FORECASTING WEAR CHARACTERISTICS OF RAIL VEHICLES

[0001]The invention relates to a method for determining a time-dependent wear behavior of a technical component of a rail vehicle. The invention further relates to a method for planning the use of rail vehicles. The invention moreover relates to an estimation device. The invention furthermore relates to a planning device.

[0002]The wear of functional units of rail vehicles must be regularly monitored by maintenance personnel to prevent breakdowns and accidents caused by a technical failure of the functional unit. The regular visual inspection of the technical components that are susceptible to wear is complex and also costs valuable operating time. In contrast, an automatic prediction of material wear in a railway environment would allow improved and plannable maintenance intervals and therefore cost saving with optimum operation of rail vehicles, in particular locomotives. Potential savings are produced particularly in the case of components with a high degree of wear, such as contact strips of current collectors or wheels, for example. Individual wear patterns usually result due to local conditions. Such local conditions include, for example, the type and level of the network voltage used, the degree of wear of the overhead line, etc.

[0003]Maintenance or replacement of components is usually carried out after a defined number of kilometers traveled or after the contact strip thickness has been measured at the depot. However, in the first procedure, the individual wear behavior of technical components is not considered, and so material resources are not optimally used, and, in the latter procedure, there is a high degree of maintenance outlay for checking the wear state of the respective technical component and so the personnel outlay is comparatively high in this case.

[0004]It is thus the object of the invention to enable more effective prediction of the future wear of a rail vehicle and more effective maintenance of a rail vehicle.

[0005]This object is achieved by a method for determining a time-dependent wear behavior of a technical component of a rail vehicle as claimed in patent claim 1, a method for planning the use of rail vehicles as claimed in patent claim 8, an estimation device as claimed in patent claim 12 and a planning device as claimed in patent claim 13.

[0006]In the method according to the invention for determining a time-dependent wear behavior of a technical component of a rail vehicle, a digital twin of the technical component, which includes a current wear state and a line-dependent wear profile of the technical component of the rail vehicle, is determined and stored. A technical component should be understood as meaning a technical functional unit of a rail vehicle that is used during operation of the rail vehicle and that is subject to usage-dependent, in particular line-dependent, wear. Such a technical component is also referred to colloquially as a wearing part. If the technical component reaches a maximum tolerable measure of wear, it must either be replaced or repaired in the context of a maintenance measure. A digital twin is to be understood as meaning a digital representation of a technical component comprising information about the physical and/or technical state of the technical component. In the specific case, the digital twin represents a current wear state and a location-dependent wear behavior of a technical component of a rail vehicle.

[0007]The current wear state indicates the most recently determined measurement value of the wear of the technical component.

[0008]A wear profile is to be understood as meaning a location-dependent measure of the wear of the technical component. The location dependency is preferably implemented by dividing the line network into line sections and determining an extent of the wear of the technical component associated with a respective line section. Since the wear of the technical component is generally very low for a single journey over a line section, the cumulative wear of a rail vehicle is measured after multiple journeys on the line section and an average wear of the rail vehicle for a single journey over the line section is calculated based on the cumulative wear. This average wear of a technical component for a journey over a specific line section ultimately results in the value of the wear profile of the rail vehicle, also referred to as line-related wear profile component, for this specific line section. A line section is formed by a railway line between two railway stations. The line section is preferably formed by a railway line between two adjacent railway stations.

[0009]In this way, it is possible to determine a location-dependent wear profile for different technical components of an individual rail vehicle. The wear profile may also be generalized and parameterized. This means that the wear profile is preferably determined based on wear values from observing a plurality of rail vehicles and scenarios with different parameters or parameter values. As explained in more detail later, the wear of a pantograph or of wheels of a rail vehicle may depend on weather conditions and the weight of the rail vehicle. Instead of a mean value for the wear, a function that is dependent on different variables and/or parameters that influence the wear of a technical component may also be associated with a wear profile for each line section.

[0010]The future time-dependent wear behavior of the rail vehicle is also estimated depending on a future timetable of the rail vehicle and based on the digital twin. As already mentioned, the digital twin includes both a current wear state of the technical component of the rail vehicle and a line-dependent wear profile, based on which the future time-dependent wear behavior of the rail vehicle is estimated.

[0011]The timetable can be used to determine the future travel route and line use of a rail vehicle. The line sections associated with the travel route can be determined based on the travel route and based on the wear profile values or wear profile functions associated with the individual line sections and known and predicted values for variables and parameters can be used to calculate wear values of an individual rail vehicle predicted for individual line sections. The predicted total wear when a timetable is observed in a predetermined time interval then results from the sum of the predicted wear values for each line section.

[0012]A maintenance interval or a subsequent maintenance time can advantageously be stipulated based on the predicted wear of a technical component of a rail vehicle. Conventionally sweepingly stipulated maintenance intervals can thus be adapted individually when the method according to the invention is applied, such that maintenance outlay can be reduced to the necessary minimum. The maintenance of an individual rail vehicle can thus be planned depending on the timetable. Rail vehicles can also be used more optimally for particular travel assignments, with their current wear state and their respective predicted subsequent maintenance time being considered.

[0013]A method for planning the use of rail vehicles can also be specified based on the aforementioned considerations and described advantageous effects. In the inventive method for planning the use of rail vehicles, a future wear behavior of a plurality of available rail vehicles is estimated for different use scenarios of the rail vehicles in order to observe a predetermined timetable using the method according to the invention for determining a time-dependent wear behavior of a technical component of a rail vehicle. Furthermore, a use scenario is selected depending on a secondary condition related to the future wear behavior or the predicted wear and the required maintenance of the rail vehicles. The secondary condition preferably includes one or more wear-related and maintenance-related criteria to be met. A use scenario should be understood as meaning a plan according to which particular vehicles are specifically used on particular lines according to a timetable in order to meet the mentioned secondary condition. For example, such a criterion may require a minimum maintenance outlay at a predetermined time interval. Use of the rail vehicles may also be adapted to maintenance capacities. For example, the wear of the rail vehicles is controlled such that, for a minimum amount of rail vehicles to be kept available and for present or predefined maintenance capacities, there are always sufficient maintenance capacities available for the rail vehicles and so there are no timetable cancelations due to maintenance queues or the like.

[0014]The estimation device according to the invention comprises a database for determining and storing a digital twin of a technical component of a rail vehicle with a current wear state and a line-dependent wear profile of the technical component of the rail vehicle. The estimation device according to the invention furthermore comprises an estimation unit for estimating the future time-dependent wear behavior of the technical component of the rail vehicle depending on a future timetable of the rail vehicle and based on the digital twin. As already mentioned, the digital twin includes information about a current wear state of the technical component of the rail vehicle and information about a line-dependent wear profile of the technical component of the rail vehicle. The estimation device according to the invention shares the advantages of the method according to the invention for determining a time-dependent wear behavior of a technical component of a rail vehicle.

[0015]The planning device according to the invention comprises an estimation device according to the invention for estimating the wear of a plurality of available rail vehicles for different use scenarios of the rail vehicles in order to observe a predetermined timetable and a selection unit for selecting a use scenario depending on a secondary condition related to the wear and the required maintenance of the rail vehicles. The planning device according to the invention shares the advantages of the method according to the invention for planning the use of rail vehicles.

[0016]The majority of the aforementioned components of the estimation device and planning device according to the invention may be implemented entirely or partly in the form of software modules in a processor of an appropriate computation system, for example a computer in a central office of a transport company or at a railway depot. A substantially software-based implementation has the advantage that even previously used computation systems can easily be retrofitted by way 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 computation system, having program sections to execute the steps of the method according to the invention for determining a time-dependent wear behavior of a technical component of a rail vehicle and the method according to the invention for planning the use of rail vehicles, at least the steps that can be executed by a computer, in particular the step of determining and storing a digital twin, the step of estimating the future time-dependent wear behavior of the rail vehicle, the step of estimating the wear of a plurality of available rail vehicles and the step of selecting a use scenario depending on a secondary condition related to the wear and the required maintenance of the rail vehicles, when the program is executed in the computation system. A computer program product of this kind may include additional components in addition to the computer program, such as documentation, and/or additional components, also hardware components, such as hardware keys (dongles, etc.) for using the software.

[0017]For transport to the computation system or the control unit and/or storage on or in the computation system or the control unit, a computer-readable medium, for example a memory stick, a hard drive or another transportable or fixedly installed data carrier, may be used, on which the program sections of the computer program that can be read and executed by the computation system are stored. To this end, for example, the computation system may comprise one or more cooperating microprocessors or the like.

[0018]The dependent claims and the subsequent description each contain particularly advantageous refinements and developments of the invention. Here, in particular, the claims of one claim category can also be developed in a manner analogous to the dependent claims of another claim category and the parts of the description pertaining thereto. In addition, the various features of different exemplary embodiments and claims can also be combined to form new exemplary embodiments within the scope of the invention.

[0019]In the method according to the invention for determining a time-dependent wear behavior of a technical component of a rail vehicle, the technical component preferably includes a technical component for energy transmission. In this case, the energy transmission preferably includes the transmission of electrical energy and/or mechanical energy. When energy is transmitted, a plurality of dissipation phenomena occur, wherein the component or the components for energy transmission are worn. The technical component likewise preferably includes a component that moves relative to a contact element that is preferably stationary. The energy transmission may include both energy supply to the rail vehicle and a portion of the traction process. The relative movement may result in friction effects, which lead to wear of the technical component. The contact element is very particularly preferably part of the stationary railway infrastructure, with the technical component moving relative to the contact element. Since the wear of a technical component is generally dependent on the line in the case of direct contact with the infrastructure, it is particularly effective to estimate the future time-dependent wear behavior of a technical component of a rail vehicle based on a line-dependent wear profile, which is part of a digital twin of the technical component of the rail vehicle, and a future timetable of the rail vehicle.

[0020]In addition to energy transmission at an interface between the rail vehicle and the infrastructure, for example an overhead line or the traversed tracks, the technical component may also be part of a technical device installed within the rail vehicle for electrical or mechanical energy transmission or energy conversion. In the case of an internal technical component of a rail vehicle, loading of the technical component that is different depending on the line may also have an individual effect on the extent of the wear thereof.

[0024]In this case, the current wear state for the contact strip preferably includes information with respect to the current wear state of the contact strip. This information preferably includes a current contact strip thickness.

[0025]The current wear state of a wheel profile preferably includes information with respect to the current wear state of the wheel profile. This information preferably includes quantitative information with respect to a deviation between the current wheel profile and a wheel profile of an unused wheel of a rail vehicle of the same type. The maintenance of this technical component that is particularly susceptible to wear can advantageously be optimized and the use of rail vehicles can be matched to a current and predicted wear of the individual rail vehicles and can be optimized.

[0026]As already mentioned, the future wear-dependent wear behavior of the technical component of the rail vehicle is estimated based on the current wear state and the line-dependent wear profile of the rail vehicle, each of which are part of the stored digital twin. There is advantageously no need for additional investigation for the wear prediction; instead, it is sufficient to retrieve the current state data of the digital twin.

[0027]The wear profile preferably comprises a digital railway network map with which location-dependent attributes which influence the wear profile are associated. These location-dependent attributes make it possible to create a “wear map”, which allows a prediction of the wear of a technical component depending on the selected travel route of the rail vehicle. Location-dependent attributes may include, for example, the type of traction power network used, the degree of wear of the overhead line and a location-dependent incline or a location-dependent downward slope. A digital railway network map makes it possible to assign attributes that influence the wear profile to points or partial lines or line sections. Digital maps may be based on map material, for example detailed map material available on the Internet such as “Openrail” or “OpenStreetMap”, which depicts each line as a graph.

[0028]In this case, the digital railway network map preferably depicts the lines as a graph with edges as the connection between adjacent stations, preferably railway stations. Line-section-related wear profile components are preferably associated with the edges of the graph. The overall wear can advantageously be calculated for a specific timetable by adding values of line-section-related wear profile components from different line sections.

[0029]As an alternative, wear patterns can also be stored in nodes of a graph when a technical component is not subject to continuous wear but rather “intermittent” wear arising discreetly. Such instantaneous wear is preferably effected by a temporary fault in the infrastructure resulting in intermittent wear. Such an effect may arise, for example, in the case of a defective/worn-out rail joint or in the case of lowering of railroad ties. A worn overhead line may also produce such an effect. Instantaneous wear can also be caused by pressure differences produced when entering and exiting a tunnel, which result in undesirable transverse forces and thus distinct wear patterns.

[0034]An expected value for wear of the technical component arising when a predetermined timetable is observed can be formed based on a statistic with respect to the wear of the technical component arising based on the line section.

[0035]Weather data provide information about the current and expected weather conditions when a timetable is observed. These weather conditions influence the wear of a technical component of a rail vehicle.

[0036]Vehicle data or vehicle-specific data also influence the expected wear of a technical component of a rail vehicle. For example, the wear of a contact strip and a wheel of a rail vehicle is dependent on the weight of the rail vehicle. Typical vehicle-specific properties include the type of pantograph used by a rail vehicle.

[0041]The vehicle characteristics include individual properties of a rail vehicle, for example a vehicle version or a vehicle type.

[0042]Assignment data relate to all data that influence the wear of the technical components and that are associated with the individual operation of the individual rail vehicles. Assignment data include, in particular, the route traversed and the length or condition thereof.

[0043]The maintenance data include the current wear state, preferably the contact strip thickness or wheel rim thickness and possibly also performed or planned maintenance activities.

[0044]The type of route traversed influences the wear behavior of different technical components of a rail vehicle. For example, wheels or contact strips on lines with an increased incline wear to a greater extent than on flat lines. The wear of contact strips and wheels is also greatly increased at increased speed. The wear of the wheels also depends on the geometry of the wheels and the track traversed.

[0045]In addition to directly storing statistical data, functional relationships for calculating the wear, either overall or for individual line sections, can also be stored in the digital twin. For example, as already briefly mentioned, the wear of a technical component can be represented as a linear or non-linear function of a plurality of variables and parameters. Such parameters preferably include the length of a line section, the average speed of a rail vehicle on the section and, specifically for a contact strip, the current density through the contact strip.

[0046]As an alternative or in addition, instead of a rigid model for the wear of a technical component as a function of variables and parameters that influence the wear, a model based on artificial intelligence can also be determined and stored. Typical methods for producing such a model include machine learning. Such a model is preferably trained individually for a plurality of line sections, particularly preferably for each line section of a line network or each edge of a graph associated with the line network. To this end, so-called labeled training data are used, said training data including both known values of variables and parameters as input data as well as wear values as label, which can be compared with the result data produced during training. Such a model can even advantageously be retrained or updated during a journey of a rail vehicle if parameter values and values for variables and wear values of the technical components are measured during or after said journey, so as to achieve greater accuracy and currency of the model. In this case, it should be noted that the wear of a technical component, for example a contact strip or a wheel rim, generally turns out to be lower than the measurement tolerance of measurement instruments used to measure the wear. Therefore, the wear can be measured with the required accuracy only after a plurality of journeys over a line section. A model based on artificial intelligence has the advantage of improved flexibility and scalability since the model is adapted to changing conditions automatically and, in contrast to determining the expected wear based on a statistic, less data has to be kept available.

[0047]Such a model preferably includes a regressor. A regressor can be used to determine a functional relationship based on statistical data.

[0048]In the method according to the invention for use planning, the secondary condition preferably includes the maximum maintenance capacities for the rail vehicles. Operation according to the timetable can advantageously be upheld by taking into account maintenance capacities.

[0049]The secondary condition preferably includes a minimum maintenance outlay for the rail vehicles. Resources for maintenance of the rail vehicles can advantageously be saved.

[0050]The secondary condition also preferably includes observing the timetable without maintenance. The intention in this variant is to avoid vehicles having to be taken out of operation temporarily in order to perform maintenance on them. This procedure may be expedient if very few rail vehicles are available and lost time due to maintenance is to be avoided.

[0051]The secondary condition may also include observing the timetable with the lowest possible number of rail vehicles. In this case, simultaneous maintenance processes on different vehicles should be kept to a minimum in order to keep as many vehicles as possible in use at the same time.

[0052]An extent of the wear of the technical component is preferably determined automatically in order to create the digital twin. A wear twin can advantageously be created without personnel outlay.

[0053]The invention is explained in more detail again below with reference to the appended figures based on exemplary embodiments. In the figures:

[0054]FIG. 1 shows a graphical illustration of a rail connection, which illustrates a wear behavior of a contact strip of a rail vehicle,

[0055]FIG. 2 shows a chart illustrating a dependency of a wear behavior on a plurality of parameters,

[0056]FIG. 3 shows a flowchart illustrating an estimation of a wear behavior based on an AI-based model,

[0057]FIG. 4 shows a flowchart illustrating a planning method according to an exemplary embodiment of the invention,

[0058]FIG. 5 shows a schematic illustration of an estimation device according to an exemplary embodiment of the invention,

[0059]FIG. 6 shows a schematic illustration of a planning device according to an exemplary embodiment of the invention.

[0060]FIG. 1 illustrates a graphical representation of a section of a digital network map with a rail connection. The graphical representation illustrates a wear behavior of a contact strip of a rail vehicle. The rail connection runs between Berlin (abbreviated to B) via Leipzig (related to L), Nuremberg (abbreviated to N) to Munich (abbreviated to M). Wear indicators V_BL, V_LN, V_NM for partial sections between the individual cities of a contact strip of a rail vehicle are illustrated or arranged at the edges of the graph as line-related wear profile components VK. It goes without saying that such a wear indication can also include indications relating to wear of a plurality of contact strips of different rail vehicles. Wear indications may also include a statistic of the wear of a contact strip for a plurality of rail vehicles. The wear profile may also illustrate a dependency of the wear on current weather data or characteristic vehicle data. A statistically expected wear or a statistically expected wear profile of a particular rail vehicle that is to travel on a predetermined line can also be calculated based on the statistical data stored for each edge of the graph and if appropriate also functional relationships given knowledge of the vehicle type and the predicted weather. If the current contact strip thickness of a contact strip of a particular rail vehicle is known, it is possible to use the aforementioned data to estimate the time at which the rail vehicle will need to have maintenance for a predetermined timetable. The collection of data represented by the graphical illustration shown in FIG. 1 is also referred to as a digital twin ZW or as a wear twin.

[0061]FIG. 2 shows a chart 200 illustrating a modeled representation or a non-linear function of the wear A (in micrometers μm per kilometer km) as a function of variables and parameters, such as the speed v (in kilometers km per hour h) and the electrical current density J (in amperes A per square centimeter cm²) through the contact strip. In this variant, a pre-calculation of the wear of a rail journey on a predetermined railway line can be determined without having to directly access a database with a large amount of statistics.

[0062]FIG. 3 shows a flowchart 300 illustrating an estimation of a wear behavior of a pantograph based on an AI-based model.

[0063]In step 3.I, data with respect to the wear profile A (P) of a pantograph type are collected for each section or each edge of a graph representing a line map. In this case, a particular pantograph type is stipulated for each edge depending on the power system with which the associated line is provided. That is to say if a pantograph of a first pantograph type has to be used on a line section, it is not expedient to also show values for a second pantograph type that is technically different from the first pantograph type. The type of rail power is also stipulated for each line section, that is to say, for example, whether DC (direct current) or AC (alternating current) is used and at what electrical voltage, for example 1.5 kV, 3 kV, 16.7 kV or 25 kV, the railway power network in this section is operated. However, different pantograph types and different types of railway power can be used by one and the same rail vehicle for a longer planned line. Other parameters P that influence the wear of a pantograph relate to the speed of the rail vehicle, weather data and the weight of the train. The weight of the train influences the electrical current density J of the electrical current flowing through the current collector. The wear of a pantograph also depends on the length of the line and the electrical current density J through the contact strip of the pantograph.

[0064]Instead of simply storing the mentioned parameters and wear values, in the exemplary embodiment illustrated in FIGS. 3, in step 3.II, an artificial neural network KN is trained. In this case, the mentioned parameters P, such as the electrical current density J through the contact strip, the speed of the rail vehicle, the type of pantograph, the weight of the train, weather data, the length of the line, etc., are used as input data that are labeled using the measured output data, that is to say the associated wear values of the wear profile A (P). The artificial neural network is trained so that it determines the associated output data based on the input data.

[0065]In step 3.III, the trained artificial neural network KN is used to predict a future wear behavior VV of a contact strip of a pantograph of a rail vehicle with a stipulated timetable FP.

[0066]FIG. 4 shows a flowchart 400 illustrating a method for planning the use of rail vehicles in accordance with an exemplary embodiment of the invention.

[0067]In step 4.I, the method for estimating a future wear behavior VV of a plurality of available rail vehicles already illustrated in FIG. 1 to FIG. 3 is carried out for different use scenarios of the rail vehicles in order to observe a predetermined timetable FP.

[0068]In step 4.II, a use scenario ES is selected depending on a secondary condition NB related to the wear VV expected in the future and the required maintenance of the rail vehicles. Such a secondary condition may require, for example, a minimum maintenance outlay for the rail vehicles and observation of the timetable by the rail vehicles.

[0069]FIG. 5 shows an estimation device 50 according to an exemplary embodiment of the invention. The estimation device 50 comprises a database 51 for determining and storing a digital twin ZW of a technical component. The digital twin ZW includes a current wear state VS and a line-dependent wear profile V for a technical component, in particular a contact strip, of a rail vehicle.

[0070]The estimation device 50 further comprises an estimation unit 52 for estimating the future time-dependent wear behavior VV of the rail vehicle depending on a future timetable FP of the rail vehicle, the current wear state VS and the line-dependent wear profile V of the rail vehicle in question.

[0071]FIG. 6 illustrates a planning device 60 according to an exemplary embodiment of the invention. The planning device 60 comprises the estimation device 50 shown in FIG. 5. Part of the planning device 60 is also a selection unit 61 for selecting a use scenario ES depending on a secondary condition NB related to the wear and the required maintenance of the available rail vehicles. The selection unit 61 requests from the estimation device 50 the digital twins ZW of the individual rail vehicles of an available vehicle fleet FUP. The digital twins ZW include a current wear state VS of one or more technical components of the individual rail vehicles of the vehicle fleet FUP and the respective wear profile V of each rail vehicle. The selection unit 61 determines a use plan or a use scenario ES for optimum use of the vehicle fleet FUP based on the digital twins ZW of the individual rail vehicles of the vehicle fleet FUP, a timetable FP and a previously stipulated secondary condition NB which, in addition to the main condition that no rail vehicle may be used over the wear limit, specifies a further optimization aim. The secondary condition NB may be, for example, that there should be a minimum maintenance out lay.

[0072]Finally, it is noted once again that the above-described methods and devices 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 to the extent that it is specified by the claims. For the sake of completeness, it is also noted that the use of the indefinite article “a” or “an” does not preclude the features in question also being present in a plurality. Likewise, the term “unit” does not preclude the latter from consisting of a plurality of components that may also be distributed spatially, if appropriate.