13 + u 9 1 I 2 u 2 + u 6 x 6 u 14 + u 10 1 I 3 u 3 + u 7 x 7 u 15 + u 11 1 I 4 u 4 + u 8 x 8 u 16 + u 12 0 0 0 0 1 M … x 9 2 + u 9 + u 5 x 5 + u 6 x 6 + u 7 x 7 + u 8 x 8
[0038] The state vector x has n=9 states and the process noise has dimension Q (9 x 9)< , where the diagonal entries of the matrix are occupied.
[0039] The wheel rotation speeds are measured and fed back to the Kalman filter 30 as observations. ω and, which provides m=4 observations. This results in an observation matrix. H (4 x 9)< and a diagonally populated covariance matrix of the measurement noise R (4 x 4)< .
[0040] The diagonal entries of the covariance matrix and the process noise matrix are preferably determined empirically.
[0041] From the entrances ( x, u The Kalman filter 30 estimates the current adhesion values as well as a velocity v_k, which is subsequently called the Kalman velocity, as frictional friction values.
[0042] The slip module 40 is calculated from the wheel speeds. ω and and the wheel radii r_i determine the axle speed value v_circulation according to: In braking mode (detected by the sum of the friction brake torque values and motor torque values), the fastest axle defines the axle speed value. In drive mode, the slowest axle defines the axle speed value.
[0043] Furthermore, the state "is sliding" is determined using the returned velocity or the motion estimate v_virt. This state expresses whether the vehicle is in all-axle sliding. The following criteria are used for this purpose: the variance of the axle speeds, the first time derivative of the axle speeds, the second time derivative of the axle speeds and the slip deviation of the axle speeds from a reference speed, which may, for example, correspond to the last determined speed estimate v_virt.
[0044] In evaluation module 50, the new velocity estimate v_virt and its time derivative a_virt are formed from the Kalman velocity v_k and the axis velocity value, preferably according to: Without all-axle sliding (is_slide == 0), the axle velocity value (v_circulation) defines the velocity estimate v_virt. If all-axle sliding is present (is_slide == 1), the Kalman velocity v_k is used for the new velocity estimate v_virt. The Kalman velocity v_k is freed from an offset by being "pulled up" or "pulled down" to the last axle velocity value (without all-axle sliding) and determines the new velocity estimate v_virt during all-axle sliding.
[0045] Differentiating and smoothing the new velocity estimate v_virt yields the new acceleration estimate a_virt, which can be used to verify the plausibility of the measured vehicle acceleration.
[0046] In summary, based on the brake cylinder pressures, the engine torques, the dynamic wheel diameters and the wheel speeds, the estimation device 1 can determine the non-measurable adhesion values as well as new speed estimates v_ and new acceleration estimates a_virt, which can be used, for example, to verify the plausibility of a measured vehicle acceleration.
[0047] A downstream controller uses the estimated adhesion values per wheelset and the speed (i.e., the estimated motion value v_virt) to adjust the slip, thus achieving the highest possible adhesion values. This minimizes the stopping distance and maximizes the traction force.
[0048] The acceleration estimate a_virt can be used – as already mentioned – to verify the plausibility of the measured acceleration. This diverse redundancy can either increase safety, eliminate the need for additional hardware, or allow for greater weighting of the readings from the installed sensors.
[0049] Additionally, the new velocity estimate v_virt can be used to adjust the time between the detection of all-axle slippage and the triggering of drift protection. Drift protection is preferably not executed after a fixed time, but rather after the difference between the "classic" reference velocity (i.e., the gradient-limited axle velocity) and the velocity estimate v_virt exceeds a predefined value. This prevents both excessively high slip values and unnecessary axle deceleration.
[0050] Estimation device 1, for example, also makes it possible to give greater weight to acceleration sensor, GPS, or radar data for determining the reference speed, as these can be validated using a diverse redundant method based on existing data. Furthermore, it enables a more accurate estimation of the coefficient of friction, which leads to an optimization of the coefficient of friction utilization in slip control during traction and braking.
[0051] Additionally, the drift protection intervention can be adapted to the environmental conditions, thus increasing the accuracy of the reference speed and shortening the stopping distance.
[0052] The Figure 2 This demonstrates the implementation of the estimating device 1 according to... Figure 1in a computer system 100, which forms, or at least also forms, the estimating unit 1. The computer system 100, and thus the estimating unit 1, comprises a computing unit 110 and a memory 120, in which a Kalman filter software module M30 is stored, which, when executed by the computing unit, generates the Kalman filter 30 according to Figure 1 forms a spoofing software module M40, which, when executed by the computing device, forms the spoofing module 40 according to Figure 1 forms, and an evaluation software module M50, which, when executed by the computing device, evaluates the module 50 according to Figure 1 forms, are stored.
[0053] Furthermore, the memory 120 contains a vertical force calculation software module M20, which, when executed by the computing device, corresponds to the vertical force calculation module 20. Figure 1 forms, and a friction brake torque calculation software module M10, which, when executed by the computing device, forms the friction brake torque calculation module 10 according to Figure 1 forms, stored.
[0054] The Figure 3 shows an embodiment of a rail vehicle 200, which is equipped with a computer system 100 according to Figure 2 and thus with an estimating device 1 according to Figure 1 is equipped.
[0055] The rail vehicle 200 also includes a slip controller 210, which, using the estimated friction values µ_i per axle, the wheel circumferential speed values and the respective motion estimate v_virt of the estimating device 1, regulates the slip of the respective axle.
[0056] Furthermore, the estimation device 1 is connected to a plausibility check device 220, which performs a plausibility check on the movement estimates of the estimation device 1 with regard to movement data from other sources.
[0057] Finally, it should be mentioned that the features of all the embodiments described above can be combined with each other in any way to form further embodiments of the invention.
[0058] Furthermore, all features of dependent claims can each be combined with each of the subordinate claims, either individually or in any combination with one or more other dependent claims, to obtain further embodiments.
Claims
1. Estimating device (1) for determining a new estimated movement value (v_virt) which indicates the speed of a rail vehicle (200) having at least two axles, comprising - a Kalman filter (30) which is supplied on the input side at least with torque values (m_p_i, m_e_i), vertical force values (Q_i), wheel size values (r_i) and wheel rotary speed values (ω_i) and, on the basis thereof and on the basis of a Kalman filtration, determines a Kalman-estimated vehicle speed value (v_k), - a slip module (40) that is supplied on the input side with the torque values (m_p_i, m_e_i), the wheel size values (r_i), the wheel rotary speed values (ω_i) and the respective prior estimated movement value (v_virt) from the estimating device (1) and determines a wheel circumferential speed value (v_circulate) and a sliding indication which indicates any sliding of the respective axle, and - an evaluating module (50) which determines the new estimated movement value (v_virt) of the estimating device (1) on the basis of the Kalman-estimated vehicle speed value (v_k), the wheel circumferential speed value (v_circulate) and the sliding indications.
2. Estimating device (1) according to claim 1, characterised in that the slip module (40) - during a braking operation of the rail vehicle (200), outputs the wheel circumferential speed value (v_circulate) of the axle with the highest wheel circumferential speed value to the evaluating module (50), - during a driving operation, outputs the wheel circumferential speed value (v_circulate) of the axle with the smallest wheel circumferential speed value to the evaluating module (50), and - during sliding of all the axles, outputs an all-sliding indication (actual_sliding) to the evaluating module (50).
3. Estimating device (1) according to claim 2, characterised in that the evaluating module (50) - during a braking or driving operation of the rail vehicle (200) and no sliding of all the axles, outputs the wheel circumferential speed value (v_circulate) as the new estimated movement value (v_virt), and - during sliding of all the axles, determines the new estimated movement value (v_virt) on the basis of the Kalman-estimated vehicle speed value (v_k) of the Kalman filter (30).
4. Estimating device (1) according to one of the preceding claims, characterised in that - the estimating device (1) has a vertical force calculating module (20) which generates, at least on the basis of the torque values (m_p_i, m_e_i) that are also used by the Kalman filter (30), the wheel size values (r_i) and the respective prior estimated movement value (v_virt) of the estimating device (1), a vertical force value (Q_i) per axle, - wherein each of the vertical force values (Q_i) of the vertical force calculating module (20) forms one of the vertical force values (Q_i) that is used by the Kalman filter (30).
5. Estimating device (1) according to one of the preceding claims, characterised in that - the estimating device (1) has a friction braking torque calculating module (10) which generates, at least on the basis of a braking pressure value (p_i) per axle, the wheel size values (r_i) and the respective prior estimated movement value (v_virt) of the estimating device (1), a friction braking torque value (m_p_i) per axle, - wherein each of the friction braking torque values (m_p_i) forms one of the torque values (m_p_i, m_e_i) that is used by the Kalman filter (30).
6. Estimating device (1) according to one of the preceding claims, characterised in that the Kalman filter (30) uses axle-related motor torque values (m_e_i) as torque values (m_p_i, m_e_i) or at least also takes these into account.
7. Estimating device (1) according to one of the preceding claims, characterised in that the Kalman filter (30) is an extended Kalman filter (30).
8. Estimating device (1) according to one of the preceding claims, characterised in that the estimating device (1) outputs, as a further estimated movement value (a_virt), a vehicle acceleration value indicating the acceleration of the rail vehicle (200).
9. Estimating device (1) according to one of the preceding claims, characterised in that - the Kalman filter (30) also determines a traction value (µ_i) relating to the force-rail contact per axle and - the estimating device (1) also outputs the traction values (µ_i).
10. Estimating device (1) according to one of the preceding claims, characterised in that the estimating device comprises a computing device (110) and a memory store (120) in which are stored - a Kalman filter software module (M30) which, when executed by the computing device, forms the Kalman filter (30), - a slip software module (M40) which, when executed by the computing device, forms the slip module (40), and - an evaluating software module (M50) which, when executed by the computing device, forms the evaluating module (50).
11. Estimating device (1) according to claim 10, characterised in that stored in the memory store (120) are - a vertical force calculating software module (M20) which, when executed by the computing device, forms the vertical force calculating module (20), and / or - a friction braking torque calculating software module (M10) which, when executed by the computing device, forms the friction braking torque calculating module (10).
12. Rail vehicle (200), characterised in that the rail vehicle is equipped with an estimating device (1) according to one of the preceding claims.
13. Rail vehicle according to claim 12, characterised in that connected to the estimating device (1) is a slip regulator (210) which, by taking account of the estimated traction values per axle, the wheel circumferential speed values and the respective estimated movement value (v_virt) from the estimating device (1), regulates the slip of the respective axle.
14. Rail vehicle according to claim 12 or 13, characterised in that connected to the estimating device (1) is a plausibility checking device (220) which carries out, with the estimated movement values from the estimating device (1), a plausibility check in relation to movement information from other sources.
15. Method for determining a new estimated movement value (v_virt) which indicates the speed of a rail vehicle (200) having at least two axles, wherein in the method - into a Kalman filter (30) on the input side are supplied at least torque values (m_p_i, m_e_i), vertical force values (Q_i), wheel size values (r_i) and wheel rotary speed values (ω_i) and, on the basis thereof and on the basis of a Kalman filtration, a Kalman-estimated vehicle speed value (v_k) is determined, - with the wheel size values (r_i), the wheel rotary speed values (ω_i) and the respective prior estimated movement value (v_virt), a wheel circumferential speed value (v_circulate) and a sliding indication (actual_sliding) are determined, which indicates any sliding of the respective axle, and - on the basis of the Kalman-estimated vehicle speed value (v_k), the wheel circumferential speed value (v_circulate) and the sliding indications, the new estimated movement value (v_virt) is determined.