Method and device for operating a device battery for optimizing the use life after the identification of a current or

Abstract

The invention relates to a method for operating a system battery, in particular a system battery in a technical system after detecting an abnormality of the system battery, comprising: providing a temporal operating variable curve of a plurality of determined operating variables of the system battery; the operating variable curve over time is predicted from the provided operating variable curve over time and the plurality of load parameters of the load pattern by means of a usage pattern model in order to obtain an artificial operating variable curve, the usage pattern model being designed to provide an artificial curve of the operating variables from the plurality of load parameters, the plurality of load parameters correspond to stress factors for aging of the equipment battery pack; determining the residual service life of the equipment battery pack by means of an aging state model according to the manual operation variable curve; executing an optimization method for adapting a plurality of load parameters in order to determine an optimized load parameter, by means of which the remaining service life can be increased when the optimized load parameter is taken into account in future operation of the technical installation; and operating the equipment battery pack according to the optimized load parameters.

Description

Technical Field

[0001] The invention relates to a method for operating a device battery pack for which an abnormal or faulty behavior has been detected, in particular to adapting the operating method after the abnormality has been detected. The invention also relates to a method for selecting suitable measures to extend the service life of the battery pack and to increasing the service life of the device battery pack with the abnormality. Background Art

[0002] The energy supply of electrical devices and machines that are operated independently of the power grid, such as electrically driven motor vehicles, is usually provided by device batteries or vehicle batteries, which provide the electrical energy for operating these devices.

[0003] Equipment batteries degrade during their service life and depending on their load or usage. This so-called aging leads to a continuous decrease in the maximum performance or storage capacity. The state of aging corresponds to a measure that describes the aging of the accumulator. Conventionally, a new equipment battery can have a state of aging (with respect to its capacity, i.e. SOH-C) of 100%, which gradually decreases during its service life. The measure of equipment battery aging (the change in the state of aging over time) depends on the individual load of the equipment battery, i.e. in the case of a vehicle battery of a motor vehicle on the driver's usage behavior, external environmental conditions and the vehicle battery type.

[0004] In battery-operated technical devices, especially in the case of high energy density, the normal operating mode of the device batteries used must be regularly monitored for errors for safety reasons. If a battery cell, a unit consisting of several battery cells or an entire device battery fails, the technical device may be restricted or completely inoperable, depending on the error that has occurred, and in the case of malfunctions that lead to a strong increase in temperature, the safety of the technical device and the user may also be affected.

[0005] Therefore, electrical devices and machines that operate independently of the power grid, especially device batteries in mobile devices, are continuously monitored in order to promptly identify sudden device battery failures (sudden death), thermal error events, rapid cycle aging or other error events (collectively referred to as anomalies in this article). Depending on the type of anomaly detected, different measures can be taken, ranging from a simple warning to the user to an immediate shutdown of the technical device.

[0006] In this case, the abnormality detection includes: based on the measured time curve of the operating variable of the device battery To detect current or future anomalies. In order to perform anomaly detection that is as independent as possible, anomaly detection models are implemented in the technical system, which evaluate the operating variable profile and signal anomalies in a suitable manner in the event of deviations from a normal operating pattern.

[0007] Publication DE 102019208372 A1 discloses a computer-implemented method for identifying anomalies in a technical system, comprising the following steps: detecting an operating variable vector, which describes the operating state of the technical system and includes a plurality of operating state variables, wherein the operating state variables include at least one environmental state variable, which describes the environmental conditions for operating the technical system, and the system state variables indicate an internal system state of the technical system; providing an environmental state model and an anomaly detection model, wherein the environmental state model describes the verifiability of the operating variable vector with respect to the presence of anomalies based on at least one of the environmental state variables by using an anomaly detection model, and wherein the anomaly detection model describes the presence of anomalies that should be expected based on the operating variable vector; signaling the presence of an anomaly or non-anomaly based on an evaluation of at least one environmental state variable of the operating variable vector based on the environmental state model and based on an evaluation of the operating variable vector based on the anomaly detection model.

[0008] In addition, the publication "LSTM-Based Battery Remaining Useful Life Prediction With Multi-Channel Charging Profiles" by K. Park, Y. Choi, W. J. Choi, H.-Y. Ryu and H. Kim, in: IEEE Access, Volume 8, Pages 20786-20798, 2020, doi: 10.1109 / ACCESS.2020.2968939 discloses the use of an LSTM model to predict the remaining useful life of a battery pack for different charging profiles (Ladeprofile). Summary of the invention

[0009] According to the present invention, a method for operating a device battery pack after identifying a current or predicted anomaly and, in particular, for determining optimization measures for extending the service life of a device battery pack after identifying a current or specific anomaly is provided according to claim 1, and a corresponding device is provided according to the parallel claims.

[0010] Further refinements are described in the dependent claims.

[0011] According to a first aspect, a computer-implemented method for operating a device battery in a technical device is provided, in particular after an abnormality of the device battery has been detected, the method comprising the following steps:

[0012] - providing a time-dependent operating variable curve of a plurality of operating variables of the determined device battery;

[0013] - predicting the temporal operating variable curve (F) based on the provided temporal operating variable curve (F) and a plurality of load parameters of the load pattern by means of a usage pattern model to obtain an artificial operating variable curve (F), wherein the usage pattern model is designed to provide an artificial curve of at least one operating variable based on the plurality of load parameters, wherein the plurality of load parameters correspond to stress factors for aging of the battery of the device;

[0014] -determine the remaining useful life of the battery pack of the device based on these artificial operating variable curves;

[0015] - performing an optimization method for adapting the plurality of load parameters in order to determine optimized load parameters, wherein the residual service life can be increased by means of these optimized load parameters when these optimized load parameters are taken into account in future operation of the technical installation;

[0016] - operating the device battery according to at least one of the optimized load parameters.

[0017] Furthermore, based on the time-dependent operating variable curve, it can be checked whether there is an abnormality in the device battery pack, which may cause the remaining service life of the device battery pack to be less than the guaranteed service life of the device battery pack, wherein the optimization method is executed only when the abnormality is determined.

[0018] In order to monitor device batteries in a large number of devices, operating variable data can be recorded continuously and transmitted block by block as operating variable curves to a central unit outside the device. In order to evaluate the operating variable data, in particular in a physical or electrochemical battery model based on differential equations, these operating variable data are sampled as curves with a relatively high time resolution (sampling rate), for example between 1 Hz and 100 Hz, and the battery state is determined therefrom using a time integration method.

[0019] In order to evaluate the operating variable data, in particular to determine the internal battery state (which also determines the aging state), an electrochemical battery model (P2D battery model) can be used, which is based on a differential equation system (Differenzialgleichungssystem) with multiple nonlinear differential equations. The operating variable data enables the current battery state to be modeled using a time integration method. Such an electrochemical battery model is known, for example, from publications US2020 / 150,185A, WO2022017984A1, EP3919925A1, DE102020206915B3 and US20210373082A1.

[0020] Providing the operating variable curve in the central unit makes it possible to use and adapt the electrochemical battery model for a large number of system batteries with similar battery cells or cells with similar battery chemistry. The calculation of the battery state by means of a system of differential equations is computationally intensive, so that the computational load in the internal computing device of the system can also be reduced by transferring it to the central unit.

[0021] In addition, the aging state can also be determined based on the operating variable curve. The aging state (SOH: State of Health) is a key variable for describing (Angabe) the remaining battery capacity or the remaining battery charge in the case of a device battery pack. The aging state represents a measure of aging or performance for a device battery pack. In the case of a device battery pack or a battery pack module or a battery pack cell, the aging state can be described as a capacity retention rate (Capacity Retention Rate, i.e., SOH-C). The capacity retention rate SOH-C, that is, the capacity-related aging state is described as the ratio of the measured instantaneous capacity to the initial capacity of the fully charged battery pack, and decreases with increasing aging. Alternatively, the aging state can be specified as an increase in internal resistance relative to the internal resistance at the beginning of the service life of the device battery pack (SOH-R). The relative change in internal resistance SOH-R increases as the battery pack ages.

[0022] Furthermore, as mentioned at the outset, the device battery is therefore continuously monitored in order to promptly detect sudden failures (sudden death), thermal error events, rapid cycle aging or other error events (collectively referred to herein as anomalies) of the device battery. Depending on the type of anomaly detected, different measures can be taken, ranging from a simple warning to the user to an immediate shutdown of the technical device. For this purpose, anomaly recognition models are used which recognize current or future anomalies based on an evaluation of the time profile of operating variables.

[0023] Abnormal identification for device batteries is usually performed by evaluating operating variable curves. For device batteries, the operating variable curves usually include battery pack voltage or cell voltage, battery pack current, battery pack temperature and charge state. These operating variable curves can be analyzed with the help of rule-based or data-based abnormal identification models, especially after subsequent feature extraction, to identify device batteries with abnormalities or errors.

[0024] With the aid of methods for anomaly detection, current or future occurring anomalies can be identified. Such methods can be rule-based or based on a plurality of data-based anomaly recognition models and can predict various error types. In particular, anomalies are all those deviations from normal operation of the device battery that lead to a shortened service life or premature failure. Thus, above-average degradation of the device battery and severe failures such as sudden death or thermal runaway or other error types can be identified.

[0025] However, in rule-based anomaly detection, errors in device batteries have so far only been detected if applied error thresholds for operating variables, such as cell voltage, module temperature, current values ​​or state of charge and state of aging values, are exceeded or not exceeded.

[0026] For example, the anomaly recognition model can also be provided in a data-based manner as a classification model, wherein, due to the merging of data from a large number of device battery applications and the high computational effort, the creation and training of the data-based anomaly recognition model must usually also be performed outside the device. For example, various data-based anomaly recognition models are known from DE 102022202111 A1 and DE 102022203343 A1 in the form of autoencoder-based methods. Alternative methods are known, for example, from DE 102022202112 A1.

[0027] Anomaly identification can also be provided in a data-based manner, for example, by means of an autoencoder or a variational autoencoder. For example, such an autoencoder or variational autoencoder can be trained, especially unsupervised, and regularly or continuously updated with the aid of a large number of operating characteristic points of similar device batteries, which have worked without errors or have operated without errors and without failures over a period of time in the past (e.g., until 3 or 6 months ago). The application of the anomaly identification model can determine anomalies of the device battery based on the evaluation of a deviation measure of the reconstruction error or distribution in the latent state space of the autoencoder or variational autoencoder.

[0028] The autoencoder-based approach for anomaly detection is able to classify the current battery pack state as normal or abnormal and, in addition, reliably predict when a failure event will occur in the future with a certain probability. Safety-critical events such as thermal runaway events or complete failure (sudden death) of the device battery pack are predicted in advance by changes in the battery pack state. The type of error that may occur can also be identified using criteria for rule-based anomaly identification or by differentiated classification of anomalies in data-based anomaly identification.

[0029] The early prediction of anomalies makes it possible to attempt to extend the service life or the remaining service life (remaining service life) of the device battery or to achieve compliance with a guaranteed service life by intervening in the operation of the device battery. The service life of the device battery is determined by the point in time at which a predetermined degradation (aging state) is reached. The guaranteed service life predetermines the shortest time period (Mindestzeitdauer) during which the degradation does not exceed a predetermined level. For example, it is common for automotive batteries to guarantee a service life of eight years, wherein the capacity-related aging state is reduced by 20% (to 80%). In particular, in the case of anomalies that lead to faster aging of the device battery and therefore failure to achieve the guaranteed service life, an extension of the expected service life can be achieved by appropriate operation.

[0030] The remaining service life can be determined with the aid of an aging model for time-series-based modeling of the aging state of a device battery pack, wherein a temporal aging state curve is determined based on a provided temporal operating variable curve and an artificial operating variable curve, wherein the remaining service life is determined as a function of the point in time when the aging state curve reaches a predetermined limit aging state.

[0031] The purpose of the above method is to determine one or more load parameters that have an influence or the greatest influence on the degradation of the battery pack as stress factors and to change these load parameters / stress factors in further operation in order to achieve the best possible extension of the service life of the device battery pack. Such load parameters may include, for example, the (maximum or average) battery pack temperature, the charge state window given by the minimum charge state and the maximum charge state or the corresponding voltage window for the electrical terminal voltage, and the charging curve indicating the maximum charging current depending on the charge state.

[0032] In order to determine the service life that can be achieved by a specific device battery, an aging state model based on a time series evaluation of operating variables can be used. The aging state model uses the recorded operating variable curves of the device battery to model the aging state of the device battery, wherein these operating variable curves can include, for example, battery current, battery voltage, battery temperature and charge state. With the help of predetermined load parameters that can be derived from the historical operating variable curves, an artificial operating variable curve can be generated accordingly. Based on the artificial operating variable curve determined in this way, the curve of the aging state can be predicted by applying the aging state model.

[0033] The predicted service life determined in this way results here from the point in time at which the prediction curve of the state of aging reaches the state of aging threshold value predefined for the guaranteed service life.

[0034] An excessively high (average) state of age gradient can result in the guarantee limit of the device battery not being exceeded, which is defined by the overall service life and the state of age limit. If the state of age reaches or falls below the state of age limit, e.g. 80%, or reaches or falls below the state of age limit after a shorter period than the determined (guaranteed) service life, the guarantee specification for the device battery is not met. Therefore, even if the current degree of age of the device battery is still above the age limit, the future warranty risk can be very high.

[0035] For example, the current aging state can be determined with the aid of a physical aging model, which represents a form of an electrochemical battery model and can be parameterized in a corresponding manner. The electrochemical battery model is based on a system of differential equations with a plurality of nonlinear differential equations. The time series of the operating variable data enables the current battery state to be modeled with the aid of a time integration method. Such electrochemical battery models are known, for example, from the publications US2016 / 023,566, US2016 / 023,567 and US2020 / 150,185.

[0036] The aging model can be characterized by its model parameters and models the internal battery state of the device battery. The aging model is based on electrochemical model equations parameterized by the model parameters, which characterize the electrochemical state of a system of nonlinear differential equations and can be continuously evaluated according to a time integration method. A description of the aging state can be derived from the parameterized model parameters in a manner known per se, for example by linear combination.

[0037] In order to improve the accuracy of the aging state model, the aging state model can be provided in the form of a hybrid aging state model, i.e. a combination of a physical aging model and a data-based correction model. In the case of a hybrid state model, the physical aging state can be determined with the aid of a physical or electrochemical aging model and a correction value can be applied to the physical aging state, in particular by addition or multiplication, wherein the correction value results from the data-based correction model.

[0038] Furthermore, the correction model of the data-based hybrid aging state model can be constructed using a probabilistic or artificial intelligence-based regression model, in particular a Gaussian process model, and can be trained to correct the aging state obtained by the physical aging model. For this purpose, there is therefore a data-based correction model for correcting the aging state of the capacity-related aging state SOH-C, and / or there is an additional correction model for correcting the resistance-related aging state SOH-R. Possible alternatives to Gaussian processes are alternative supervised learning methods, which are based on, for example, random forest models, AdaBoost models, support vector machines or Bayesian neural networks.

[0039] The correction model can use operating characteristics as input variables, which are determined from the operating variable curve over time by means of characteristic extraction or feature extraction methods, wherein the characteristics or features are calculated by means of operations in signal processing technology. The operating characteristics assigned to the operating variable curve define operating characteristic points of the relevant device battery for the determined time points, which reflect the load state caused by cyclic operation (cyclic aging) and calendar aging (duration since commissioning or the beginning of the service life) of the device battery.

[0040] The operating characteristics may include cumulative or aggregate characteristics based on load and / or statistical variables determined throughout a previous service life.

[0041] In particular, the feature can be determined as an operating feature based on histogram data, which are created based on the curve of the operating variable. Thus, for example, a histogram of the battery current at the battery temperature and the state of charge of the vehicle battery, a histogram of the battery temperature at the state of charge of the vehicle battery, a histogram of the charging current at the battery temperature, and a histogram of the discharge current at the battery temperature can be created. In addition, as operating features, the total accumulated charge (Ah) since the device battery was put into operation, the average capacity increase during the charging process (especially for charging processes with a charge increase above a threshold share of the total battery capacity [e.g. 20% ΔSOC]), the charging capacity, and the extreme values ​​(e.g. local maximum) of the smoothed differential capacity during the measured charging process with a sufficiently large charge state deviation (smooth curve of dQ / dU: charge change divided by battery voltage change) or the accumulated driving power can be considered. Further operating characteristics may correspond to local extreme values ​​of the spectral kurtosis evaluated based on the charging process of the current or voltage signal, one or more coefficients of the wavelet transform and / or one or more coefficients of the Fourier transform, which are respectively evaluated for the charging process of the current or voltage signal or the spectral values ​​of the transform assigned to a defined frequency band.

[0042] Therefore, the operation characteristics can be derived from the histogram of the operation variables. Thus, the operation characteristics, such as the mean, standard deviation and multidimensional statistical values ​​of the histogram, such as the mean, median, minimum, maximum, moment, etc. of the distribution, can be extracted by means of feature engineering or feature extraction methods.

[0043] Provision may be made to create an artificial operating variable curve as a function of a plurality of load parameters. For this purpose, a data-based usage pattern model may be provided, which converts a predefined load pattern into an artificial curve of a load variable L, which may include partial operating variables and reflect the load of the device battery to which the device battery is subjected during the use and operating mode described by the usage pattern.

[0044] The load patterns can characterize load parameters of the previous use and loading of the vehicle battery or can be defined differently therefrom. The load patterns can be predefined and determined in such a way that, when used together with the usage pattern model, they result in the output of time curves of the battery current and the battery temperature as load variables. With the aid of a suitable known battery dynamic model, the curves of the battery voltage and the state of charge SOC can be determined from the curves of the load variables in order to generate an artificially generated curve of the operating variable together with the load variables.

[0045] The load pattern can also be defined by load parameters that are derived individually for previous operations from historical operating variable curves and / or operating characteristics. The load pattern can also be adapted by changing one or more load parameters in order to implement desired measures for extending the remaining service life of the device battery.

[0046] These load patterns can be evaluated by a usage pattern model, preferably by means of a data-based method, and used to describe the typical usage behavior of individual users with respect to the vehicle battery in question.

[0047] The usage pattern model, if it is itself implemented based on data, can be designed as, for example, a transformer, which can be based on multi-head attention, or alternatively as a recurrent neural network, such as an LSTM or GRU, in particular as a Bayesian LSTM network, and trained based on curves of load variables or operating variables that illustrate typical and representative usage types of vehicle battery packs. The curves of the load variables or operating variables considered here should be based on time periods of the same usage type and the same operating type of the vehicle battery pack. In addition, the usage pattern model can be designed based on physical domain knowledge, as described in US20230306803A1, or it can also be designed as a hybrid model including a physical model and a data-driven model.

[0048] The load parameters that describe the load pattern N then correspond to the input variables of the usage pattern model and characterize the driving behavior of the driver or the type of use of the vehicle. The usage pattern model can be trained in such a way that, given a specified load pattern and a calendar time specification, an operating variable curve is predicted, which, in the case of a suitable evaluation, results in a load parameter that corresponds to the specified load pattern or from which the load parameter can be derived in a corresponding evaluation.

[0049] For evaluation purposes, the usage pattern model can acquire calendar time specifications and load parameters of a previously used load pattern or changed load parameters of a changed load pattern on the input side and output corresponding artificial curves of load variables (current, temperature, preferably as a time series) on the output side.

[0050] The usage pattern model can be designed as an overall model or with a sub-model for each curve of the load variable to be predicted. In this case, one sub-model of the usage pattern model is trained with respect to the time temperature curve and another sub-model of the usage pattern model is trained with respect to the current curve based on a predefined load pattern. Thus, by predefining calendar time information, such as date and clock time and the desired load parameters, an artificial curve of the load variable and / or operating variable can be generated. The calendar time information can also contain the day of the week, the month and information about public holidays and, in particular, take seasonality into account by feature engineering.

[0051] By predicting an artificial operating variable curve based on load parameters of a predetermined load pattern, the predicted aging state curve can be extended into the future with the aid of an aging state model. This enables the remaining service life for a specific load pattern to be determined. The predicted operating variable curve can be modeled based on specifications for load parameters or stress factors, including, for example, upper and lower limits for the state of charge, limitations on battery pack temperature or battery pack temperature range, average or maximum ampere-hour throughput, and selection of a specific charging curve. Other possible load parameters may include aggregated or abstract variables, such as: usage category, annual mileage, fast charging frequency, driving style (e.g., average acceleration), temperature range (e.g., ambient temperature), state of charge offset (regarding SOC), and distributions regarding temperature, state of charge, etc.

[0052] By varying the load parameters or stress factors, an artificial operating variable curve is determined accordingly and the aging state curve is predicted accordingly until the aging state limit defined by the guarantee condition is reached, such as 80% SoH, and the remaining remaining service life can be determined for each varied load pattern. By maximizing the remaining service life in an optimization method, the optimal load pattern can be determined.

[0053] The optimization is used to determine those load parameters that have the greatest influence on the aging of the device battery in which the anomaly was detected. If the load parameter or parameters found in this way are taken into account in future operation, it can be achieved that, despite the anomaly detected in the device battery, the guarantee conditions predefined by the manufacturer can still be complied with.

[0054] The optimization method can be designed as a simple gradient descent method. Alternatively, due to the probabilistic modeling of the aging state model, the optimization method can also be performed in a manner known per se with the aid of a Bayesian optimization method in order to optimize the load parameters and at least in a predictive manner the temperature set points in the state of charge window, the charging curve and the operating strategy, i.e. the limit values ​​of the battery pack temperature. Thus, a suitable adaptation of the load parameters or stress parameters can be carried out for each device battery pack in which an abnormality has been determined.

[0055] From the specific load parameters, those load parameters that have the greatest impact on mitigating aging of the device's battery pack may then be selected.

[0056] According to one specific embodiment, one of the at least one load parameter can be excluded for the optimization depending on the usage behavior of the device battery, wherein the usage behavior is determined based on a provided operating variable profile.

[0057] If necessary, load parameters can be restricted with the aid of additional conditions to meet usage specifications or user expectations. For example, if the device battery pack is always operated in fast charging mode, this is a desired mode that should not be restricted, and therefore it should not be allowed as a load parameter to restrict the charging profile to a gentler charging process. However, in this case, as a load parameter, service life improvements can be made via thermal management or charge state offset. For each device battery pack, the allowed changes in load parameters can be learned regularly from the data, for example by quantile analysis.

[0058] By using a data-based probabilistic model, in particular by means of a Gaussian process, to predict the curve of the aging state, the predicted curve of the aging state can be provided with a confidence interval. With the help of an optimization method, the corresponding curve of the predicted aging state can now be simulated by changing the load parameters. The goal of the optimization is to achieve a predetermined guaranteed service life, for example eight years, with a certain probability, for example 95%, wherein the exact probability of reaching the predetermined service life can be determined by a probability density function, wherein the probability is derived from the probabilistic modeling of the aging state model.

[0059] By means of the optimization, individual load parameters which have the greatest possible influence on the extension of the remaining service life can be identified. In particular, the one or more identified load parameters can be optimized again after having been identified while maintaining the values ​​of the remaining load parameters, so as to adapt the operating strategy according to the reoptimized identified load parameters.

[0060] The optimized load parameter can then be transmitted from the central unit back to the relevant technical device in order to be taken into account in the operating strategy during the further operation of the device battery. For this purpose, at least one parameter in the operating strategy can be adapted so that the value of the optimized load parameter or the values ​​of a plurality of optimized load parameters can be reached as a target state. It is also possible to take into account in the operating strategy only the load parameter with the greatest change or one or more load parameters with the greatest influence on the extension of the remaining service life. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The embodiments are explained in more detail below with reference to the accompanying drawings, wherein:

[0062] Figure 1A schematic diagram of a system for providing driver and vehicle individual operating variables for identifying predicted anomalies and for determining optimized operating load parameters for a vehicle battery in a central unit is shown;

[0063] Figure 2 A schematic diagram showing a hybrid aging state model for probabilistic modeling of aging state curves based on load patterns; and

[0064] Figure 3 A flow chart illustrating a method for determining optimized operating parameters for extending the service life of a vehicle battery pack is shown;

[0065] Figure 4 Schematically, curves of predicted aging states are shown for the case of an anomaly in the vehicle battery and for the vehicle battery with adapted load parameters. DETAILED DESCRIPTION

[0066] The method according to the invention is described below using a vehicle battery as a generic device in a large number of motor vehicles. To this end, one or more electrochemical battery models are evaluated and parameterized in a central unit based on the operating variable profile.

[0067] The above examples represent a large number of fixed or mobile devices with energy supply devices independent of the power grid, such as vehicles (electric vehicles, electric assisted bicycles, etc.), facilities, machine tools, household appliances, Internet of Things devices, etc., which are connected to a central unit (cloud) outside the device via corresponding communication connections (such as LAN, Internet).

[0068] Figure 1 A system for collecting fleet data in a central unit 2 is shown, wherein the fleet data is used to create and run and evaluate an electrochemical battery model for modeling the internal battery state of a vehicle battery and an aging state model for determining the current and predicted aging state of a vehicle battery in a motor vehicle.

[0069] Figure 1 A fleet 3 of multiple motor vehicles 4 is shown. One of the motor vehicles 4 is Figure 1 The motor vehicles 4 each have a vehicle battery 41 with battery cells 45, an electric drive motor 42 and a control unit 43. The control unit 43 is connected to a communication device 44 which is suitable for transmitting data between the respective motor vehicle 4 and the central unit 2 (the so-called cloud).

[0070] The control unit 43 is designed in particular to record detected operating variables of the vehicle battery 41 with a high time resolution of, for example, between 1 Hz and 50 Hz, for example 10 Hz, by means of a battery management system 46 and to transmit them to the central unit 2 via a communication device 44 .

[0071] The motor vehicle 4 sends operating variables F to the central unit 2, which at least describe variables that influence the aging state of the vehicle battery 41 or are influenced by the aging state of the vehicle battery 41 and are required for determining the internal battery state, the aging state, the parameterization of the electrochemical battery model. In the case of a vehicle battery, these operating variables F can describe the instantaneous battery current, the instantaneous battery voltage, the instantaneous battery temperature and the instantaneous state of charge (SOC: State of Charge) in the case of a vehicle battery, both at the pack level, the module level and / or at the cell level. The operating variables F are recorded as operating variable curves in a fast time grid of 0.1 Hz to 50 Hz and can be regularly transmitted to the central unit 2 in uncompressed and / or compressed form. For example, the time series can be transmitted to the central unit 2 in a block-by-block manner at intervals of 10 minutes to several hours by using a compression algorithm for the purpose of minimizing the data traffic to the central unit 2.

[0072] The central unit 2 has a data processing unit 21 , in which a part of the method described below can be executed, and a database 22 for storing data points, model parameters, states and the like.

[0073] The central unit 2 is designed to receive the operating variable curve F. The central unit 2 can determine the current aging state in a manner known per se from the operating variable curve of the respective vehicle battery 41, for example by means of the aging state model described below. Furthermore, the central unit 2 can derive a load pattern from the operating variable curve and generate an artificial operating variable curve, which will result in an actual load (load intensity) of the vehicle battery corresponding to the historical load of the vehicle battery.

[0074] For example, Figure 2 The functional structure of an embodiment of a data-based aging state model 9 designed in a hybrid manner is schematically shown. The aging state model 9 comprises a physical aging model 5 and a data-based correction model 6 .

[0075] The physical aging model 5 is a mathematical model based on nonlinear differential equations. The evaluation of the physical aging model 5 of the aging state model 9 using the course of operating variables, in particular the course of operating variables since the beginning of the service life of the vehicle battery, results in the following internal states of the system of physical differential equations that correspond to the physical internal states of the vehicle battery. Since the physical aging model is based on physical and electrochemical laws, the model parameters of the physical aging model are variables that describe physical properties.

[0076] Therefore, the time series of the operating variables x(t) of the vehicle battery pack 41 to be evaluated is directly included in the physical aging state model 5, wherein the physical aging state model is preferably implemented as an electrochemical model and the corresponding internal electrochemical states z(t), such as layer thickness (e.g. SEI thickness), changes in cyclable lithium due to anode / cathode side reactions, rapid consumption of electrolyte, slow consumption of electrolyte, loss of active materials in the anode, loss of active materials in the cathode, etc., are modeled in a multidimensional state vector with the aid of nonlinear differential equations.

[0077] The physical aging model 5 thus corresponds to an electrochemical model of the relevant vehicle battery 3. This model determines the internal physical battery state z(t) as a function of the operating variable course x(t) in order to describe a physically based aging state SOHph=Phys[x(t)] that depends on at least one dimension of these electrochemical states z(t), wherein the electrochemical state z(t) can be linearly or nonlinearly mapped to a capacity retention rate (SOH-C) and / or an internal resistance increase rate (SOH-R) in order to provide them as aging states (SOH-C and SOH-R).

[0078] However, the model values ​​for the physical aging state SOHph provided by the electrochemical model are inaccurate in certain cases, and therefore it is provided that the model values ​​are corrected by a correction variable k. The correction variable k is provided by a data-based correction model 6, wherein the data-based correction model is trained with the aid of a training data set determined by the test bench 1.

[0079] Preferably, the data-based correction model is equivalent to a Gaussian process model Where μ is an m x 1 dimensional vector that describes the predicted mean at evaluation time point t, and ∑ is the m x m dimensional prediction covariance matrix for the Gaussian process model. The formulas for the mean and covariance of the Gaussian process are as follows:

[0080]

[0081] Where N corresponds to the number of labels is of mxd dimensions and describes the set of m new points in the input space, where and y correspond to N x 1 dimensional vectors, respectively, and wherein y describes the aging state measured or determined as a label, respectively. k corresponds to an N x m dimensional matrix of kernel evaluations describing the correlations encoded in the Gaussian process kernel between the m new points and the N measured points. C corresponds to an N x N dimensional matrix of kernel evaluations between the N measured points, and c corresponds to an m x m matrix of kernel evaluations between the m new points. T denotes transpose. See also Bishop, "Pattern Recognition and Machine Learning", 2006.

[0082] The correction model 6 receives an operating characteristic m(x(t)) on the input side, wherein the operating characteristic is derived from the course x(t0 . . . t n ) is determined for a future evaluation time point and may also include one or more of the internal electrochemical states of the differential equation group of the physical model 5. In addition, the correction model 6 can obtain the physical aging state SOHph obtained according to the physical aging model 5 on the input side. The operating characteristics m(t) of the current evaluation time point t are generated in the feature extraction block 8 based on the operating parameter change process x(t0…t). The correction model 6 is also provided with the internal state z(x(t)) of the state vector from the electrochemical physical aging model 5 and advantageously the physical aging state SOHph. The characteristic vector m(x(t)) is robust because it does not depend on the model quality of the physical aging model or its state. Therefore, considering the characteristic vector m(x(t)) is a valuable supplement to the internal state z(x(t)) of the physical or electrochemical aging model.

[0083] According to the predicted or predetermined course of the operating variable x(t0 ... t n ), an operating characteristic m(x(t)) can be generated in the central unit 2 for each fleet 3 or in other embodiments also in the corresponding motor vehicles, which operating characteristic is related to the evaluation time interval. For the determination of the aging state, the evaluation time interval between two evaluation time points can be a few hours (e.g. 6 hours) to several weeks (e.g. 1 month). A common value for the evaluation time interval is one week.

[0084] These operating characteristics m(x(t)) can be, for example, characteristics that are dependent on the evaluation time interval and / or cumulative characteristics and / or statistical variables that have been determined during the entire service life to date. In particular, these operating characteristics can include, for example: electrochemical state, such as SEI layer thickness, changes in cyclable lithium due to anode / cathode side reactions, rapid absorption of electrolyte solvents, slow absorption of electrolyte solvents, lithium deposition, loss of anode active material and loss of cathode active material, information about impedance or internal resistance, histogram characteristics, such as temperature at charge state, charging current at temperature and discharge current at temperature, in particular multi-dimensional histogram data about the battery pack temperature distribution at charge state, charging current distribution at temperature and / or discharge current distribution at temperature, current throughput in ampere hours, accumulated total charge (Ah), average capacity increase during charging (in particular for charging processes with a charge increase above a threshold share of the total battery pack capacity [e.g. 20% ΔSOC]), charge capacity and extreme values ​​(maximum values) of the differential capacity during a measured charging process with a sufficiently large charge state deviation (smooth curve of dQ / dU: charge change divided by change in battery pack voltage) or accumulated driving power. These variables are preferably converted in such a way that they characterize the actual usage behavior as well as possible and are normalized in the feature space. These operating characteristics m(t) can be used in their entirety or only in part for the method described below.

[0085] In order to determine the corrected aging state SOH to be output, the outputs SOHph,k of the physical aging model 5 and of the data-based correction model 6 are applied to one another, wherein the correction model is preferably implemented as a Gaussian process model. In particular, these outputs SOHph,k can be added or also multiplied in a summation block 7 (not shown) in order to obtain the modeled aging state SOH to be output at the current evaluation time interval. The confidence of the Gaussian process can also be used as a confidence of the corrected aging value SOH to be output of the hybrid model in the case of addition. The confidence or confidence value of the Gaussian process model thus characterizes the modeling uncertainty of the mapping of the operating characteristic points to the aging state.

[0086] The initial training of the hybrid aging state model 9 is carried out in the test bench 1. For this purpose, a training data set is created, which assigns the operating parameter changes of the vehicle battery pack operated according to the load curve to the aging states determined empirically or based on the model as labels. The goal is to enable the hybrid aging state model to accurately predict nonlinear, cyclic capacity decline. This can be achieved by combining the effect chain (Wirkkette) of electrochemical modeling and machine learning.

[0087] For example, a basic model can be provided for the training of a hybrid or data-based aging state model for determining the aging state as a signature, wherein according to the basic model, the SOH-C measurement is performed under defined ambient conditions by coulomb counting or by forming a current integral over time during the charging process, wherein the current integral over time is divided by the offset of the charge state between the beginning and the end of the relevant charging and / or discharging phase. If during the charging process the vehicle battery reaches a fully charged state from a fully discharged charge state starting from a defined relaxation state under reproducible load and ambient conditions, for example, a sufficiently reliable statement about the aging state can be obtained. The maximum charge thus detected can be linked to the initial maximum charge capacity of the vehicle battery. The aging state related to resistance (SOH-R value) can also be calculated from the voltage change related to the current change. They are usually related to a defined time interval as well as defined ambient conditions and the energy flow direction of the system.

[0088] The determination of the aging state of the tag can be carried out in a manner known per se under defined load conditions and ambient conditions for producing the tag, for example at constant temperature, constant current, etc., by evaluating the operating variable profile with an additional aging model. For this purpose, other models can be used for determining the aging state. The data-based correction model can be trained in a conventional manner based on the training data set and the residuals of the modeled aging state.

[0089] One object of the method described below is to adapt, after determining an anomaly in a vehicle battery that leads to increased aging or degradation of the vehicle battery, load parameters that characterize or determine the further operation of the vehicle battery in order to reduce further aging of the vehicle battery. As a result, the degradation of the vehicle battery 41 with the anomaly can be reduced to such an extent that predetermined guarantee conditions (defined by aging state thresholds and service life) can be complied with despite the anomaly.

[0090] In order to determine an artificial operating variable curve for predicting the aging state, a load variable L can first be generated. For this purpose, a data-based usage mode model can be set, for example. The usage mode model 10 converts a predetermined load pattern N into a curve of the load variable L, which reflects the load of the device battery pack as described in the usage and operating mode described by the usage mode. The load pattern N characterizes the load parameters of the previous use and load of the vehicle battery pack 41. The load patterns can be predetermined and determined in such a way that these load patterns, when used together with the usage mode model 10, result in the output of the time curves of the battery pack current I and the battery pack temperature T as the load variable L. With the help of a suitable known battery pack dynamic model 11, the curves of the battery pack voltage U and the state of charge SOC can be determined based on the curve of the load variable L, so as to generate an artificially generated curve of the operating variables together with the load variable. The usage mode model 10 and the battery pack dynamic model 11 can be operated according to the aging state.

[0091] The load pattern N can be defined by load parameters, which are derived from historical operating variable curves and / or operating characteristics for previous operations alone. The load pattern can also be adapted by changing one or more load parameters in order to implement desired measures for extending the remaining service life of the vehicle battery 41.

[0092] The load pattern can be evaluated by a usage pattern model, preferably by means of a data-based method, and used to describe the typical usage behavior of individual users with respect to the relevant vehicle battery.

[0093] The usage pattern model 10 can be designed as a recurrent neural network, for example, as an LSTM or GRU, in particular as a Bayesian LSTM network, and is trained based on a curve of a load variable L or an operating variable F that illustrates a typical and representative usage type of a vehicle battery pack. The curve of the load variable L or the operating variable F to be considered here should be based on a time period of the same usage type and the same operation type of the vehicle battery pack 41.

[0094] The load parameters that describe the load pattern N then correspond to the input variables of the usage pattern model 10 and characterize the driving behavior of the driver or the type of use of the vehicle. The usage pattern model 10 is trained in such a way that, given a specified load pattern N and a calendar time specification, an operating variable curve is predicted, which, in the case of a suitable evaluation, results in a load parameter that corresponds to the specified load pattern N or can be derived from the operating variable curve in a corresponding evaluation.

[0095] For evaluation purposes, the usage mode model 10 can be applied on the input side with calendar time specifications and load parameters of a previously used load mode or changed load parameters of a changed load mode and output an artificial curve of the load variables (current, temperature, preferably as a time series) on the output side.

[0096] The usage pattern model 10 can be designed as an overall model or with a sub-model for each curve to be predicted for the load variable L. In this case, one sub-model of the usage pattern model 10 is trained with regard to the individual time-temperature curve of the driver, and another sub-model of the usage pattern model 10 is trained with regard to the individual current curve of the driver. Thus, by predefining calendar time information, such as date and clock time, and the expected load parameters, artificial curves of load variables and / or operating variables can be generated. The calendar time information can also contain information about the day of the week, the month and about public holidays, and in particular take into account seasonality by feature engineering.

[0097] As long as no driver change is detected, the usage pattern model 10 can be trained regularly, for example once a month, taking into account new data. In this case, the training data set can usually include the curves of the operating variables or load variables from the last identified driver change to the latest recorded operating variables or load variables. Usually, in the middle of this section, the training data set is decomposed. A time period that is not used for training but is used to validate the usage pattern model. Typically, a Bayesian LSTM network is used to characterize the current profile individually for the driver. In addition, a Bayesian LSTM network is used to characterize the temperature profile individually for the driver. Alternatives to the LSTM approach are the Transformer model based on multi-head attention and the ARIMAX model.

[0098] The usage pattern model 10 can be formed directly from the raw data of the curves of the load variables L and / or the operating variables F. For the vehicle battery 41, typical patterns of current distribution are thus identified and reproducible, for example due to repeated commuting routes and typical idle and rest times and for loads of temperature ranges.

[0099] Therefore, the load profile may describe the type of loading of the vehicle battery 41 based on the load parameters, in particular cyclic loading.

[0100] The load pattern N is predefined by the vehicle individually and characterizes the type of use and operation or the use and operating behavior of the respective vehicle battery 41 based on the load parameters.

[0101] The use of the usage pattern model 10 represents only one possibility for generating an artificial operating variable curve from the load parameters of the previous load of the vehicle battery pack 41. In the simplest case, time periods of the past operating variable curves can be combined together part by part to generate an artificial operating variable curve for prediction. For example, an artificial operating variable curve can be generated by the comparable past load of the vehicle battery pack 41, wherein for each day of the week, the operating variable curve of the previous corresponding (same) day of the week that is closest to the expected load parameters of the day of the week with respect to the observed load parameters is copied. In this case, load parameters or stress factors are taken into account, which include, for example, the average or maximum ampere-hour throughput, the upper and lower limit values ​​of the state of charge (charging is limited by the upper limit of the state of charge, and discharging is limited by the lower limit of the state of charge), the maximum and / or average battery pack temperature, the maximum discharge current, and the selection of a specific charging curve with the maximum charging current.

[0102] The method for determining measures for extending the service life of the vehicle battery pack 41 with an abnormality is executed in the data processing device of the central unit 2 and is used below. Figure 3 The flowchart in is described in more detail.

[0103] In step S1 , firstly an operating variable profile is continuously transmitted from each of the vehicles 4 to the central unit 2 .

[0104] The operating variable curve can be evaluated in step S2 in order to determine an abnormality of the vehicle battery pack. For this purpose, methods known per se, such as those disclosed in DE 102022202111 A1 and DE 102022203343A1 in the form of methods based on autoencoders, can be used. Alternative methods are known, for example, from DE 102022202112 A1. Other methods include PCA or variational autoencoder methods and normalized streams. Reference is made to the prior art in publications US20220101666A1, DE102021211158A1, DE102022203343A1, DE102021205879A1, US20230280406A1 and DE102020214234A1.

[0105] In step S3, it is checked whether there is an abnormality in the vehicle battery pack 41. If an abnormality is determined (decision: yes), the method continues with step S4, otherwise it jumps back to step S1.

[0106] In step S4, the following ranges of load parameters or stress factors are defined, which can be used as a basis for the future operation of the vehicle battery pack 41 with an abnormality. By means of the load parameters, upper and lower limits of the state of charge or upper and lower limits of the battery pack voltage at idling, a derating temperature (maximum temperature) for thermal management and a plurality of charging curves (each with a predetermined maximum charging current) for carrying out the charging process, as well as a maximum discharge current can be set. Since an abnormality has been determined, the goal is to mitigate one or more of these load parameters / stress factors, i.e., to change the value of the load parameter accordingly, so that the further operation of the vehicle battery pack is limited by the load parameter.

[0107] With the aid of a suitable optimization method, an artificial operating variable curve is predicted based on different load patterns, wherein the load patterns take into account the corresponding combination of load parameters. For example, charging and discharging occur within the limits of the state of charge, so that the artificial operating variable curve is designed accordingly. The operation of the vehicle battery pack is also designed so that the battery pack temperature remains below the temperature value predetermined by the derating temperature. In addition, a charging curve can be selected for performing the charging process. The charging curve usually predetermines the maximum charging current depending on the corresponding state of charge. For these driving cycles, the maximum discharge current can be set according to the load parameters related to this. Other load parameters that influence the degradation of the vehicle battery pack can also be considered and optimized in the optimization method.

[0108] In step S4, an optimization method is performed, in which the load pattern is changed in a simulated manner and an artificial operating variable profile is determined for the future vehicle operation. The remaining service life is optimized, which results from the predicted aging state profile. The optimization method iteratively solves the problem of optimizing the service life maximization, i.e. the time period until a predetermined limit aging state is reached, for the changes in the individual load parameters. The optimization method can be performed based on a gradient descent method.

[0109] The purpose of the optimization is to achieve an extended service life by adapting the load parameters, i.e., to extend the remaining service life of the vehicle battery pack with an abnormality. To this end, an artificial operating variable curve is iteratively generated and the aging state curve is predicted from it, for example, using the above-mentioned method. The aging state curve is analyzed as follows to determine the time point at which the extreme aging state is reached. This time point marks the end of the guaranteed service life and is used as a decisive point for calculating the remaining service life.

[0110] Figure 4Schematically, the curves of the predicted aging states with an abnormal vehicle battery 41 (dash-dotted curve) and with adapted load parameters (dotted curve) are shown as aging states SOH over a calendar service life t (in years). In the case of an adaptation of the load parameters, an extension of the service life is detected. When a hybrid aging state model 9 is used, the aging state curve can be described with a confidence interval K.

[0111] For this purpose, the optimization method changes the load parameters, for example according to a gradient descent method, in order to obtain an optimal load pattern with optimized load parameters for extending the service life.

[0112] As an optimization method, Bayesian optimization can be applied, for example. Here, for various combinations of the load parameters listed above, a probabilistic model is created to determine the remaining service life as a function of the load parameters. Evaluation of the model then results in optimized load parameters for the further operation of the vehicle battery pack.

[0113] By means of the optimization, individual load parameters which have the greatest possible influence on the extension of the remaining service life can be identified. In particular, the one or more identified load parameters can be optimized again after having been identified while maintaining the values ​​of the remaining load parameters, so as to adapt the operating strategy according to the reoptimized identified load parameters.

[0114] The determined optimized load parameters can now be transmitted in step S5 to the relevant vehicle in order to restrict the operation of the vehicle battery 41 there during further use of the vehicle battery 41 according to the adapted operating strategy.

[0115] During the optimization method, additional conditions can be taken into account, in particular specifications can be taken into account for the determined load parameters, for example: the charging of the vehicle battery 41 must always be carried out using a fast charging process. In particular, if the aging state model 9 is designed as probabilistic as described above, corresponding confidence intervals can be taken into account when reaching the extreme aging state. Thus, the remaining service life resulting from reaching the extreme aging state can be stated with the determined predicted probability.

[0116] The above method can be repeated at regular or predetermined times to check whether the limitation caused by the change in the load parameters leads to a corresponding reduction in aging. If necessary, the load parameters can be further adapted by re-running the optimization method, for example, in order to try to comply with the guarantee limits in the case of deviations from the previously predicted aging curve. This can also be performed if it is determined that the aging is significantly slower than predicted and the predicted remaining service life of the vehicle battery 41 is therefore significantly above the guarantee limit.

[0117] According to another embodiment, at least one load parameter for the optimization can also be determined based on the historical use of the vehicle battery pack. For example, if the driver drives long distances with multiple charging stations, he relies on the highest possible state of charge deviation, so that the load parameters that specify the lower and upper state of charge should be changed as little as possible or not changed at all. In other cases where the user frequently charges his vehicle with a fast charging process, changes in the charge profile may not be taken into account in the optimization, wherein there may not be sufficient user acceptance in the limitation of the charge profile.

[0118] Those vehicle battery packs 41 to which the load parameters or operating strategies changed according to the above method are then monitored, wherein in particular the actual aging state curve (evaluated using one of the above calculation models) is compared with the previously simulated predicted aging state curve. If there is a deviation between the actual aging state curve and the predicted simulated aging state curve, the above optimization method can be re-executed.

Claims

1. A method for operating a device battery (41) in a technical device (4), in particular after an abnormality (S3) of the device battery (41) has been detected, the method comprising the following steps: - providing (S1) a time-dependent operating variable curve (F) of a plurality of operating variables of the determined device battery (41); - predicting the temporal operating variable curve (F) based on the provided temporal operating variable curve (F) and a plurality of load parameters of the load pattern with the aid of a usage pattern model (10) to obtain an artificial operating variable curve (F), wherein the usage pattern model (10) is designed to provide an artificial curve of at least one operating variable based on the plurality of load parameters, wherein the plurality of load parameters correspond to stress factors for aging of the device battery pack (41); - determining the remaining useful life (RUL) of the device battery (41) based on the artificial operating variable curve (F) with the aid of an aging state model (9); - performing (S4) an optimization method for adapting or determining a plurality of load parameters of the load pattern in order to determine optimized load parameters, wherein the remaining useful life (RUL) can be increased or maximized by the optimized load parameters when the optimized load parameters are taken into account in an operating strategy during future operation of the technical installation (4); - operating (S5) the device battery (41) according to at least one of the optimized load parameters.

2. The method according to claim 1, wherein: Based on the time operating variable curve, it is checked whether there is the following abnormality in the device battery pack (41), which will cause the remaining service life of the device battery pack (41) to be less than the guaranteed service life of the device battery pack (41), wherein the optimization method is only executed when the abnormality is determined.

3. A method according to claim 1 or 2, wherein the remaining service life is determined with the aid of an aging state model (9) for modeling the aging state of the device battery pack (41) based on a time series, wherein a temporal aging state curve is determined based on the provided temporal operating variable curve and the artificial operating variable curve, wherein the remaining service life (RUL) is determined depending on the time point when the aging state curve reaches a predetermined limit aging state.

4. The method according to any one of claims 1 to 3, wherein: The plurality of load parameters include at least one of the following limitations: an upper limit of the state of charge, a lower limit of the state of charge, a limit on the maximum battery pack temperature, a limit on the average battery pack temperature, a maximum discharge current, an average discharge current during a driving cycle, and a selection of a specific charging curve having a maximum charge current.

5. The method according to any one of claims 1 to 4, wherein: The artificial operating variable curve is created based on the load pattern (N) with the aid of a data-based usage pattern model (10), wherein the usage pattern model (10) is trained in such a way that one or more curves of the operating variable are predicted when a plurality of load parameters of the load pattern (N) and in particular calendar time information are predefined, and the one or more curves of the operating variable, when appropriately evaluated, result in load parameters corresponding to the predefined load pattern (N).

6. The method according to any one of claims 1 to 5, wherein for optimization, one of the at least one load parameter is excluded depending on the usage behavior of the device battery pack (41), wherein the usage behavior is determined based on a provided operating variable curve.

7. The method according to any one of claims 1 to 5, wherein: The aging state model (9) is designed in a data-based manner with the aid of a probabilistic model, wherein the optimization method is a Bayesian optimization method.

8. A method according to any one of claims 1 to 7, wherein the aging state model (9) is designed in a data-based manner with the aid of a probability model, wherein the determination of the remaining useful life (RUL) is performed based on a prediction of a time point at which a predetermined aging state limit is reached with a predetermined probability, in particular a predetermined probability between 90% and 99.999%, wherein the predetermined probability can be derived by integration of a probability density, wherein the probability density is derived from a probabilistic prediction model.

9. The method according to any one of claims 1 to 8, wherein the method is performed periodically or continuously.

10. The method according to any one of claims 1 to 9, wherein the aging state model (9) and the usage pattern model (10) are retrained regularly.

11. Apparatus for carrying out the method according to any one of claims 1 to 10.

12. Computer program product comprising instructions which, when the program is executed by at least one data processing device, cause the data processing device to perform the steps of the method according to any one of claims 1 to 10.

13. A machine-readable storage medium comprising instructions which, when executed by at least one data processing device, cause the data processing device to perform the steps of the method according to any one of claims 1 to 10.

Images