Method for detecting the vehicle category of a railway vehicle and device suitable for use of the method

Abstract

The invention relates to a method for identifying the characteristics of a rail vehicle (FZ), in which an axle counter (AZ1 ... AZ2) records measurement data during the passage of the rail vehicle (FZ). The measurement data is analyzed by a computer to determine the axle spacing of the rail vehicle (FZ). Based on the determined axle spacing, a first characteristic is determined by the computer, namely whether the rail vehicle (FZ) is a passenger train or a freight train. In a first test step, the determination of this first characteristic is checked to see if identical or similar patterns can be found in at least a predominant part of the sequence of axle spacings. If no pattern can be found, the rail vehicle (FZ) is assigned the characteristic of a freight train as its first attribute.Once a pattern has been identified, the railway vehicle (FZ) is assigned the characteristic of a passenger train as its first property, and / or a further test step is carried out to validate or extend the test result. This process can be supported by machine learning. Furthermore, the invention comprises a device for determining the properties of railway vehicles (FZ), a computer program, and a delivery device.

Description

[0001] The invention relates to a method for recognizing properties of a rail vehicle, in which An axle counter (meaning at least one axle counter, possibly also several axle counters) records measurement data during the passage of the rail vehicle, the measurement data is analyzed by computer and the axle spacing of the rail vehicle is determined, and a first characteristic is determined by computer based on the determined axle spacing, namely whether the rail vehicle is a passenger train or a freight train.

[0002] Furthermore, the invention relates to a device for determining the properties of rail vehicles, comprising at least one axle counter for recording measurement data when the rail vehicles pass over, a computer that is set up to analyze the measurement data and thereby determine axle spacings of the rail vehicle, and to determine, based on the determined axle spacings, a first characteristic as to whether the rail vehicle is a passenger train or a freight train.

[0003] Finally, the invention relates to a computer program product and a delivery device for this product.

[0004] Computer program product, wherein the computer program product is equipped with program instructions for carrying out this procedure.

[0005] Traditional train control systems do not recognize many train characteristics, such as the train type (e.g., freight train, regional train, locomotive), but only logical properties, such as whether a track occupancy detection section is present. Operational control systems, on the other hand, do recognize this train type and potentially other characteristics, but these cannot usually be used as a basis for safe decisions because they themselves do not provide the necessary level of safety. Nevertheless, in certain operational situations, it is essential for train operation to determine the train type with the required level of safety. This is currently achieved through a combination of technical and operational procedures, sometimes requiring considerable effort.

[0006] Train categories, called "Zugkategorie" in Switzerland, are categories of different railway trains. Trains are classified according to their use, their importance to traffic, and their operational handling. Each train is identified by its train category and a train number.

[0007] The designations for train categories vary; in addition to colloquial terms, there are also technical designations, namely terms from transport science, designations originating from regulations of the state railway era, and brand names of railway undertakings. Regardless of which train categories are used, however, they allow for more precise information about the trains in operation. This information can, for example, be stored in a railway automation system and used for control tasks. Document EP 2 718 168 B1 concerns a method for operating a railway signaling system with at least one trackside device, taking into account a speed measurement recorded when the rail vehicle enters the activation section of the railway signaling system.When the rail vehicle enters the activation section, the system checks, based on the speed measurement, whether a correction time needs to be set for forwarding a message from one trackside device to an associated railway safety device, according to the speed measurement. Subsequently, the set correction time is checked to ensure it remains effective regardless of at least one other factor influencing the rail vehicle's travel time.

[0008] KIEFFER EBERHARD ET AL: "Mixed Traffic - Better Utilization of High-Speed ​​Lines with Many Tunnels", DEINE BAHN, December 2010, pages 43-47, XP055788752, describes high-speed lines (SFS) of DB Netz AG which, due to their track characteristics, are generally suitable for mixed traffic (for passenger and freight trains). However, on the tunnel-heavy sections of these high-speed lines, there is a prohibition on passing between freight and passenger trains. For the SFS between Fulda and Burqsinn, which is also used by freight trains, this was made possible by the introduction of a day and night window.

[0009] HIEBENTHAL T ET AL: "Conflict warning system - detection of unauthorized encounters of passenger and freight trains", SIGNAL UND DRAHT Vol. 103, No. 11, November 2011, pages 16-19, XP001569642, describes a method for identifying the train type based on the recorded axle pattern.

[0010] For this purpose, wheel sensors are attached to the track, which are used to detect the axle patterns and other characteristics of the passing train. The detected axle pattern is compared with patterns from an axle pattern database, in which each vehicle that regularly travels the line is assigned an axle pattern and a train type. After identifying all vehicles of the train, the train type of the entire train is then determined. If a train passes that contains vehicles whose axle patterns are not stored in the database, an unrecognized train type is reported for this train. Furthermore, the publication of patent application CN 1 378 935 A describes essentially the same method as that of Hiebenthal et al.

[0011] The object of the invention is to reliably identify the train type, ideally without the need for additional sensors, so that the identified train type can be used in railway signaling systems. To this end, a method and a device suitable for applying the method are to be provided. Furthermore, the invention aims to provide a computer program and a delivery device for this computer program with which the aforementioned method can be carried out.

[0012] This problem is solved according to the invention with the subject matter of the claim (method) specified at the outset by checking, in a first test step when determining the first property, whether identical or similar patterns can be found in at least a predominant part of the sequence of axis distances and If no pattern could be identified, the railway vehicle is assigned the property of a freight train as its first characteristic, or if a pattern was identified, the railway vehicle is assigned the property of a passenger train as its first characteristic and / or a further test step is carried out.

[0013] In other words, the invention provides for estimating the distances between the axles of the train consist from the raw data of the axle counter during the crossing. Passenger trains, such as ICE or regional trains, in particular, consist of fixed units that generally remain together, without any cars being uncoupled. Therefore, patterns exist that can be measured repeatedly by coupling these units and that are similar to one another. Passenger trains thus have, so to speak, a fixed "fingerprint" that is only altered by measurement errors, etc.

[0014] The advantage of using pattern recognition in train operations lies in the fact that operational parameters, such as level crossing closing times or track clearances, can be flexibly adapted to the vehicles assigned to them based on the pattern recognition of the first characteristic. This allows, for example, greater track utilization. Another example is the reliable detection of hazard hotspots, for which safety measures can be initiated.

[0015] In contrast, freight trains exhibit different, usually variable, and therefore not similar or identical patterns, depending on which and how many units are coupled together. This allows the data to be represented as multidimensional vectors whose components estimate the distances between the axes, i.e., axis 1 to axis 2 up to axis n-1 to axis n (for n axes in the train, up to 250 in reality).

[0016] The terms "identical" and "similar" should be understood in the context of pattern recognition. This means that a comparison of patterns can lead to them being assessed as identical or similar (or not identical and not similar, i.e., not related). This assessment is preferably computer-aided.

[0017] Patterns are considered identical if all test criteria, when compared to the pattern, result in a match. Since the test criteria are based on measured values, a tolerance interval can be defined for the measurement, within which the test criterion must fall to be considered identical.

[0018] Patterns are considered similar if an evaluation of the test criteria shows that they correspond to each other, at least to a large extent. It should be noted that similarity also exists if the patterns are identical.

[0019] The criteria for determining when the criteria correspond, at least largely, must be defined for pattern recognition to be carried out. Generally, the following relationship applies to the recognition of the aforementioned first characteristic (freight train or passenger train): The stricter the criteria for recognizing similarity, the greater the probability that the recognized similar patterns will always lead to the identification of passenger trains. However, the probability that passenger trains will not be recognized also increases. Conversely, the less strict the criteria for recognizing similarity, the higher the probability that all passenger trains will be recognized. However, the probability that freight trains will be mistakenly identified as passenger trains also increases.

[0020] Regardless of the stringency of the criteria, the method according to the invention functions in a technical sense. However, in practice, it is necessary to determine where the optimum lies with regard to the stringency of the criteria in relation to safe operation (more on defining the criteria below).

[0021] It should be noted that the fact that no at least similar patterns are recognized leads to the assessment of the first characteristic as that of a freight train. It is worth noting that freight trains are more critical to assess with regard to train operations. For example, a tunnel meeting between two passenger trains whose external dimensions can be reliably determined would likely be permitted, but not a tunnel meeting between two freight trains or between a freight train and a passenger train. Another example is level crossings. Due to its slow speed, a freight train will require longer closing times at the level crossing than would be necessary for a passenger train.

[0022] It can be deduced from this that the procedure must be configured so that, in cases of doubt, the train should be assumed to be a freight train. For the pattern recognition procedure, this means that train operation can be carried out with a high safety standard (e.g., SIL 4) if the criteria are interpreted strictly. This means that under no circumstances should a freight train be identified as a passenger train. On the other hand, it is less critical if a passenger train is not identified based on its characteristic patterns, since its treatment as a freight train is generally unproblematic in train operation.

[0023] A predominant proportion of the sequence of axle spacings is present if the repeating pattern can be observed for more than 50% of the axle spacings. Preferably, it can also be defined that the threshold at which a predominant proportion is assumed is more than 60%, particularly preferably more than 70%, 80% or 90%.

[0024] In the context of the invention, "computer-aided" or "computer-implemented" can be understood as an implementation of the method in which at least one computer or processor performs at least one process step of the method.

[0025] The term "computer" covers all electronic devices with data processing capabilities. Computers can include, for example, personal computers, servers, handheld computers, mobile phones, and other communication devices that process data using a computer system, as well as processors and other electronic devices for data processing, which may preferably also be connected to a network.

[0026] In the context of the invention, a "processor" can be understood to mean, for example, a converter, a sensor for generating measurement signals, or an electronic circuit. In particular, a processor can be a central processing unit (CPU), a microprocessor, a microcontroller, or a digital signal processor, possibly in combination with a memory unit for storing program instructions, etc. A virtualized processor or a soft CPU can also be understood as a processor.

[0027] In the context of the invention, a "storage unit" can be understood to mean, for example, a computer-readable memory in the form of a working memory (Random-Access Memory, RAM) or data storage device (hard drive or data carrier).

[0028] "Interfaces" can be implemented in hardware, for example via wired or wireless connections, and / or in software, for example as interaction between individual program modules or program parts of one or more computer programs.

[0029] The term "cloud" refers to an environment for "cloud computing" (also known as a computer cloud or data cloud). It describes an IT infrastructure that is made available via network interfaces such as the internet. This typically includes storage space, computing power, or software as a service, without requiring installation on the local computer using the cloud. The services offered within the framework of cloud computing encompass the entire spectrum of information technology and include, among other things, infrastructure, platforms, and software.

[0030] The term "program modules" refers to individual functional units that enable the program flow according to the invention. These functional units can be implemented in a single computer program or in several communicating computer programs. The interfaces implemented in this way can be implemented in software within a single processor or in hardware if multiple processors are used.

[0031] Unless otherwise specified in the following description, the terms "create," "determine," "calculate," "generate," "configure," "modify," and the like primarily refer to processes that create and / or modify data and / or convert data into other data. The data is primarily in the form of physical quantities, such as electrical impulses or measured values. The necessary instructions (program commands) are compiled into a computer program, which is software. Furthermore, the terms "send," "receive," "read," "extract," "transmit," and the like refer to the interaction of individual hardware components and / or software components via interfaces.

[0032] According to one embodiment of the invention, it is provided that in the first test step, a number of axle distances at the beginning of the sequence and / or a number of axle distances at the end of the sequence are disregarded.

[0033] By disregarding a number of axle spacings at the beginning or end of a sequence, it can be advantageously avoided that locomotives or power cars, which, for example, form the beginning or end of a passenger train, are not checked for the presence of axle spacing patterns. This is because both the locomotives and often the power cars have different axle spacings (and are therefore characterized by different patterns) than the vehicles in the middle of the train, which are generally identical in a passenger train and therefore form similar or identical patterns. The first check can therefore be carried out more quickly and reliably if it is limited specifically to the middle section of the train.

[0034] The number of axles to be disregarded depends on the train operation being monitored. If the locomotives or power cars used are known, the sequence of axle spacings to be disregarded corresponds to that of the locomotives or power cars used. However, even with unknown locomotives and power cars, a general value can be assumed. This could be, for example, four, six, or eight axles.

[0035] According to one embodiment of the invention, it is provided that in this or a further test step the magnitude of the axle spacing is determined, wherein the rail vehicle is assigned the property of a passenger train as a first property only if the magnitude of the largest axle spacing occurring in the model exceeds a defined limit value.

[0036] The threshold value that reliably identifies passenger trains depends, not least, on the specific characteristics of the train operation being monitored. This threshold value can therefore be determined on a route-specific basis if it is known which passenger trains operate on the route in question. It is important that the longest axle spacing of the relevant train cars is taken into account. However, if train cars with different axle spacings are used, the shortest of these axle spacings must be used as the threshold value.

[0037] According to a particularly advantageous embodiment of the invention, this limit value can also differ from typical axle spacings of freight wagons, so that the axle spacing can be used as a particularly reliable criterion for distinguishing between freight wagons. Since the axle spacing is used as an additional criterion to the patterns to be identified (i.e., supplementarily), compliance with this difference is not mandatory.

[0038] According to one embodiment of the invention, it is provided that in the first or a further test step the determined patterns of axle spacings are compared with reference patterns of axle spacings and, in the event of a recognized match of the pattern with a reference pattern, a train category linked to the reference pattern is assigned to the rail vehicle as a second property.

[0039] The reference patterns can be stored, for example, in a storage device. A server can provide the reference patterns to enable comparison with the detected patterns. Alternatively, the reference patterns can be stored in a storage device that forms part of the axle counter. This allows the axle counters to be technically modified with a certain degree of intelligence, in other words, as autonomous or partially autonomous units.

[0040] The advantage of storing reference patterns in a memory device is that they are always available and can be retrieved without delay when needed. The memory devices can also store the various reference patterns for specific routes, so that only certain reference patterns are available to specific axle counters on specific sections of track.

[0041] According to one embodiment of the invention, it is provided that the axle counter can be used to determine further properties of the rail vehicle, in particular the direction of travel of the train and / or the speed of the train when passing over an axle and / or the average speed when passing over and / or the acceleration when passing over and / or the wheel diameter.

[0042] Besides recording the number of axles, the axle counter, due to its standard dual-sensor design, is also fundamentally suitable for determining other data such as those mentioned above. Furthermore, it is relatively easy to supplement the axle counter with simple sensors that, for example, measure the axle load during transit.

[0043] The following measurement principles can be used as examples. Direction of travel of the train: by comparing the influence of dual sensors (e.g., by evaluating the time offset during signal generation). Speed ​​of the train when passing an axle: from the distance between the dual sensors, e.g., by evaluating the time offset during signal generation or the time interval between passing the estimated wheel centers with known axle spacing. Average speed during the passage and / or the acceleration during the passage: from averaging over different wheels or numerical derivation of the speed. Wheel diameter: from the duration of the influence on the axle counter.

[0044] From a parameter set determined in this way, a suitable parameter set for the application is advantageously selected. For example, the wheel diameter can be omitted if it is more or less the same for all trains on the line. For the parameters under consideration, location-specific, representative data are now collected or measured and classified, e.g., passenger train, freight train. This involves a finite number of integer or real-valued measurements; for example, these could be the speed and the number of axles, to give a clear two-dimensional example. In principle, this results in a classification problem, as described below. Figure 5 described.

[0045] Overall, collecting additional parameters beyond those used for comparison makes the detection of vehicle characteristics more robust against errors. This can advantageously lead to a higher degree of reliability in train detection, enabling more effective train traffic control. Which parameters should be considered for a given train traffic control task depends on the specific circumstances. They must be selected appropriately when designing the control procedure.

[0046] According to one embodiment of the invention, it is provided that the criteria for the first test step and / or the further test steps are evaluated using a machine learning method.

[0047] Machine learning offers the advantage of optimizing operational processes, specifically the reliable identification of train characteristics, particularly train types, during operation. This allows the system to automatically adapt to changing operating conditions. For example, additional patterns can be generated when a new type of passenger train is deployed on a particular section of track. Neural networks or other artificial intelligence systems can be used for this purpose.

[0048] In the context of this invention, artificial intelligence (hereinafter also abbreviated as AI) refers specifically to computer-aided machine learning (hereinafter also abbreviated as ML). This involves the statistical learning of algorithm parameterization, preferably for complex applications. Using ML, the system recognizes and learns patterns and regularities in the acquired process data based on previously inputted training data. With the aid of suitable algorithms, ML can independently find solutions to emerging problems. ML is divided into three areas: supervised learning, unsupervised learning, and reinforcement learning, with more specific applications such as regression and classification, structure recognition and prediction, data generation (sampling), or autonomous action.

[0049] In supervised learning, the system is trained by observing the relationship between input and corresponding output of known data, thereby learning approximate functional relationships. The availability of suitable and sufficient data is crucial, because if the system is trained with unsuitable (e.g., non-representative) data, it will learn incorrect functional relationships. In unsupervised learning, the system is also trained with example data, but only with input data and without a connection to a known output. It learns how to form and extend data groups, what is typical for the application, and where deviations or anomalies occur. This allows for the description of use cases and the detection of errors.In reinforcement learning, the system learns through trial and error by proposing solutions to given problems and receiving positive or negative feedback on these proposals. Depending on the reward mechanism, the AI ​​system learns to perform corresponding functions.

[0050] Machine learning can be performed, for example, using artificial neural networks (ANNs). Artificial neural networks are usually based on the interconnection of many neurons, such as McCulloch-Pitts neurons or slight modifications thereof. In principle, other artificial neurons can also be used in ANNs, such as the high-order neuron. The topology of a network (the mapping of connections to nodes) must be determined depending on its task. After the construction of a network, the training phase follows, in which the network "learns." A network can learn using methods such as: Developing new connections; deleting existing connections; changing the weights (of neuron j to neuron i); adjusting the thresholds of neurons, if they have thresholds; adding or deleting neurons; modifying activation, propagation, or output functions.

[0051] Furthermore, the learning behavior changes when the activation function of the neurons or the learning rate of the network is altered. In practice, an ANN learns primarily by modifying the weights of the neurons. Threshold adjustments can be handled concurrently by an "on" neuron. This enables ANNs to learn complex nonlinear functions via a learning algorithm that attempts to determine all parameters of the function from existing input and desired output values ​​through an iterative or recursive approach. ANNs are a realization of the connectionist paradigm, as the function consists of many simple, similar parts. Only their sum totals create complexity in the behavior.

[0052] According to one embodiment of the invention, it is provided that probability densities for the properties are determined from the measurement data of a large number of measurements.

[0053] Knowledge of the probability densities makes it possible to define classification limits for assigning the properties. The method is advantageously very robust with respect to the classification limits, because for the comparatively low-dimensional problems according to the invention, the probability densities for the two classes can be estimated from the data (e.g., using density estimation of the measurement results), and thus the error probabilities for an incorrect classification can also be determined.

[0054] According to one embodiment of the invention, the method is applied to determine the characteristics of the rail vehicle as it approaches a hazard on the track intended for it.

[0055] A hazard point could be a place on the track where either the rail vehicle is potentially at risk (for example, a tight curve, bridge, tunnel) or a place where the rail vehicle potentially endangers others (for example, a track construction site or a level crossing).

[0056] The method according to the invention can therefore also be advantageously applied selectively to mitigate hazardous areas. In other words, train traffic at hazardous areas can be carried out more reliably while maintaining greater flexibility. In particular, trains can be routed through a hazardous area at different speeds depending on their characteristics. This means, for example, that passenger trains can pass the hazardous area at a higher speed than freight trains. This enables smoother passenger train traffic. Passengers thus reach their destinations sooner.

[0057] According to one embodiment of the invention, the method is applied to rail vehicles approaching a level crossing, wherein the closing time of the level crossing is selected depending on the identified rail vehicle once the characteristics of the approaching rail vehicle have been determined.

[0058] The major advantage of applying this method to level crossings, which are considered hazardous locations, lies in the fact that the closing times of the level crossing can be advantageously adjusted individually depending on the characteristics of the approaching train. At least when the train's characteristics can be reliably identified, the closing time can often be shortened without compromising safety standards in the operation of the level crossing. Crossing traffic benefits, as it often has to wait less time at the level crossing.

[0059] According to one embodiment of the invention, it is provided that a data pool is used to determine the closing time, in which closing times are linked to the determinable properties of the rail vehicles, in particular train categories.

[0060] The data pool can be determined deterministically and / or created and / or further developed during operation using the machine learning methods already described above. Once the data is available in the data pool, it can be used advantageously with short access times. During operation, the data in the data pool can be further optimized, thus increasingly streamlining train operations.

[0061] According to one embodiment of the invention, it is provided that in the event that the characteristics of the rail vehicle could not be determined, a standard closing time for the level crossing is selected.

[0062] According to the invention, the standard closing time for the level crossing is defined as the closing time that reliably prevents any danger to crossing vehicular and pedestrian traffic, regardless of the characteristics of the trains operating on the line. Slow-moving freight trains are critical in this regard, as they take the longest to travel from the trigger point of the track safety system to the level crossing and therefore require the longest closing time. This can thus be defined as the standard closing time.

[0063] The advantage of using the standard closing time is that, on the one hand, it ensures the safe operation of the level crossing without exception, and on the other hand, it allows for flexible adjustment of the closing times if the characteristics of the approaching train can be determined with sufficient reliability.

[0064] The aforementioned problem is alternatively solved according to the invention with the subject matter of the claim (device) specified at the outset by the fact that the computer is configured to check, in a first test step when determining the first property, whether a repeating pattern can be found at least in a predominant part of the sequence of determined axial distances and If no pattern could be identified, the first characteristic of the rail vehicle should be assigned the characteristic of a freight train, or if a pattern was identified, the first characteristic of the rail vehicle should be assigned the characteristic of a passenger train and / or a further examination step should be carried out.

[0065] The device offers the advantages already explained in connection with the method described in more detail above. The statements made regarding the method according to the invention also apply accordingly to the device according to the invention.

[0066] Furthermore, a computer program product with program commands for carrying out the said inventive method and / or its embodiments is claimed, wherein the inventive method and / or its embodiments can be carried out by means of the computer program product.

[0067] Furthermore, a provisioning device for storing and / or providing the computer program product is required. The provisioning device is, for example, a storage unit that stores and / or provides the computer program product. Alternatively and / or additionally, the provisioning device is, for example, a network service, a computer system, a server system, in particular a distributed, for example cloud-based, computer system and / or virtual computer system, which preferably stores and / or provides the computer program product in the form of a data stream.

[0068] The provision of the computer program product takes the form of a program data block as a file, in particular as a download file, or as a data stream, in particular as a download data stream. This provision can also, for example, take the form of a partial download consisting of several parts. Such a computer program product is, for example, read into a system using the provisioning device, so that the method according to the invention is executed on a computer.

[0069] Further details of the invention are described below with reference to the drawing. Identical or corresponding drawing elements are each provided with the same reference numerals and are only explained more than once to the extent that differences arise between the individual figures.

[0070] The exemplary embodiments described below are preferred embodiments of the invention. In these exemplary embodiments, the described components each represent individual features of the invention that can be considered independently of one another. Each of these features further develops the invention independently and can therefore be considered part of the invention individually or in a combination other than that shown. Furthermore, the described components can also be combined with the features of the invention described above. They show:

[0071] Figure 1 A schematic embodiment of the device according to the invention, showing its interactions. Figures 2 and 3 schematically partly identical or similar patterns of axle spacing for a passenger train and a freight train, Figure 4An embodiment of the method according to the invention as a flowchart, wherein the functional units and interfaces are shown according to Figure 1 are indicated by example, Figure 5 This is symbolic of two normal distributions for the measured data, but in principle it works for all distributions.

[0072] In Figure 1 The diagram depicts a track system with one track GL, a control center LZ, and a signal box SW. A vehicle FZ, in the form of a train, is approaching a level crossing BU on track GL. A first axle counter AZ1 and a second axle counter AZ2 are installed on track GL, configured in a known manner to count the axles of vehicle FZ.

[0073] The axle counter AZ1 is connected to the SW signal box, specifically to a CP computer located within the signal box, via a first interface S1, and the second axle counter AZ2 via a second interface S2. The CP computer also has a third interface S3 for the BU level crossing. Furthermore, the CP computer is connected to a storage unit SE via a sixth interface S6.

[0074] The SW signal box has a first antenna system A1, the LZ control center a second antenna system A2, and the FZ vehicle a third antenna system A3. This enables communication between the SW signal box and the LZ control center via a fourth interface S4, as well as communication between the FZ vehicle and the LZ control center via a fifth interface S5. The fourth interface S4 and the fifth interface S5 are therefore radio interfaces. The first interface S1, the second interface S2, and the third interface S3 can be either wired or radio interfaces; however, the antenna technology required for radio interfaces is not shown.

[0075] As the vehicle FZ moves along track GL towards the level crossing BU, the axles of the vehicle FZ first pass the second axle counter AZ2 and then the first axle counter AZ1. The recorded measured values ​​can be transmitted to the computer CP via the first interface S1 and the second interface S2, the computer CP being configured to carry out the method according to the invention. The computer CP can also directly control the level crossing BU. Another possibility is that the computer CP can communicate with another computer (in the third interface S3) via the third interface S3. Figure 1 (not shown) is connected, which is used via another interface to control the level crossing BU.

[0076] In Figure 2 is a vehicle FZ on track GL according to Figure 1A moving passenger train PZ is depicted. This passenger train PZ consists of a locomotive LK, several passenger cars PW and a power car TK at the end of the passenger train PZ opposite the locomotive LK.

[0077] Furthermore, the axle spacings between the individual axles (indicated by wheels) are shown schematically. It is evident that several axle spacings occur multiple times in the passenger train PZ, allowing the sequence of axle spacings to be examined for patterns. The axle spacings are labeled with the capital letters A to G. The sequence of axle spacings is FFEFFGABACABACABACADA.

[0078] If we disregard the locomotive LK and the power car TK, since their axle spacings differ from those of the passenger cars PW, the successive passenger cars, which are identical in construction, exhibit a repeating sequence of axle spacings. These form a pattern MT, which is indicated by a curly bracket for the passenger car PW following the locomotive LK. The sequence of axle spacings in the Figure 2 The MT pattern shown is ABAC. This sequence of axle spacings also applies to the two following passenger cars.

[0079] The situation is different in the Figure 3The depicted freight train GZ on track GL consists of a locomotive LK and a first freight car GW1, a second freight car GW2, and a third freight car GW3. These have different lengths and numbers of axles, resulting in several different axle spacings, which are labeled with the capital letters A to H. Figure 3 It becomes clear that no repeating patterns can be discovered in the depicted sequence ABACDEDFGFH, which allows the conclusion that it is a freight train.

[0080] In Figure 4 Figure 1 illustrates how the method according to the invention can proceed. First, it is started in a first step, START. This is followed by a measurement step MS by the respective axle counters AZ1, AZ2 (see Figure 2). Figure 1 This measurement step is followed by a first test step PS1, in which the sequence of axle distances (as shown) is checked. Figure 2 and Figure 3The pattern MT (as described) can be determined and verified. Either it is possible to recognize the pattern MT in the sequence of axle spacings, or it is not. In a subsequent query step, GZ,PZ?, it is checked whether the sequence of axle spacings (by finding patterns) indicates a freight train GZ or a passenger train. If this is not the case, a standard closing time SZS for the level crossing BU is output to the storage device SE. A separate memory area is reserved for this purpose in the storage device, which a controller (for example, the computer CP or a unit in the Figures 1 to 3 (another computer not shown in detail) of the level crossing can access it to retrieve the currently stored closing time.

[0081] Once the first characteristic, i.e., whether it is a freight train (GZ) or a passenger train (PZ), has been determined, the computer CP performs a further query step (GZ?) to check if it is a freight train (GZ). If so, a first calculated closing time (SZ1) is transferred to the storage unit SE (replacing the previously stored closing time). If it is not a freight train or if there is no clear result, the computer CP performs a second check (PS2).

[0082] The second test step, PS2, serves to determine the magnitudes of the axle distances. In a subsequent test step |A| <GW kann daher gefragt werden, ob die ermittelten Beträge der Achsabstände kleiner eines für Güterwagen GW typischen Grenzwerts sind. Ist dies der Fall, handelt es sich um einen Güterzug GZ, sodass an die Speichereinheit SE die erste Schließzeit SZ übergeben werden kann (ersetzt vorher abgespeicherte Schließzeit). Ist dies nicht der Fall, wird im Computer CP ein dritte Prüfschritt PS3 angestoßen.

[0083] In the third test step, PS3, reference patterns RMT are loaded from the storage unit SE. The axle spacings and their values ​​are then compared with the reference patterns. In a test step MT=RMT, it can be checked whether the determined patterns MT correspond to a reference pattern RMT. If this is not the case, a second closing time, SZ2, is transferred to the storage unit SE (replacing the previously stored closing time), which can be understood as a standardized closing time for passenger trains PZ. However, if a pattern MT has been recognized, a third closing time, SZ3, can be transferred to the storage unit SE (replacing the previously stored closing time), which is individually tailored to the reference pattern RMT. This individual third closing time, SZ3, may, for example, already have been stored in the storage unit along with the reference patterns RMT, so that its transfer to the aforementionedseparate storage area of ​​the storage unit SE based on the data already available in the storage unit SE.

[0084] Should the query step MT=RMT return a negative result, the determined pattern MT can also be transferred to the control center LZ via interface S4. Simultaneously, driving data FD from the vehicle FZ can also be transferred to the control center LZ via the fifth interface S5. Based on the data available in the control center LZ, a new fourth closing time SZ4, adapted to the determined train type, can then be calculated in a modification step MOD and transferred to the storage unit SE via an output step OUT. This fourth closing time SZ4 can then be used as the individual closing time for the level crossing BU (replacing the previously stored closing time). At the same time, an output can be made to the storage unit SE such that the fourth closing time SZ4, together with the newly determined reference pattern RMT, which belongs to the vehicle FZ currently being analyzed, is written to the database in the storage unit SE as an addition.

[0085] In the storage unit SE, a closing time for the level crossing BU is now stored in a separate memory area. Depending on the procedure, this could be the standard closing time SZS, the first closing time SZ1, the second closing time SZ2, the third closing time SZ3, or the fourth closing time SZ4 (or other closing times not included in the example). Figure 4 are described) act.

[0086] This closing time is now available in the separate storage area of ​​the storage device SE in order to control the level crossing BU (see below). Figure 1 ), i.e., to be transferred to the computer CP or another control system for the level crossing BU. The level crossing BU can therefore be operated with an individually determined closing time.

[0087] In Figure 5Two parameters measured or determinable by the axle counter according to the invention are shown as examples in a plane, which could also be called the xy-plane, and on which the measurement distribution MV of the measured values ​​is clearly visible. Accordingly, the velocity GSW would be shown on the x-axis and the axle distances A ... H on the y-axis. The z-axis serves to represent the (for example, estimated) probability densities.

[0088] For the parameters in question, location-specific, representative data are collected or measured and classified in this example, e.g., passenger trains as normal distribution NV2 and freight trains as normal distribution NV1, as described above. This involves a finite number of integer or real-valued measurement data from the axle counters; for example, these could be the speed and axle spacing, to give a clear two-dimensional example. In principle, this results in a classification problem, as described in Figure 5 schematically represented.

[0089] Given representative data, it is known how to solve such pattern recognition problems using machine learning methods, e.g., neural networks. In this application with axle counters, there is considerable flexibility in setting the classification boundary, because with such low-dimensional problems, the probability densities for the two classes can also be estimated from the data (e.g., using density estimation). This allows the error probabilities for an incorrect classification to be determined (see, e.g., Duda et al.: Pattern Classification, Wiley, 2001). Figure 5 This is shown symbolically for a first normal distribution NV1 and a second normal distribution NV2, but in principle this also works for distributions other than normal distributions.

[0090] In the example, assuming the small ellipse represents the first classification boundary KG1 for freight trains and the large ellipse the classification boundary KG2 for passenger trains, the error probabilities could be calculated using the estimated distributions. If the misclassification probability for freight trains were too high, the classification boundaries would be changed. In the example according to... Figure 5This would result in a smaller ellipse for the first classification boundary, KG1. However, there can also be applications where the classification errors are asymmetrical, meaning the errors do not have the same significance. For example, in the case of time-controlled activation of a level crossing, it would be irrelevant from a safety perspective if a slow freight train were classified as a fast passenger train, whereas this would be dangerous in the case of a tunnel passing prohibition. Therefore, the safety aspect must always be considered when evaluating the types and probabilities of errors.

Claims

1. Method for identifying characteristics of a rail vehicle (FZ) in which • an axle counter (AZ1 ... AZ2) captures measurement data while the rail vehicle (FZ) is passing over, • the measurement data is analysed in a computer-aided manner and in this case wheelbases (A ... H) of the rail vehicle (FZ) are determined, • a first characteristic is determined in a computer-aided manner with reference to the determined wheelbases (A ... H), namely whether the rail vehicle (FZ) is a passenger train (PZ) or a freight train (GZ), characterised in that in the case of determining the first characteristic in a first testing step (PS1), a check is performed as to whether at least in a predominant part of the sequence of the wheelbases (A ... H) it is possible to determine identical or, if at least a specified predominant part of testing criteria for the patterns match, similar patterns (MT) and • if it has not been possible to determine an identical or similar pattern (MT), the characteristic of a freight train (GZ) is allocated to the rail vehicle (FZ) as the first characteristic, or • if an identical or similar pattern (MT) has been determined, the characteristic of a passenger train (PZ) is allocated to the rail vehicle (FZ) as the first characteristic and / or a further testing step (PS2, PS3) is performed.

2. Method according to claim 1, characterised in that in the case of the first testing step (PS1) in the sequence of the wheelbases (A ... H) a number of wheelbases (A ... H) remain unconsidered at the start of the sequence and / or a number of wheelbases (A ... H) remain unconsidered at the end of the sequence.

3. Method according to one of the preceding claims, characterised in that in the or a further testing step (PS2) the quantity of wheelbases (A ... H) is determined, wherein the characteristic of a passenger train (PZ) is allocated to the rail vehicle (FZ) as the first characteristic as long as the quantity of the largest wheelbase (A ... H) that is provided in the pattern (MT) exceeds a specified limit value (GW).

4. Method according to one of the preceding claims, characterised in that in the or a further testing step (PS3) the patterns (MT) of the wheelbases (A ... H) are compared with reference patterns (RMT) of wheelbases (A ... H) and in the event of an identified match of the pattern (MT) with a reference pattern (RMT) a train type that is linked to the reference pattern (RMT) is allocated to the rail vehicle (FZ) as a second characteristic.

5. Method according to one of the preceding claims, characterised in that further characteristics of the rail vehicle (FZ), in particular the direction of travel of the train and / or the speed of the train when an axle is passing over and / or the average speed when an axle is passing over and / or the acceleration when an axle is passing over are determined using the axle counter (AZ1 ... AZ2) .

6. Method according to one of the preceding claims, characterised in that the criteria for the first testing step (PS1) and / or the further testing steps (PS2, PS) are evaluated using a method of machine learning.

7. Method according to one of the preceding claims, characterised in that probability densities for the characteristics are determined from the measurement data of a plurality of measurements.

8. Method according to one of the preceding claims, characterised in that the method is implemented in order to determine the characteristics of the rail vehicle (FZ) while this rail vehicle is approaching a risk site on the track that is provided for said rail vehicle.

9. Method according to claim 8, characterised in that the method is implemented for rail vehicles (FZ) that are approaching a level crossing (BU), wherein the closing time (SZ1 ... SZ4) of the level crossing (BU) is selected in dependence upon the determined rail vehicle (FZ) if characteristics of the approaching rail vehicle (FZ) have been determined.

10. Method according to claim 9, characterised in that for a determination of the closing time (SZ1 ... SZ4) a data pool is used in which closing times (SZ1 ... SZ4) are linked to the determinable characteristics of the rail vehicles (FZ), in particular train types.

11. Method according to one of claims 9 or 10, characterised in that for the case that it was not possible to determine the characteristic of the rail vehicle (FZ), a standard closing time (SZS) is selected for the level crossing (BU).

12. Apparatus for determining characteristics of rail vehicles (FZ), said apparatus comprising • at least one axle counter (AZ1 ... AZ2) for capturing measurement data when rail vehicles (FZ) pass over, • a computer (CP) that is configured so as to analyse the measurement data and in this case to determine wheelbases (A ... H) of the rail vehicle (FZ), and also • with reference to the determined wheelbases (A ... H) to determine a first characteristic of whether the rail vehicle (FZ) is a passenger train (PZ) or a freight train (GZ), characterised in that the computer (CP) is moreover configured for the purpose of, during the determination of the first characteristic in a first testing step (PS1), checking whether it is possible to determine repeating identical patterns or, if at least a specified predominant part of testing criteria for the patterns match, similar patterns (MT) at least in a predominant part of the sequence of the wheelbases (A ... H) that are determined and • if it has not been possible to determine an identical or similar pattern, to allocate the characteristic of a freight train (GZ) to the rail vehicle (FZ) as the first characteristic, or • if an identical or similar pattern has been determined, to allocate the first characteristic of a passenger train (PZ) to the rail vehicle (FZ) and / or to perform a further testing step (PS2, PS3).

13. Computer program product having program commands for implementing the method according to one of claims 1 - 11.

14. Providing apparatus for the computer program product according to claim 13, wherein the providing apparatus stores and / or provides the computer program product.

Images