Automatic detection of wheel profile wear

Abstract

The invention relates to a method for automatically detecting wear (V) of a wheel of a rail vehicle due to travel. According to the method, the rail vehicle travels over a measuring track (MS), the length (ZS) of which is known or ascertained to a highly precise degree. Furthermore, the number (AU) of wheel rotations (U) required to travel over the measuring track (MS) is ascertained. The current diameter (Dm) of the wheel of the rail vehicle is ascertained on the basis of the length (ZS) and the ascertained number (AU) of wheel rotations (MS). Additionally, the wear (V) of the wheel is ascertained by comparing the ascertained current diameter (Dm) with a reference value (Dn, Dm°). The invention also relates to a method for training an artificial intelligence-based algorithm to ascertain the current value of a geometric parameter (Sd, Sh, qR, S, HL) of the wheel profile of a rail vehicle and to a wear detection device (40).

Description

Automatic detection of wheel profile wear The invention relates to a method for automatically determining the travel-related wear of a wheel of a rail vehicle. The invention also relates to a method for training an artificial intelligence-based algorithm to determine a current value of a geometric parameter of a wheel profile of a rail vehicle. The invention further relates to a wear determination device. Rail vehicles are subject to severe wear during operation on account of the exceptionally large masses which are being moved, the high energy flow which occurs during this process, and the pronounced dissipation which results from the high energy flow. This relates in particular to the contact between the wheels of a rail vehicle and the rails traveled on. The wheel profile of rail vehicles, i.e. the part of a wheel of a rail vehicle which ensures stable contact with the rail, is produced according to a standardized geometry (DIN EN 13715) by turning. Owing to the dynamic metal-metal contact, the wheel profile is subject to considerable wear. This wear can be divided into two parts. One part is due to rolling during movement, which is always accompanied by slippage and micro-sliding. On the one hand, this wear manifests itself in a geometric change in the profile, which is "deformed" in relation to the profile prescribed in standards. On the other hand, the wear is attributable to the correction of the profile changes. This correction is effected by turning the underfloor wheel set (reprofiling), in which process material is always removed in order to restore the (standard-compliant) wheel profile, this of course being accompanied by a reduction in the wheel diameter. In order to determine when reprofiling is required, it is customary in most cases to carry out regular measurements of the diameter and profile of a wheel with special devices which measure certain geometric parameters of the wheel and its profile. When these parameters reach certain limits, reprofiling is planned. This is where the first problem lies: the use of most measuring instruments presupposes that the vehicle is stopped in a workshop suitable for testing the wheels (i.e. equipped with a suitable test pit) to enable the personnel trained in the use of these measuring instruments to intervene. One of the required measurements is the diameter of the circle which corresponds to the point D0 (see figure 1) of the profile, which is located in the center of the profile in the transverse direction. When measuring the diameter, the measuring devices currently used provide results which are subject to a measurement error due to their operating principle. For example, some measuring instruments measure the diameter of the flange (diameter at the location of the apex SP in figure 1), then measure the flange height (Sh in figure 1), and then estimate the measured diameter, which relates to the center of the profile. Other instruments, which are arc-shaped, (see figure 2) are mounted on the tread of the wheel and allow indirect measurement of the diameter by measuring the curvature of the arc. In this case, small local deformations (flat areas), surface damage, soiling etc. lead to measurement errors of an order of magnitude (millimeters) which is compatible with the operating limits. The error which occurs in these measurements can (and should) be reduced by taking measurements in different arc sectors on one and the same wheel and then forming the mean value of the results obtained. However, this is particularly awkward since, in a certain position of the stationary vehicle in the workshop, it is almost always only a limited sector of each wheel which is accessible. In order to gain access to other or larger sectors, the vehicle would have to be moved in such a way that the wheels rotate by a sufficient number of degrees to enable different arc sectors to be accessible to the measuring device. This is almost never done in practice because it is often too expensive and because railway workshops are not always equipped to carry out such movements safely. The measurement error is sometimes so great that, when the measurement protocols of two successive inspections are compared, the diameter which is associated with one protocol is not infrequently greater than that which is associated with the preceding protocol, but of course this cannot correspond to the facts since there is increasing wear on the wheels during operation. The value of the wheel diameter is important for the safety and reliability of the vehicle. Since it contributes to the determination of the instantaneous speed with the aid of wheel impulse generators, there is the problem of reducing these measurement errors, checking the accuracy of the measurements carried out with conventional methods and possibly applying more precise, innovative, methods. Other geometric parameters of the wheel profile (which are likewise relevant for vehicle safety) also constantly change as a result of the operation of the vehicle and generally approach their operating limits. Since these parameters (e.g. Sh = flange height, Sd = flange thickness, qR = flange flank dimension) are also determined only during workshop inspections, there is the problem of estimating them more frequently using innovative methods. The object is therefore to develop a method and a device for the simplified determination of wheel wear of a wheel of a rail vehicle. This object is achieved by a method for automatically determining the travel-related wear of a wheel of a rail vehicle according to the independent patent claims. In the method according to the invention for the automatic determination of the travel-related wear of a wheel of a rail vehicle, the rail vehicle traverses a measurement section, the length of which is known with high accuracy or is determined with high accuracy. As will be explained in detail later, there are different methods for accurately measuring a route section. In the simplest case, the route section has markings whose spacing is exactly known. When the rail vehicle passes the marking, the counting of the revolutions is started or ended. Alternatively, exact position measurement of a rail vehicle can also be carried out on the basis of a satellite navigation system, the measurement deviations of which are corrected with additional information. As will be explained in detail below, a differential satellite navigation system is particularly well suited for highly accurate measurement of a measurement section traveled by the rail vehicle. If the length of the measurement section is determined by an internal measuring device of the rail vehicle, wheel wear can be automatically determined independently of measurements by an infrastructure operator. In the case of a single measurement, in particular by the infrastructure operator, on the other hand, it is advantageously possible to increase the accuracy of wear determination in regions in which exact measurement of the length of a measurement section by an internal measuring device is not possible. Such a case may occur, for example, when no differential satellite navigation system is available in one area and the measurement of a simple satellite navigation system needs to be refined. Furthermore, the required number of revolutions of the wheel to traverse the measurement section is determined. To measure the required number of revolutions, use is preferably made of a wheel impulse generator, with which an axle of a modern rail vehicle, in particular a modern locomotive, is equipped. Inter alia, such a wheel impulse generator transmits the instantaneous angular velocity of each axle to the safety systems and to the anti-slip and anti-skid systems. A modern rail vehicle comprises a computing system with which the number of revolutions of each axle and also fractions thereof can be determined with high accuracy in a predetermined time interval or on a predetermined route section. A current diameter Dm of the wheel of the rail vehicle is determined on the basis of the length ZS of the route section and the determined number AU of revolutions of the wheel. The diameter Dm of the wheel is defined as the diameter in a measurement circle plane. The measurement circle plane is preferably arranged at a distance of 70 mm from the inner face of the wheel. The diameter Dm of the wheel is obtained according to the following formula: 7S = ^ (1) A wheel of a rail vehicle has a wheel profile with a generative curve defined by standards, viewed in the cross-sectional direction. The generative curve has a bulge on the part directed toward the inside of the track and a slight taper on the part directed toward the outside of the track, the flange. For simplification, this body can be approximated by a truncated cone. This special geometry of the wheels of a wheel set of a rail vehicle has the effect that, in the case of a slight movement of an axle in the axial direction, for example to the left, the left wheel ideally rolls on a disk with a somewhat larger diameter than the right wheel. Since, however, both wheels are rigidly fastened to the same axle and must therefore rotate in solidarity, that is to say through the same angle at the same time, a restoring moment is produced on contact with the rail, which restoring moment lies in the running plane and returns the wheel set to the center of the track. In the case of a rectilinear movement, this leads practically to an oscillating approximately sinusoidal movement of the axle in the rolling plane. On the other hand, from moment to moment each individual wheel rolls over an ideal point, the rail contact point, which is at a distance from the center of the axle and varies between a minimum and a maximum. The contact point, or more precisely the very small contact area, between the wheel and the rail thus describes an open curve during each revolution, since the end point of a revolution generally does not coincide exactly with its starting point. This is due to the "frustoconical" geometry of the wheel and the resulting transverse movements. However, the point at which the revolution begins and the point at which the revolution ends are very close together. It can therefore be assumed that the distance which the wheel travels during each revolution corresponds to the length of the open curve described above. This is true assuming that the wheel rolls without creeping. This length can be related to the mean diameter of the open curve associated with each revolution. Moreover, if a sufficiently large number of revolutions AU is assumed and if it is known that the wheel rotates on circles with alternately larger and smaller diameters, it can be assumed as an approximation that the mean of the length of the curves described during each revolution is equal to the circumference of the circle which corresponds to the point D0 of the wheel profile in the measurement circle plane. In reality, the contact between the wheel and the rail is not point contact but occurs over a small contact area. The dimensions of this area for the materials in question are very small compared to the surface area of a wheel with a diameter of more than one meter. Therefore, it can be assumed that this area can be approximated to a point and that, after a very large number AU of revolutions, the distance covered by rolling on a circle with a diameter between a minimum and a maximum is virtually identical to the distance which would have been covered if rolling had taken place only on a disk of negligible thickness and a diameter Dm. Assuming that the measurement section or route section ZS covered is larger by many orders of magnitude than the diameter of a wheel, which is approximately one meter, i.e. that the number AU of revolutions is so large that the average of the circles on which the route section ZS is traveled practically corresponds to the diameter Dm of the reference circle for the measurements, the value of the diameter Dm can be calculated according to formula (1). Finally, wear of the wheel profile is determined on the basis of a comparison, preferably by comparison, of the determined current diameter with a reference value of the diameter of the wheel. Wear is to be understood as meaning wear of the tread of a wheel of a rail vehicle, which leads to a change in the wheel profile. As will be explained in detail below, the wear is characterized in particular by the variables flange height Sh, flange thickness Sd, flange flank angle qR, and the rollover S. In this case, use is made of the knowledge that these variables are related to the change in the wheel diameter. For example, the flange height Sh increases if the diameter in the center of the wheel is reduced by wear, since the point D0 in the center of the track profile is used as a reference point for determining the flange height Sh. The flange thickness Sd, on the other hand, is reduced when the diameter of the wheel is reduced by wear. The flank angle qR is also reduced by increasing wear. The rollover S, on the other hand, increases as a result of increasing wear. Furthermore, "hollow wear" HL can also be determined as wear on the basis of the diameter Dm. If considerable wear is detected in the area of the point D0, the term "false flange" or "hollow wear" is used in accordance with standard EN 15313:2016 (D)-§ C.2.14. Hollow wear HL in the form of a false flange is present if the outer tread region is higher than the tread in the measurement circle plane. Hollow wear is formed when considerable wear of the tread occurs in the region of the measurement circle plane. Consequently, on the basis of the measurement according to the invention, it is also possible to estimate the hollow wear, which according to the standard must not be greater than 2 mm. Advantageously, the wear of the wheel profile during travel of the rail vehicle is determined and the rail vehicle does not have to go to a workshop for determination of the wear. A workshop visit is only required when it has been determined during one of the automatic wear measurements that a wear limit has been reached. Advantageously, the number of workshop visits can be reduced in comparison with the conventional procedure, in which the wear of a wheel profile itself is determined in a workshop. With the method according to the invention, the current values which characterize the wheel wear can be continuously estimated and an inspection can be indicated in good time. The method according to the invention for training an artificial intelligence-based algorithm to determine a current value of a geometric parameter of a wheel profile of a rail vehicle is preferably intended to be used in the method according to the invention for automatically determining the travel-related wear of a wheel of a rail vehicle in order to determine the wear on the basis of the determined current diameter of the wheel. In the method according to the invention for training an artificial intelligence-based algorithm to determine a current value of a geometric parameter of a wheel profile of a rail vehicle, labeled training data are provided, which comprise a value of a wheel diameter as input data and a target value of an additional parameter of the wheel profile as output data. Such an additional parameter or its value characterizes the shape of the wheel profile and thus also the rolling properties of the wheel. The wear state of the wheel is characterized by these additional parameters. As already briefly explained, the wear is characterized in particular by the variables flange height Sh, flange thickness Sd, flange flank angle qR, and the rollover S and hollow wear HL. Labeled data should be understood here to mean training data comprising target data which can be compared with the result data of the algorithm when the input data of the training data have been input into the algorithm. The artificial intelligence-based algorithm preferably has an artificial neural network which is subject during the training process to adaptation to the training data used for training. Furthermore, a value of an additional parameter of the wheel profile is determined by the algorithm on the basis of the input data. The algorithm is adapted on the basis of the output data or result data of the algorithm and the output data or the target data of the training data. If an artificial neural network is used to implement the algorithm, the weights of the artificial neural network are updated during the adaptation of the algorithm on the basis of an error function, into which the result of the algorithm and the corresponding target value of the labeled training data are incorporated. The "comparison" is thus carried out in such a way that the target value and the result of the algorithm are used as an input value for the error function during a training step, and the result of the error function decides on the change in the weights of the artificial neural network. The artificial intelligence-based algorithm is trained by the training method to estimate the additional parameters of the wheel profile on the basis of labeled training data, which comprise both measured wheel diameters as input data for the algorithm and target values of an additional parameter of the wheel profile as desired result values of the algorithm. The training method can also be carried out on the basis of the measurements of the additional parameters carried out in the workshop during an inspection, with the result that the algorithm becomes more and more reliable and accurate over time. The wear determination device according to the invention has a section length determination unit for highly accurate determination of a length of a measurement section to be traversed by a rail vehicle. Another part of the wear determination device according to the invention is a revolution number determination unit for determining the required number of revolutions of the wheel for traversal of the measurement section. Such a revolution number determination unit has a revolution counter or angle meter. The wear determination device according to the invention also comprises a diameter determination unit for determining a current diameter of the wheel of the rail vehicle on the basis of the length and the determined number of revolutions of the wheel. In addition, the wear determination device according to the invention comprises a wear determination unit for determining the wear by comparison of the determined current diameter with a reference value. The wear determination device according to the invention shares the advantages of the method according to the invention for automatic determination of the travel-related wear of a wheel of a rail vehicle. A part of the aforementioned components of the wear determination device according to the invention can be implemented entirely or partially in the form of software modules in a processor of a corresponding computing system. A largely software-based implementation has the advantage that even computing systems that have already been used in controlling rail vehicles can be upgraded in a simple manner by means of a software update in order to operate in the manner according to the invention. In this respect, the object is also achieved by a corresponding computer program product having a computer program which can be loaded directly into a computing system, having program sections in order to carry out the steps of the method according to the invention for automatically determining the travel-related wear of a wheel of a rail vehicle, if appropriate using suitable additional measurement sensors, when the program is executed in the computing system. In addition to the computer program, a computer program product of this kind can advantageously optionally comprise additional elements such as documentation and / or additional components including hardware components, such as hardware keys (dongles etc.) for using the software. For transport to the computing system and / or for storage on or in the computing system, it is possible to use a computer-readable medium, for example a memory stick, a hard drive or some other transportable or built-in data medium on which the program sections of the computer program that can be read in and executed by a computing system are stored. For this purpose, the computing system can for example comprise one or more cooperating microprocessors or the like. The dependent claims and the following description each contain particularly advantageous configurations and developments of the invention. In this case, in particular, the claims in one claim category can also be developed analogously to the dependent claims in another claim category and the parts of the description relating thereto. In addition, in the context of the invention, the various features of different exemplary embodiments and claims can also be combined to form new exemplary embodiments. In one variant of the method according to the invention for the automatic determination of the travel-related wear of a wheel of a rail vehicle, the length of the measurement section is measured with high accuracy in advance. As already mentioned, such a procedure is particularly advantageous if access to a differential satellite navigation system is not possible. Alternatively, the length of the measurement section is measured with high accuracy during traversal by means of a differential satellite navigation system. A differential satellite navigation system is suitable for determining the position of a rail vehicle with an error of a few centimeters, and, with optimum signal reception, even of a few millimeters. Particularly precise differential satellite navigation systems additionally evaluate the phase shift of the carrier wave and thus achieve accuracies of a few millimeters. If, therefore, the starting point and the end point of the route section or measurement section lie on a straight section that lies completely within a region covered by a high-quality differential satellite navigation system, it is possible to determine the distance ZS between the starting point and the end point with high accuracy, and this can be made available to an algorithm for calculating the diameter Dm of a wheel. The accuracy of position determination depends on the distance of the receiver of the satellite navigation system from the reference station. As already mentioned, the length of the measurement section should be orders of magnitude greater, preferably a few kilometers long, than the diameter of the wheels of the rail vehicle. This ensures that the number of revolutions is so high that the effects of the transverse movement of the axes, also referred to as "snaking", can be neglected. Preferably, the measurement section is also traversed several times and the measured distances are averaged. In this case, the traversing of the measurement section is thus preferably repeated several times. A statistical mean value of the length of the measurement section is then determined. The measured values of the number of revolutions which deviate from a statistical mean value by more than a predetermined threshold value are discarded. Finally, a mean value of the number of revolutions is determined on the basis of the remaining measured values, and this mean value is used as a basis for determining the statistically corrected current diameter Dm* of the wheel of the rail vehicle. The accuracy of the measurement result is advantageously further improved by averaging and discarding "outliers". This is because the determination of the diameter of a wheel can be impaired by a systematic error, which is mainly due to microsliding and slip of the motor-driven axles. As already explained above, the diameter of the wheel preferably relates to the center of the wheel profile at the point D0 in the center plane of the wheel. Here, the term "point" is intended to refer to a projection or sectional representation of a wheel. In a three-dimensional view, the "point" D0 corresponds to a circle which runs around the wheel profile in the center plane of the wheel. The center of the wheel profile lies approximately in the region in which a wheel of a rail vehicle is in contact with the rail. The wheel diameter Dm at the center of the wheel profile can advantageously be inferred from the number of revolutions of the wheel and the length of the route section or measurement section traversed. More precisely, the "point" D0 can be defined as a small, elliptically shaped region which is slightly stretched in the direction of movement. The geometry of this contact surface varies dynamically during rolling and is strongly influenced by the physical properties of the metals in contact (wheel / rail) and the mutual dynamic interactions due to travel. Preferably, a correction value is determined as a reference value after the replacement or reprofiling of a wheel by a first determination of a first value Dm* of a current diameter directly after the replacement or the reprofiling of the wheel, and a comparison of the current diameter Dm* with the value Dn of the diameter of the wheel given in design documents is carried out. In this case, an offset |Dn - Dm*| of the first value Dm* relative to the value Dn given in design documents is determined as the correction value, if appropriate. It is expedient to use this offset in order to correct the error which could have influenced the estimation of the diameter. This offset, the value of which is calculated once as described, is then applied to all subsequent estimates Dm of the diameter which are carried out with the formula (1) and the statistical correction already mentioned in order to eliminate from this measurement and all further measurements the systematic error which affects the first value Dm*. The first value of a current diameter corrected by this offset is also referred to below as Dm° and corresponds to the value Dn from the design documents. Alternatively, the reference value can also be corrected by the correction value in order to take account of the systematic measurement error. That is to say, in this case, the first value Dm* is used as a reference value and the wear is determined on the basis of the difference between subsequent values Dm of the diameter of the wheel and the first value Dm*. A systematic measurement error can advantageously be compensated. Likewise as a preference, a current value of a diameter corrected by the correction value |Dn - Dm*| is compared with the data entered for the safety systems and travel control systems of the rail vehicle and / or with measurements carried out during the last inspection, and corrections are made to said data if appropriate. Advantageously, further measurement errors that have occurred in connection with measurements that were carried out using methods other than the method according to the invention are likewise detected and corrected, if appropriate. In particular, it is possible in this way to correct safetyrelevant and reliability-relevant errors. Furthermore, a current value of at least one of the following additional geometric parameters of the wheel profile is preferably determined on the basis of the determined current diameter Dm by an artificial intelligence-based algorithm: - the flange thickness Sd, - the flange height Sh, - the flange flank dimension qR, - the rollover S, - the hollow wear HL. After a wear state of the profile of a wheel of a rail vehicle on the rolling strip, i.e. at the point D0 of the profile which is located in the center of the profile in the transverse direction, has been determined with the value of the current diameter Dm, this information can be used to determine the additional parameters of the wheel profile. Advantageously, further wear parameters of the wheel profile are determined, on the basis of which a maintenance strategy can be defined. In this case, the determined operating conditions of the rail vehicle are preferably also used as input data for the algorithm in addition to the current value Dm of the diameter, preferably in addition to the deviation of the current value Dm of the diameter from a reference value Dm°. A correlation between the development of the rollover S on account of wear, in conjunction with the change in other parameters (on account of the same wear), can be determined statistically in the training phase of the algorithm and subsequently refined further during the implementation of the algorithm. This makes the determination of the wear even more precise. The operating conditions preferably comprise at least one of the following types of operating conditions: - the average speed of the rail vehicle, - the curve profile of the sections traveled by the rail vehicle since the last test measurement, - the gradient profile of the sections traveled by the rail vehicle since the last test measurement, - weather conditions during the operation of the rail vehicle, in particular snow or ice, - the general state of the track elements traveled on, - temperatures prevailing during operation. Advantageously, these influencing variables which affect the wear of the wheels can also be taken into account by the algorithm for determining the additional parameters which characterize the wear of the wheels, thereby making the final result more precise and more reliable for determining the wear of the wheels of a rail vehicle. In the following, the invention is explained once again in greater detail with reference to the accompanying figures on the basis of exemplary embodiments. In the drawings: Figure 1 shows a sectional view of a wheel profile of a rail vehicle, Figure 2 shows a conventional arcuate measuring device for indirectly measuring the diameter of a wheel of a rail vehicle by measuring the curvature of the wheel profile in the circumferential direction, Figure 3 shows a flowchart illustrating a method for automatically determining the travel-related wear of a wheel of a rail vehicle, Figure 4 shows a schematic illustration of a wear determination device according to one exemplary embodiment of the invention, Figure 5 shows a flowchart illustrating a method for training an artificial intelligence-based algorithm to determine a current value of a geometric parameter of a wheel profile of a rail vehicle, Figure 6 shows a schematic illustration of a training device according to one exemplary embodiment of the invention. Figure 1 shows a sectional view 10 of a wheel profile of a rail vehicle, wherein the "length" L is plotted against the "height" H of the wheel profile (in each case in mm). The radial extent of the wheel is designated here as the "length" L and the axial extent of the wheel is designated here as the "height" H. More precisely, figure 1 shows two wheel profiles, a wheel profile N (indicated by a solid line) of a new wheel and a wheel profile W (illustrated by dashed lines) of a worn wheel. For the wheel profile N of the new wheel, a flange height Sh is entered as a parameter of the wheel profile, showing the height at the highest point of the flange over the height of the wheel profile in the wheel center at the point D0. Figure 1 also shows the flange thickness Sd. In addition, it also illustrates the flange flank dimension qR, which represents a flank width of the flange. In addition, figure 1 shows a depression or hollow wear HL of the wheel profile of the worn wheel. Moreover, the rollover S is shown on the right of the image in figure 1. The rollover S represents a material redistribution of the wheel material from the center of the wheel profile toward the edge. Figure 2 shows two illustrations 20 of a conventional arcuate measuring device for indirectly measuring the diameter of a wheel of a rail vehicle by measuring the curvature of the wheel profile in the circumferential direction. Figure 3 shows a flowchart 300 which describes a method for automatically determining the travel-related wear of a wheel of a rail vehicle. In step 3.I, a measurement section MS is traversed by the rail vehicle, wherein the length ZS of the rail section or measurement section MS is determined with high accuracy by a differential satellite navigation system. In step 3.II, the number AU of revolutions U of the wheel required to traverse the measurement section is determined. In step 3.III, a current diameter Dm of the wheel of the rail vehicle is determined on the basis of the length ZS and the determined number AU of revolutions of the wheel. In step 3.IV, the determined current diameter Dm is compared with a reference value Dn. In the event that the difference between the two values exceeds a predetermined threshold value, which is denoted by "y" in figure 1, a transition is made to step 3.V and a plurality of wear parameters V of the wheel is determined by an artificial intelligence-based algorithm. In the event that the threshold value is not exceeded in step 3.IV, this being indicated by "n" in figure 3, a transition is made to step 3.I and the method is carried out again after a specific period of time. Figure 4 shows a schematic illustration of a wear determination device 40 according to one exemplary embodiment of the invention. The wear determination device 40 has a section length determination unit 41 for highly accurate determination of a length of a measurement section to be traversed by a rail vehicle on the basis of satellite navigation data SND. Another part of the wear determination device 40 is a revolution number determination unit 42 for determining the number AU of revolutions U of the wheel required to traverse the measurement section. The wear determination device 40 comprises a diameter determination unit 43 for determining a current diameter Dm of the wheel of the rail vehicle on the basis of the length ZS and the determined number AU of revolutions of the wheel. The wear determination device 40 also has a wear determination unit 44 for determining the wear V by comparing the determined current diameter Dm with a reference value Dn. Figure 5 shows a flowchart 500 which illustrates a method for training an artificial intelligence-based algorithm to determine a current value of a geometric parameter of a wheel profile of a rail vehicle that characterizes the wear V of a wheel. In step 5.I, labeled training data TD are provided, comprising as input data a value Dm of a wheel diameter and as output data a target value of an additional parameter Sd, Sh, qR, S, HL of the wheel profile that characterizes the wear V of a wheel. In step 5.II, a value Dm of a diameter of a wheel is input as the input value for the algorithm ALG. On the basis of the input data Dm, the algorithm ALG furthermore determines an additional parameter ZP of the wheel profile as an output value. In step 5.III, the algorithm ALG is adapted on the basis of the value of the determined additional parameter ZP and the corresponding value ZPL of this additional parameter, which is assigned to the labeled training data TD. In step 5.IV, a difference AZ between the determined additional parameter ZP and the corresponding parameter ZPL of the labeled training data TD is determined. In step 5.V, it is determined whether the difference AZ falls below a predetermined threshold value SW. If the threshold value SW is undershot, this being indicated by "y" in figure 5, a transition is made to step 5.VI and the algorithm for calculating additional wear parameters ZP is made available as part of the method in order to automatically determine the travel-related wear of a wheel of a rail vehicle. In the event that the threshold value SW is exceeded, this being indicated by "n" in figure 5, a transition is made to step 5.I and the algorithm ALG is further adapted to the training data TD. Figure 6 illustrates a schematic representation of a training device 60 according to one exemplary embodiment of the invention. The training device 60 has an AI unit 61, which is configured to determine additional parameters ZP on the basis of input data Dm from labeled training data TD. The additional parameters ZP and the additional parameters ZPL assigned to the labeled training data TD are transmitted to an adaptation unit 62, which adapts the algorithm ALG of the AI unit 61 to the received data ZP, ZPL. The algorithm ALG is based, for example, on an artificial neural network with weights which are adapted to the target values. The received data ZP, ZPL are also transmitted to a difference forming unit 63, which determines a difference value AZP therefrom. Another part of the training device 60 is a test unit 64, which, on the basis of the difference value AZP, checks whether the algorithm ALG has already been sufficiently adapted to the training data TD. If this is the case, the algorithm is provided for the calculation of additional parameters ZP. Otherwise, the algorithm ALG is sent back to the AI unit 61 and the adaptation process is continued. In conclusion, it is once again noted that the above-described methods and apparatuses are merely preferred exemplary embodiments of the invention and that the invention can be varied by a person skilled in the art without departing from the scope of the invention as specified by the claims. For the sake of completeness, it is also noted that the use of the indefinite article “a”, “an” or “one” does not exclude the features in question also being present in a plurality. Likewise, the term “unit” does not exclude this consisting of a plurality of components, which can optionally also be spread out in space. Irrespective of the grammatical gender of any particular term, persons identifying as male, female or any other gender are also included.

Claims

Patent claims1. A method for automatically determining the travel-related wear of a wheel of a rail vehicle, comprising the steps of:- traversing a measurement section (MS) by the rail vehicle, the length (ZS) of which is known with high accuracy or is determined with high accuracy,- determining the required number (AU) of revolutions (U) of the wheel for traversal of the measurement section (MS),- determining a current diameter (Dm) of the wheel of the rail vehicle on the basis of the determined length (ZS) of the measurement section (MS) and the determined number (AU) of revolutions of the wheel,- determining the wear (V) of the wheel on the basis of a comparison of the determined current diameter (Dm) with a reference value (Dn, Dm°) of the diameter.

2. The method as claimed in claim 1, wherein the length (ZS) of the measurement section (MS) is measured with high accuracy in advance.

3. The method as claimed in claim 1, wherein the length (ZS) of the measurement section (MS) is measured with high accuracy by means of a differential satellite navigation system during the traversal of the measurement section (MS).

4. The method as claimed in claim 3, wherein- the traversing of the measurement section (MS) is repeated several times,- a statistical mean value of the length (ZS) of the measurement section (MS) is determined,- measured values of the number (AU) of revolutions (U) which deviate from a statistical mean value by more than a predetermined threshold value (SW) are discarded,- a mean value of the number (AU) of revolutions (U) is determined on the basis of the remaining measured values, andthis mean value is used as a basis for determining the current diameter (Dm) of the wheel of the rail vehicle.

5. The method as claimed in one of the preceding claims, wherein the current diameter (Dm) of the wheel relates to the center (D0) of the wheel profile.

6. The method as claimed in one of the preceding claims, wherein a correction value is determined as a reference value after the replacement or reprofiling of a wheel by a first determination of a first value (Dm*) of a current diameter directly after the replacement or the reprofiling of the wheel and by a comparison of the first value (Dm*) with the value (Dn) of the diameter of the wheel given in design documents, and this correction value is taken into account in the subsequent measurement of a current diameter (Dm) in order to determine the current wear (V).

7. The method as claimed in claim 6, wherein a value of the diameter of the wheel corrected by the correction value is compared with the data entered for the safety systems and travel control systems of the rail vehicle and / or with measurements carried out during the last inspection, and corrections are made to said data if appropriate.

8. The method as claimed in one of the preceding claims, wherein a current value of at least one of the following additional geometric parameters (Sd, Sh, qR, S, HL) of the wheel profile is determined on the basis of the determined current diameter (Dm) by an artificial intelligence-based algorithm:- the flange thickness (Sd),- the flange height (Sh),- flange flank dimension (qR),- the rollover (S),- the hollow wear (HL).

9. The method as claimed in claim 8, wherein, in addition to the current diameter (Dm), the determined operating conditions of the rail vehicle are also used as input data for the algorithm.

10. The method as claimed in claim 9, wherein the operating conditions comprise at least one of the following types of operating conditions:- the average speed of the rail vehicle,- the curve profile of the sections traveled,- the gradient profile of the sections traveled,- weather conditions during the operation of the rail vehicle,- the state of the track elements traveled on,- temperatures prevailing during operation.

11. The method as claimed in one of claims 8 to 10, in which the artificial intelligence-based algorithm is trained to determine a current value of a geometric parameter (Sd, Sh, qR, S, HL) ofa wheel profile of a rail vehicle, comprising the steps of:- providing labeled training data (TD) which comprise a value (Dm) of a wheel diameter as input data and a target value (ZPL) of an additional parameter (Sd, Sh, qR, S, HL) of the wheel profile as output data,- determining a value (ZP) of an additional parameter (Sd, Sh, qR, S, HL) of the wheel profile as output data by means of the algorithm (ALG) on the basis of the input data,- adapting the algorithm (ALG) on the basis of the output data (ZP) of the algorithm (ALG) and the output data (ZPL) of the labeled training data (TD).

12. The method as claimed in claim 11, wherein the input data comprise operating conditions that have occurred during use of the rail vehicle.

13. A wear determination device (40), having:- a section length determination unit (41) for highly accurate determination of a length (ZS) of a measurement section (MS) to be traversed by a rail vehicle,- a revolution number determination unit (42) for determining the required number (AU) of revolutions (U) of the wheel for traversal of the measurement section (MS),- a diameter determination unit (43) for determining a current diameter (Dm) of the wheel of the rail vehicle on the basis of the length (ZS) of the measurement section (MS) and the determined number (AU) of revolutions (U) of the wheel,- a wear determination unit (44) for determining the wear (V) of the wheel by comparison of the determined current diameter (Dm) with a reference value (Dn, Dm°) of the diameter.

14. A computer program product having a computer program which can be loaded directly into a memory unit of a computing system, having program sections in order to carry out a method as claimed in one of claims 1 to 12 when the computer program is executed in the computing system.

15. A computer-readable medium on which program sections executable by a computer unit are stored in order to carry out a method as claimed in one of claims 1 to 12 when the program sections are executed by the computer unit.

Images