Method for determining at least one cell characteristic parameter of a cell system, cell system and computer program

Abstract

The invention relates to a method for determining at least one cell characteristic parameter, in particular an aging characteristic parameter, of a cell system (30) having at least one electrochemical cell (32). According to the invention, a cell characteristic parameter is determined from an operating characteristic parameter of the cell system (30) on the basis of a physical model of the cell system (30).

Description

Summary of the invention

[0001] A method for determining at least one cell characteristic parameter of a cell system having at least one electrochemical cell has been proposed. Summary of the invention

[0003] The invention is based on a method for determining at least one cell characteristic parameter, in particular for determining an aging characteristic parameter of a cell system having at least one electrochemical cell.

[0004] It is proposed to determine the cell characteristic parameters according to the operating characteristic parameters of the cell system based on the physical model of the cell system.

[0005] By means of the design of the method according to the invention, at least one cell characteristic parameter or a plurality of cell characteristic parameters can be determined in a computational manner and in particular without sensors based on the operating characteristic parameters of the cell system. Advantageously, at least one cell characteristic parameter can be determined with particularly low sensor technology overhead. Advantageously, the cell system can be operated particularly efficiently, precisely and / or smoothly based on the determined cell characteristic parameters. For example, at least one aging state of the cell system can be advantageously determined.

[0006] Advantageously, the expected failure time of the battery cell system can be estimated. Advantageously, the time of maintenance of the battery cell system can be particularly accurately adapted to the maintenance requirements of the battery cell system. Advantageously, the manufacturing process of the battery cell system can be particularly accurately and efficiently adapted, in particular optimized, based on the at least one determined battery cell characteristic parameter.

[0007] The electrochemical cell is preferably designed as a solid oxide fuel cell. However, it is also conceivable that the electrochemical cell is configured as other fuel cells, for example, as a polymer electrolyte fuel cell, etc., an electrolyte cell or a storage cell. For example, it is also conceivable that the electrochemical cell is used in an electrolyzer as an alternative. It is conceivable that the cell system has only one electrochemical cell, especially the electrochemical cell mentioned above, or the cell system has a plurality of electrochemical cells, such as two, three or hundreds of electrochemical cells. Preferably, the electrochemical cells of the cell system having a plurality of electrochemical cells are combined into stacks and are especially connected in series. It is also conceivable that the cell system has a plurality of stacks formed by a plurality of electrochemical cells, which are preferably combined into modules or towers.

[0008] The physical model is preferably a simulation model of a battery system, which in particular describes the operation of the battery system in a steady state. Preferably, the physical model is based on a finite element method. The physical model in particular has clear reversibility about at least one battery characteristic parameter (preferably an aging characteristic parameter). It can be imagined that, at least by filtering out physically irrelevant results of the physical model, the physical model can have clear reversibility about at least one battery characteristic parameter (preferably an aging characteristic parameter). For example, a physical model according to a publication entitled "A distributed real-time model of degradation in a solid oxide fuel cell" (2016, Journal of Power Sources, 311, 175-181) by Zaccaria, Tucker and Traverso can describe the degradation of a battery system with a solid oxide fuel cell, wherein the ohmic portion of the internal voltage loss is provided with an aging characteristic parameter as a pre-factor. Alternatively, it is also conceivable, for example, that other resistances other than the ohmic part of the internal voltage loss of the battery system (especially concentration losses or polarization losses), or the total internal resistance of the battery system, are multiplied by an aging characteristic parameter as a pre-factor. In a preferred physical model, in particular the total specific surface resistance of the battery system is multiplied by the aging characteristic parameter. The aging characteristic parameter describes in particular the aging state of the battery system. For example, in the unaged state of the battery system, the aging characteristic parameter has a value of 1, while for the aged state of the battery system it preferably has a value greater than 1. The aging characteristic parameter is preferably a scalar.

[0009] The physical model has in particular at least one input variable and at least one output variable. Preferably, at least one input variable of the physical model has at least one operating characteristic parameter of the operating characteristic parameters of the battery system, in particular at least one input characteristic parameter of the battery system, preferably at least one operating parameter of the battery system. Preferably, at least one other input variable of the physical model is a battery characteristic parameter, preferably an aging characteristic parameter. Operating parameters are, for example, requested load current, temperature (in particular gas temperature), volume flow, air utilization, fuel utilization, gas composition, pressure, pressure difference, etc. Preferably, at least one output variable of the physical model includes at least one operating characteristic parameter of the battery system, in particular at least one output characteristic parameter of the battery system, preferably at least one performance characteristic parameter of the battery system. The performance characteristic parameter of the battery system is in particular the output voltage of the electrochemical cell, preferably the output voltage of a stack formed by a plurality of electrochemical cells. It is conceivable that the output variable of the physical model includes at least one other battery characteristic parameter of the battery system, in particular at least one internal characteristic parameter.

[0010] The internal characteristic parameter is, for example, an internal temperature characteristic parameter of the battery system. The internal temperature characteristic parameter is preferably the temperature of the battery system, preferably the temperature of at least one electrochemical cell or a stack formed by a plurality of electrochemical cells. The internal characteristic parameter is in particular a local field variable. Alternatively, it is conceivable that the internal characteristic parameter is, for example, a fluid distribution in the battery system, preferably an oxygen distribution or a hydrogen distribution, or a current density distribution in the battery system, etc. The physical model in particular represents a mapping that maps the input variables of the physical model (preferably at least one operating parameter of the battery system and an aging characteristic parameter of the battery system) to the output variables of the physical model (preferably at least one performance characteristic parameter of the battery system and in particular at least one other battery characteristic parameter of the battery system, preferably at least one internal characteristic parameter).

[0011] The battery system preferably includes a control unit. The control unit is particularly configured to execute a method for determining at least one battery cell characteristic parameter. "Setting" should be understood as specially configured, specially designed and / or specially equipped. The function that an object is set to determine should be understood as that the object implements and / or executes the specific function in at least one application state and / or operating state. The control unit particularly includes at least one processor and a storage element and an operating program stored on the storage element. The storage element is preferably configured as a digital storage medium, such as a hard disk, etc. The battery system particularly includes a housing, in which at least one electrochemical cell or a stack formed by a plurality of electrochemical cells is preferably arranged. It is conceivable that the control unit is at least partially arranged on the housing, in particular at least partially arranged in the housing. Alternatively or additionally, it is also conceivable that the battery system has an external unit, wherein the external unit includes at least a part of the control unit. The external unit can be, for example, a cloud, a server, etc. It is conceivable that the operating characteristic parameters are transmitted to the external unit, for example wirelessly or wiredly, wherein the external unit preferably determines at least one battery cell characteristic parameter of the battery system based on the operating characteristic parameters of the battery system based on a physical model. It is also conceivable that the external unit comprises the memory element of the control unit.

[0012] Furthermore, it is proposed that the operating characteristic parameter comprises at least one, in particular already mentioned, input characteristic parameter of the battery cell system. It is conceivable that the operating characteristic parameter, from which the battery cell characteristic parameter is determined based on a physical model of the battery cell system, comprises only one input characteristic parameter of the battery cell system or a plurality of input characteristic parameters of the battery cell system. Advantageously, the battery cell characteristic parameter can be determined using characteristic parameters of the battery cell system that can be detected particularly easily, in particular at least one input characteristic parameter of the battery cell system. Advantageously, the battery cell characteristic parameter can be determined with particularly low technical expenditure.

[0013] It is further proposed that the input characteristic parameter comprises at least one operating parameter of the battery cell system, in particular one already mentioned above. In particular, the input characteristic parameter corresponds to the operating parameter. It is conceivable that at least one operating parameter is a control variable or a controlled variable of the battery cell system, in particular an actual value or a target value of the controlled variable. Advantageously, the battery cell characteristic parameter can be determined using characteristic parameters of the battery cell system that can be detected particularly easily, in particular at least one operating parameter of the battery cell system. Advantageously, the battery cell characteristic parameter can be determined with particularly low technical expenditure. Advantageously, the battery cell characteristic parameter can be determined at least partially using the following characteristic parameters of the battery cell system, which are necessary for the operation of the battery cell system and are detected.

[0014] It is further proposed that the operating characteristic parameter comprises at least one output characteristic parameter of the battery cell system, in particular the one already mentioned above. It is conceivable that the operating characteristic parameter (the battery cell characteristic parameter is determined based on the physical model of the battery cell system) has only one output characteristic parameter of the battery cell system or a plurality of output characteristic parameters of the battery cell system. Advantageously, the battery cell characteristic parameter can be determined using characteristic parameters of the battery cell system that are particularly easy to detect, in particular at least one output characteristic parameter of the battery cell system. Advantageously, the battery cell characteristic parameter can be determined with particularly low technical expenditure. Advantageously, the battery cell characteristic parameter can be determined at least partially using the following characteristic parameters of the battery cell system, which are necessary for the operation of the battery cell system and are detected.

[0015] Furthermore, it is proposed that the output characteristic parameter comprises at least one, in particular already mentioned, performance characteristic parameter of the cell system. In particular, the output characteristic parameter corresponds to the performance characteristic parameter of the cell system. Advantageously, the cell characteristic parameter can be determined using characteristic parameters of the cell system that can be detected particularly easily, in particular at least one performance characteristic parameter of the cell system. Advantageously, the cell characteristic parameter can be determined with particularly low technical expenditure. Advantageously, the cell characteristic parameter can be determined at least partially using characteristic parameters of the cell system that have already been detected during operation of the cell system.

[0016] Furthermore, it is proposed that the operating characteristic parameters are detected at least partially on the cell system. Preferably, at least a performance characteristic parameter of the cell system, in particular the output voltage, is detected on the cell system, preferably by means of a sensor. Preferably, the cell system has a detection unit for at least partially detecting the operating characteristic parameters. The detection unit is preferably connected to the control unit by data technology, in particular for wireless and / or wired data transmission. It is also conceivable that the control unit at least partially comprises the detection unit. It is conceivable that the operating characteristic parameters formed as input characteristic parameters of the cell system and / or the operating characteristic parameters formed as output characteristic parameters of the cell system are at least partially detected, preferably by means of a sensor, preferably by means of the detection unit. The detection unit comprises, for example, at least one sensor unit for at least partially detecting the operating characteristic parameters. The detection unit is preferably configured to detect at least an output characteristic parameter formed as a performance characteristic parameter, in particular by means of a sensor. The sensor unit comprises, for example, at least one voltage sensor, in particular for detecting a performance characteristic parameter of the cell system formed as an output voltage. It is conceivable that the detection unit is configured to detect at least an input characteristic parameter of the cell system formed as an operating parameter. Additionally or alternatively, it is conceivable that at least one input characteristic parameter, preferably an input characteristic parameter of a control variable or a controlled variable, in particular a target value of an input characteristic parameter of a controlled variable, is read directly by means of the control unit, preferably provided by the control unit. Advantageously, at least one cell characteristic parameter can be determined particularly accurately. Advantageously, the cell system can be operated particularly efficiently, precisely and / or smoothly based on the cell characteristic parameter that can be determined particularly accurately.

[0017] It is further proposed that a time series is formed based on the operating characteristic parameters. Preferably, in a method step, in particular in a detection step, the operating characteristic parameters (preferably at least one input characteristic parameter and at least one output characteristic parameter) are recorded over time. Preferably, the recorded operating characteristic parameters are stored as a time series on the control unit, in particular on a storage element of the control unit. Advantageously, the cell characteristic parameters can be calculated from the time series of the operating characteristic parameters at any time at different times. Advantageously, the cell characteristic parameters can be determined particularly conveniently. Advantageously, the cell characteristic parameters can be calculated and evaluated for different times.

[0018] Furthermore, it is proposed that, in order to determine the cell characteristic parameter, the physical model is inverted and optimized, in particular for the cell characteristic parameter. Preferably, in order to determine the cell characteristic parameter, which is in particular one of the input variables of the physical model, preferably in order to determine the aging characteristic parameter, the physical model is inverted and optimized, preferably for the cell characteristic parameter, in particular for the aging characteristic parameter. Preferably, in a method step, in particular in the selection step, a moment is selected from the time series of the operating characteristic parameters, preferably by a control unit, wherein the control unit is in particular arranged to determine at least one cell characteristic parameter, preferably an aging characteristic parameter, for the selected moment. Preferably, the control unit is arranged to select a moment from the time series of the operating characteristic parameters, so that the system state of the cell system associated with this moment can be described by the physical model. For example, it is conceivable that the physical model can be applied only to steady-state conditions, in particular to the input characteristic parameters of the cell system, so that correspondingly sufficiently steady-state operating characteristic parameters are used. It is also conceivable that the operating characteristic parameters are at least partially preprocessed, for example by filtering to remove oscillations, etc. It is conceivable that the operating characteristic parameters, in particular at least one input characteristic parameter and at least one output characteristic parameter, are at least partially combined. It is also conceivable that, for adapting to the physical model, the operating characteristic parameter associated with the time is at least partially converted, in particular by the control unit, preferably in the conversion step. Preferably, the operating characteristic parameter associated with the time is converted, if necessary, in particular in the conversion step, into an input variable and / or output variable of the physical model. For example, it is conceivable that the cell system has a plurality of stacks, each of which has a plurality of electrochemical cells, wherein an averaging is performed for the individual stacks, or the detected current of the cell system is converted into a current density. Based on the physical model, preferably, in particular in the determination step, at least one cell characteristic parameter is determined at a particularly previously selected time, for which the physical model predicts at least one preferably detected output characteristic parameter at a particularly previously selected time together with at least one input characteristic parameter at a particularly previously selected time. In particular, for determining the cell characteristic parameter, an optimization method is preferably applied in the determination step, such as Bayesian optimization, simulated annealing methods, gradient methods, etc., in particular involving an output characteristic parameter of the cell system detected for at least one input characteristic parameter at a particularly previously selected time and an output characteristic parameter predicted by the physical model for the input characteristic parameter at the selected time. Advantageously, at least one cell characteristic parameter can be determined from the operating characteristic parameters of the cell system in a sensorless manner. Advantageously, the cell characteristic parameter can be determined with particularly low expenditure in terms of sensor technology.

[0019] In addition, it is proposed to use a physical model to train a regression model. Preferably, the regression model is trained based on the results determined by means of the physical model, for example by means of a statistical experimental design. Preferably, the regression model is set as a substitute for the physical model, especially in an alternative determination step. The regression model is set in particular, preferably in an alternative determination step, to be used in an optimization method to determine at least one cell characteristic parameter or directly determine at least one cell characteristic parameter. In particular, it is conceivable that the regression model is trained so that the regression model directly determines at least one cell characteristic parameter of the cell system from operating characteristic parameters, preferably at least one input characteristic parameter and at least one output characteristic parameter, preferably in an alternative determination step, wherein in particular, preferably, the inversion of the physical model is eliminated at least in the determination step. In particular, a statistical experimental design is generated from a combination of the following operating characteristic parameters of the physical model, for example by means of Latin hypercube sampling, active learning, etc., and at least one cell characteristic parameter is determined based on the physical model for these operating characteristic parameters, wherein the regression model is trained using the results. The training of the regression model is based on, for example, linear regression, random forest, Gaussian process, neural network, interpretable boosting machine, etc. Preferably, in particular in the alternative determination step, at least one cell characteristic parameter is determined based on a regression model. Advantageously, the cell characteristic parameter can be determined particularly efficiently in terms of time and computation using a regression model.

[0020] Furthermore, it is proposed to train the regression model using active learning, in particular safe active learning. Active learning is in particular used to select information-optimal training data for training the regression model. Preferably, safe active learning is used to avoid training data for which the physical model does not converge, in particular diverges. Preferably, a convergence characteristic parameter is determined, for example, when determining at least one cell characteristic parameter. The convergence characteristic parameter is preferably an indicator of how close a solution method, in particular a numerical one, of the physical model is to non-convergence, in particular divergence. The convergence characteristic parameter can be determined, for example, based on the number of iterations of the solution method required to determine at least one cell characteristic parameter. Alternatively, it is also conceivable to determine the convergence characteristic parameter based on an output variable of the physical model, in particular based on at least one output characteristic parameter of the cell system and / or based on an internal characteristic parameter. Preferably, in an initialization step for initializing the training of the regression model, the following operating characteristic parameters of the cell system are selected for the system state of the cell system, for which the solution method of the physical model is known to converge. Based on the physical model, in particular by inversion and optimization of the physical model, preferably at least one cell characteristic parameter is determined for the operating characteristic parameter selected in the initialization step. Based on the result of the initialization step, in particular based on the operating characteristic parameters selected in the initialization step and the cell characteristic parameters determined in the initialization step, the value of the convergence characteristic parameter associated with the result is preferably determined. Based on the result from the initialization, a regression model preferably of probability is trained, preferably a Gaussian process model. Preferably, based on the value of the convergence characteristic parameter determined in the initialization step, a safety model, preferably a Gaussian process, is additionally trained, in particular a probabilistic one. After the initialization step, a loop is preferably performed to train the regression model. The loop preferably has a first loop step, in which the following operating characteristic parameters are selected for another system state of the cell system, the operating characteristic parameters preferably having a maximum value of the information content, and the solution method for the physical model converges with a probability under the operating characteristic parameters, which is in particular higher than a probability threshold. Particularly preferably, the probability threshold is 0.5. However, it is also conceivable that the probability threshold is between 0.1 and 0.9, preferably between 0.2 and 0.8, preferably between 0.3 and 0.7, and particularly preferably between 0.4 and 0.6. Furthermore, it is also conceivable, however, as an alternative, that the probability threshold lies outside the value range between 0.1 and 0.9. The information content is preferably evaluated by entropy, in particular in a Gaussian process where the entropy is proportional to the prediction variance. In particular, a restricted optimization problem is formed, in which the operating characteristic parameter is calculated for which the information content has a maximum value and the probability of convergence of the solution method for the physical model is greater than the probability threshold. The probability threshold is preferably settable by an operator, for example via an input unit. Preferably, the input unit is connected or connectable to the control unit by data technology, in particular for wireless and / or wired data transmission.The input unit can be configured, for example, as a keyboard, a keypad, a touch screen, a knob, a slide adjuster, a button, etc. It is conceivable that the control unit or the external unit has an input unit. As an alternative to the probability of convergence of the solution method for the physical model for determining at least one cell characteristic parameter, a criterion can also be conceived that the probability of interference-free simulation with the help of the physical model is greater than a probability threshold. The probability of interference-free simulation with the help of the physical model preferably takes into account the convergence of the solution method for the physical model, storage requirements, processor requirements and simulation duration. Alternatively or in addition, the distance between the cell characteristic parameter determined by the physical model and the known physical boundary can also be used as an indicator of the probability of interference-free simulation. In the second loop step of the loop, the value of at least one cell characteristic parameter and the corresponding convergence characteristic parameter is determined using the operating characteristic parameter of another system state. In the third loop step of the loop, the regression model and / or the safety model are preferably updated according to the at least one cell characteristic parameter determined in the second loop step and / or according to the value of the convergence characteristic parameter determined for another system state of the cell system. If the termination criterion is not met after the third loop step, the loop restarts the first loop step. The termination criterion may be, for example, the exhaustion of the calculation time, the exhaustion of the number of evaluated system states of the cell system, the information content falling below an information content threshold or the maximum error on the validation data set falling below a threshold. It is conceivable that the information content threshold is adaptable, for example automatically by a control unit or manually by an operator via an input unit. It is also conceivable that multiple system states are selected when calculating the physical model and that they are calculated in parallel, wherein the entropy of the Gaussian process is proportional to the predicted covariance. Preferably, the optimization can be performed by mapping the predicted covariance to a one-dimensional variable with the aid of an optimality criterion, for example with the aid of a determinant, preferably with the aid of a maximum eigenvalue. It is also conceivable that, in particular when calculating multiple system states in parallel with the aid of a model, a greedy algorithm is applied. It is also conceivable that the calculation with the aid of the physical model is carried out in a staggered manner in time. Advantageously, the regression model is trained particularly efficiently. Advantageously, at least one cell characteristic parameter can be determined particularly efficiently.

[0021] Furthermore, it is proposed to determine the time course of the aging characteristic parameter of the cell system, in particular the one already mentioned above. Additionally or alternatively, it is also conceivable to determine the time course of at least one other cell characteristic parameter of the cell system, in particular the time course of at least one internal characteristic parameter of the cell system. Preferably, the aging characteristic parameter at a further time, in particular a plurality of further times, preferably from the time series of the operating characteristic parameter, is determined based on the physical model according to the operating characteristic parameter of the cell system, preferably similarly to the determination of the aging characteristic parameter at a previously selected time. Based on the determined aging characteristic parameter at different times, the time course of the aging characteristic parameter can be determined and in particular displayed. It is conceivable to extrapolate the time course of the aging characteristic parameter, in particular in order to predict the future course of the aging characteristic parameter. It is also conceivable that the time course of the aging characteristic parameter is correlated with at least one operating characteristic parameter of the cell system, which is preferably aging-related, and / or an internal characteristic parameter of the cell system. Advantageously, the aging rate of the cell system can be determined and monitored based on the time course of the aging characteristic parameter. Advantageously, a possible failure time of the cell system can be predicted.

[0022] Furthermore, it is proposed that at least one other cell characteristic parameter, in particular one already mentioned above, of the cell system is determined based on the determined cell characteristic parameter, preferably at least one internal characteristic parameter. Preferably, at least one other cell characteristic parameter, preferably at least one internal characteristic parameter of the cell system (which is in particular one of the output variables of the physical model) is determined based on a physical model based on an aging characteristic parameter determined based on the physical model, in particular by inverting and optimizing the physical model with respect to the aging characteristic parameter or by a regression model generated with the aid of the physical model. Advantageously, the cell system can be made to operate particularly efficiently, precisely and / or smoothly based on the determined other cell characteristic parameters. Advantageously, cell characteristic parameters that require particularly high sensor technology overhead when detecting can be determined particularly easily.

[0023] It is further proposed that at least one other cell characteristic parameter is determined in a spatially resolved manner. Preferably, the local distribution of the other cell characteristic parameter determined as an internal characteristic parameter is determined. The local distribution of at least one other cell characteristic parameter is in particular within the cell system, preferably within at least one electrochemical cell or within a stack formed by a plurality of electrochemical cells. Preferably, the local distribution of at least one other cell characteristic parameter is determined based on a physical model according to the operating characteristic parameters of the cell system, preferably by an optimization method for applying an inverted physical model with respect to aging characteristic parameters or by a regression model generated with the aid of a physical model. In particular, at least one other cell characteristic parameter, preferably an internal characteristic parameter, can be determined at a position in the cell system or at a position in a list. It is also conceivable to determine a combination of multiple other cell characteristic parameters of the cell system (preferably multiple internal characteristic parameters of the cell system) at the position in the cell system or at the position in the list, such as a product, a sum, a nonlinear function, etc. Furthermore, it is also conceivable to determine the derived value of at least one other cell characteristic parameter or the derived value of a combination of multiple other cell characteristic parameters, such as the maximum value, minimum value, average value in a partial volume or the total volume of the cell system, for describing the distribution, such as the first-order area moment, the second-order area moment, the support points of the cumulative distribution density, etc., the gradient or the rate of change. Advantageously, the knowledge of the local distribution of at least one other cell characteristic parameter allows, for example, to improve the efficiency and dynamics of the cell system by using the optimization of the operating strategy of the cell system. Advantageously, a particularly smooth, precise and efficient operation of the cell system can be achieved.

[0024] It is further proposed that at least one further cell characteristic parameter comprises a temperature distribution in the cell system. It is conceivable that at least one further cell characteristic parameter corresponds to a temperature distribution in the cell system. Advantageously, knowledge of the temperature distribution in the cell system, in particular in a stack or electrochemical cell formed from a plurality of electrochemical cells, allows ensuring compliance with predefined limits of the distribution, such as a minimum temperature and / or a maximum temperature, in order to prevent local damage in the cell system. Advantageously, a particularly safe operation of the cell system can be achieved. Advantageously, a particularly durable cell system can be provided.

[0025] It is further proposed that at least one other cell characteristic parameter comprises a fluid distribution in the cell system, in particular an oxygen distribution and / or a hydrogen distribution. It is conceivable that at least one other cell characteristic parameter corresponds to a fluid distribution in the cell system, in particular an oxygen distribution and / or a hydrogen distribution. Advantageously, knowledge of the fluid distribution, in particular the oxygen distribution and / or the hydrogen distribution, allows ensuring compliance with preset limits of the distribution, such as a minimum concentration, in order to prevent local damage in the cell system. Advantageously, a particularly safe operation of the cell system can be achieved. Advantageously, a particularly durable cell system can be provided.

[0026] It is further proposed that at least one other cell characteristic parameter has a current density distribution in the cell system. It is conceivable that at least one other cell characteristic parameter corresponds to the current density distribution in the cell system. Preferably, based on the physical model, a plurality of cell characteristic parameters can be determined according to the operating characteristic parameters of the cell system, in particular the temperature distribution in the cell system, the fluid distribution in the cell system, the current density distribution in the cell system and / or the aging characteristic parameters of the cell system. Advantageously, knowledge of the current density distribution in the cell system allows ensuring compliance with preset limits of the distribution, such as a maximum current density, in order to prevent local damage in the cell system. Advantageously, particularly safe operation of the cell system can be achieved. Advantageously, a particularly durable cell system can be provided.

[0027] In addition, it is proposed to use a physical model to generate training data for a fault recognition model. In particular, for at least one other cell characteristic parameter, preferably at least one internal characteristic parameter, a critical value range is preset, which is stored, for example, on a storage element of the control unit. It is conceivable that the critical value range is adaptable, for example, automatically by the control unit or manually by an operator via an input unit. The physical model can be used in particular to determine the following operating characteristic parameters, for which at least one other cell characteristic parameter, especially an internal characteristic parameter, is located in a critical value range. For example, a database can be determined with the help of a physical model, which has operating characteristic parameters and associated values ​​of at least one other cell characteristic parameter (preferably an internal characteristic parameter) determined in particular according to the physical model. Preferably, the operating characteristic parameters can be used in a targeted manner with the help of the database to train the fault recognition model, wherein the respective associated values ​​of at least one other cell characteristic parameter are located or not in the critical value range. Preferably, the fault recognition model is trained by machine learning, in particular for performing a classification method to classify the determinable fault states of the cell system. The training of the fault recognition model is based on, for example, random forests, support vector machines, neural networks, etc. Preferably, the fault recognition model is used during the operation of the battery cell system to recognize and especially classify fault states in the battery cell system based on operating characteristic parameters of the battery cell system detected during operation, for example, by associating with at least one other battery cell characteristic parameter, preferably at least one internal characteristic parameter. For example, in the case of multiple other battery cell characteristic parameters of the battery cell system, especially multiple internal characteristic parameters, it is conceivable to train the fault recognition model so that the fault recognition model recognizes which of the multiple other battery cell characteristic parameters, especially multiple internal characteristic parameters of the battery cell system are within the corresponding critical value range during operation based on the operating characteristic parameters of the battery cell system. Advantageously, the fault recognition model can be trained particularly accurately and efficiently. Advantageously, faults in the battery cell system can be identified particularly simply and reliably. Advantageously, the association of faults in the battery cell system with at least one battery cell characteristic parameter can be achieved.

[0028] Furthermore, a cell system, in particular as already mentioned above, is proposed, which has at least one electrochemical cell, in particular as already mentioned above, and at least one control unit, in particular as already mentioned above, for carrying out the method according to the invention. Advantageously, a particularly durable cell system can be provided. Advantageously, a cell system can be provided, which allows particularly precise and convenient maintenance planning.

[0029] Furthermore, a computer program is proposed, which comprises instructions which, when the computer program is executed by a computer, cause the computer to execute the method according to the invention. The computer program is preferably stored on a storage element of a control unit. Alternatively, it is also conceivable that the computer program is stored on a portable data carrier, such as a USB stick, a portable hard disk, an optical data carrier (especially a CD, etc.), a cloud, a server, etc. In particular, the control unit instructs the computer to execute the computer program. Advantageously, the method for determining the cell characteristic parameters of a cell system can be particularly conveniently and flexibly adapted. Advantageously, the method can be applied to different cell systems and can be particularly conveniently adapted to the corresponding cell system.

[0030] The method according to the invention, the battery system according to the invention and / or the computer program according to the invention should not be restricted to the above-mentioned applications and embodiments in this case. In particular, the method according to the invention, the battery system according to the invention and / or the computer program according to the invention can have quantities of individual elements, components and units and method steps that differ from the quantities mentioned herein in order to realize the mode of action described herein. In addition, in the case of value ranges given in the present disclosure, the values ​​within the mentioned limits should also be regarded as disclosed and can be used arbitrarily. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Further advantages can be derived from the following description of the drawings. The drawings show embodiments of the invention. The drawings, the description and the claims contain many combinations of features. A person skilled in the art will also consider these features individually and combine them into other meaningful combinations.

[0032] Figure 1 A schematic diagram of a cell system according to the invention is shown, which has at least one electrochemical cell and at least one control unit for executing the method according to the invention for determining at least one cell characteristic parameter of a cell system,

[0033] Figure 2 shows a schematic flow chart of the method according to the present invention,

[0034] Figure 3 shows a schematic process for training a regression model based on a physical model, and

[0035] Figure 4 A schematic process for applying a fault identification model to a battery cell system is shown. DETAILED DESCRIPTION

[0036] Figure 1 A cell system 30 is shown having a plurality of electrochemical cells 32 (in Figure 1 Only one electrochemical cell 32 is shown by way of example). Alternatively, it is also conceivable that the cell system 30 has only one electrochemical cell 32. The electrochemical cells 32 are combined into stacks. It is also conceivable that the cell system 30 includes a plurality of stacks formed by a plurality of electrochemical cells 32, the stacks preferably being combined into modules or towers. The electrochemical cells 32 are connected in series. The electrochemical cells 32 are configured as solid oxide fuel cells. Alternatively, it is also conceivable that the electrochemical cells 32 are configured as other fuel cells, for example, as polymer electrolyte fuel cells, etc., electrolyte cells or storage cells. For example, it is also conceivable that the electrochemical cells 32 are used in electrolyzers as an alternative.

[0037] The battery system 30 has at least one control unit 34. The control unit 34 is configured to execute a method to determine at least one battery cell characteristic parameter of the battery system 30, in particular to determine an aging characteristic parameter of the battery system 30, and to determine a plurality of other battery cell characteristic parameters of the battery system 30, in particular to determine a plurality of internal characteristic parameters of the battery system 30. Alternatively, it is also conceivable that only the battery cell characteristic parameters of the battery system 30, in particular the aging characteristic parameters, can be determined by the method. The control unit 34 has at least one processor (not shown here) and a storage element (not shown here), as well as an operating program stored on the storage element. The storage element is configured as a digital storage medium, such as a hard disk or the like.

[0038] The cell system 30 comprises a housing (not shown here), in which a stack formed by a plurality of electrochemical cells 32 is arranged. It is conceivable that the control unit 34 is at least partially arranged on the housing, in particular at least partially arranged in the housing. Alternatively, it is also conceivable that the cell system 30 comprises an external unit (not shown here), wherein the external unit comprises at least a part of the control unit 34. The external unit may be, for example, a cloud, a server, etc. It is also conceivable that the external unit comprises a storage element of the control unit 34. A computer program 36 is stored on the storage element, which comprises instructions, which, when the computer program 36 is executed by a computer, prompt the computer to execute a method for determining cell characteristic parameters of the cell system 30. The control unit 34 has a computer. Alternatively, it is also conceivable that the computer program 36 is stored on a portable data carrier, such as a USB stick, a portable hard disk, an optical data carrier (in particular a CD, etc.), a cloud, a server, etc.

[0039] Figure 2 The schematic flow chart of a method for determining a cell characteristic parameter, especially an aging characteristic parameter, and determining a plurality of other cell characteristic parameters, especially a plurality of internal characteristic parameters, of a cell system 30 is shown. The cell characteristic parameter is an aging characteristic parameter of the cell system 30. A first cell characteristic parameter among the other cell characteristic parameters (especially the internal characteristic parameters) includes a temperature distribution in the cell system 30, and in particular the first other cell characteristic parameter corresponds to the temperature distribution in the cell system 30.

[0040] The second cell characteristic parameter of other cell characteristic parameters (especially internal characteristic parameters) includes the current density distribution in the cell system 30, and in particular, the second other cell characteristic parameter corresponds to the current density distribution in the cell system 30. The third cell characteristic parameter of other cell characteristic parameters (especially internal characteristic parameters) includes the fluid distribution in the cell system 30, and in particular, the hydrogen distribution, and in particular, the third other cell characteristic parameter corresponds to the hydrogen distribution in the cell system 30. The fourth cell characteristic parameter of other cell characteristic parameters (especially internal characteristic parameters) includes the fluid distribution in the cell system 30, and in particular, the oxygen distribution, and in particular, the fourth other cell characteristic parameter corresponds to the oxygen distribution in the cell system 30.

[0041] The cell characteristic parameters are determined based on the operating characteristic parameters of the cell system 30 based on the physical model of the cell system 30. It is conceivable that the operating characteristic parameters are transmitted to an external unit, for example, wirelessly or by wired connection, wherein the external unit preferably determines at least one cell characteristic parameter of the cell system 30 based on the operating characteristic parameters of the cell system 30 based on the physical model. The physical model is a simulation model of the cell system 30, which in particular describes the operation of the cell system 30 in a stable state. The physical model is based on the finite element method. The physical model has clear reversibility with respect to the cell characteristic parameters, in particular the aging characteristic parameters. It is conceivable that the physical model has clear reversibility with respect to the cell characteristic parameters, in particular the aging characteristic parameters, at least by filtering out physically irrelevant results of the physical model. For example, a physical model according to the publication of Zaccaria, Tucker and Traverso entitled: "A distributed real-time model of degradation in a solid oxide fuel cell" (2016, "Journal of Power Sources", 311, 175-181) describes the degradation of the battery system 30, wherein the ohmic component of the internal voltage loss is provided with an aging characteristic parameter as a prefactor. Alternatively, it is also conceivable, for example, that other resistances of the internal voltage loss of the battery system 30 that are different from the ohmic component (in particular concentration losses or polarization losses) or the total internal resistance of the battery system 30 are multiplied by the aging characteristic parameter as a prefactor. In a preferred physical model, the total specific surface resistance of the battery system 30 is multiplied by the aging characteristic parameter. The aging characteristic parameter describes the aging state of the battery system 30. In the unaged state of the battery system 30, the aging characteristic parameter has a value of 1, while for the aged state of the battery system 30, the aging characteristic parameter has a value greater than 1. The aging characteristic parameter is a scalar.

[0042] The physical model has multiple input variables and multiple output variables. Alternatively, it is also conceivable that the physical model has only one input variable and / or only one output variable. The input variables of the physical model are at least partially the operating characteristic parameters of the battery system 30. The operating characteristic parameters include multiple input characteristic parameters of the battery system 30, or alternatively include only one input characteristic parameter of the battery system 30. The input characteristic parameters of the battery system 30 are operating parameters of the battery system 30, which are exemplarily requested load current, temperature (especially gas temperature), volume flow, gas composition, pressure, pressure difference, etc. It is conceivable that the operating parameters are at least partially the control variables or controlled variables of the battery system 30, especially the actual value or target value of the controlled variable. The input variables of the physical model include aging characteristic parameters of the battery system 30. The input variables of the physical model include operating parameters of the battery system 30.

[0043] The operating characteristic parameter comprises at least one output characteristic parameter of the cell system 30. It is also conceivable that the operating characteristic parameter comprises a plurality of output characteristic parameters of the cell system 30. The output characteristic parameter of the cell system 30 comprises a performance characteristic parameter of the cell system 30. The performance characteristic parameter of the cell system 30 is the output voltage of the cell system 30, in particular the output voltage of a stack formed of a plurality of electrochemical cells 32.

[0044] The first output variable among the output variables of the physical model has at least one of the operating characteristic parameters of the battery system 30, in particular the output characteristic parameters of the battery system 30. The first output variable is a performance characteristic parameter of the battery system 30. The other output variables of the physical model correspond to other battery cell characteristic parameters of the battery system 30, in particular other battery cell characteristic parameters of the battery system 30, preferably the first battery cell characteristic parameter, the second battery cell characteristic parameter, the third battery cell characteristic parameter and the fourth battery cell characteristic parameter among the other battery cell characteristic parameters. The physical model especially represents a mapping, which maps the input variables of the physical model, preferably the operating parameters of the battery system 30 and the aging characteristic parameters of the battery system 30, to the output variables of the physical model, preferably to the performance characteristic parameters of the battery system 30 and other battery cell characteristic parameters of the battery system 30.

[0045] In one method step, in particular in the detection step 10, an operating characteristic parameter is at least partially detected on the battery system 30. The battery system 30 has a detection unit (not shown here) for at least partially detecting the operating characteristic parameter of the battery system 30. The detection unit includes a sensor unit for at least partially detecting the operating characteristic parameter. The detection unit is configured to detect at least an output characteristic parameter that is configured as a performance characteristic parameter, in particular in a sensor manner. The sensor unit includes, for example, at least one voltage sensor, in particular for detecting a performance characteristic parameter of the battery system 30 that is configured as an output voltage. The detection unit is connected to the control unit 34 in terms of data technology, in particular for wireless and / or wired data transmission. It is also conceivable that the control unit 34 at least partially includes the detection unit.

[0046] It is conceivable that the operating characteristic parameter formed as an input characteristic parameter of the cell system 30 can be detected at least partially by means of a detection unit, preferably by means of a sensor. Additionally or alternatively, it is also conceivable that the input characteristic parameter, in particular the input characteristic parameter formed as a control variable or a controlled variable, preferably a target value of the controlled variable, is read at least partially directly by means of the control unit 34 in the detection step 10, preferably provided by the control unit 34.

[0047] In the detection step 10, the operating characteristic parameters of the cell system 30, in particular the input characteristic parameters and the output characteristic parameters, are recorded over time. A time series is formed based on the recorded operating characteristic parameters. The recorded operating characteristic parameters are stored as a time series on the control unit 34, in particular on a storage element of the control unit 34.

[0048] In one method step, in particular in the selection step 12, a time is selected from the time series of the operating characteristic parameters, preferably by the control unit 34. The control unit 34 is configured to determine the cell characteristic parameters, in particular the aging characteristic parameters, for the selected time. The control unit 34 is configured to select a time from the time series of the operating characteristic parameters such that the system state of the cell system 30 associated with this time can be described by the physical model. For example, it is conceivable that the physical model can be applied only to steady-state conditions, in particular to the input characteristic parameters of the cell system 30, so that correspondingly sufficiently steady-state operating characteristic parameters are used.

[0049] It is conceivable that, for adaptation to the physical model, the characteristic operating parameters associated with the instant are at least partially converted in a method step, in particular in a conversion step 14, preferably by the control unit 34. The characteristic operating parameters associated with the instant are converted into input variables and / or output variables of the physical model, if necessary, in particular in the conversion step 14. It is also conceivable that the characteristic operating parameters are at least partially preprocessed, for example by filtering to remove oscillations or the like. Furthermore, it is also conceivable that, in particular in the conversion step 14, the characteristic operating parameters are at least partially combined.

[0050] In one method step, in particular in the determination step 16, the cell characteristic parameters at the selected moment are determined based on the physical model, for which the physical model predicts the output characteristic parameters detected at the selected moment together with the input characteristic parameters at the selected moment, in particular the performance characteristic parameters. In order to determine the cell characteristic parameters, the physical model is inverted and optimized in the determination step 16, in particular for the cell characteristic parameters. In the determination step 16, an optimization method is applied to determine the cell characteristic parameters, such as Bayesian optimization, simulated annealing method, gradient method, etc., in particular, the output characteristic parameters detected for the input characteristic parameters at the selected moment of the cell system 30 and the output characteristic parameters predicted by the physical model for the input characteristic parameters at the selected moment.

[0051] In one method step, in particular in the evaluation step 18, the time course of an aging characteristic parameter of the cell system 30 is determined. To determine the time course of the aging characteristic parameter, the selection step 12, the conversion step 14 and the determination step 16 are performed at least a second time, wherein in the selection step 12, a further time different from the previously selected time is selected from the time series of the operating characteristic parameter. Based on the operating characteristic parameter of the cell system 30, the aging characteristic parameter at the further time, in particular a plurality of further time from the time series of the operating characteristic parameter is determined based on the physical model, in particular similarly to the aging characteristic parameter determined at the previously selected time from the time series of the operating characteristic parameter of the cell system 30. Based on the determined aging characteristic parameter at different time, the time course of the aging characteristic parameter can be determined, in particular displayable. It is conceivable to extrapolate the time course of the aging characteristic parameter, in particular to predict a future course of the aging characteristic parameter. It is also conceivable to correlate the time course of the aging characteristic parameter with at least one, preferably aging-related, operating characteristic parameter of the cell system 30 and / or with other cell characteristic parameters of the cell system 30.

[0052] It is also conceivable to use a physical model to train a regression model. Based on the results determined by means of the physical model, in particular based on the determined cell characteristic parameters, the regression model is trained, for example, by means of a statistical experimental design. In order to train the regression model, a statistical experimental design is generated from the following combinations of operating characteristic parameters, for which the cell characteristic parameters are determined according to the physical model, wherein the results are used in particular for training the regression model. The training of the regression model is based, for example, on linear regression, random forests, Gaussian processes, neural networks, interpretable boosting machines, etc. The regression model is arranged to replace the physical model in the alternative determination step 16. In particular, at least one cell characteristic parameter of the cell system 30 is determined according to the regression model in the alternative determination step 16. The regression model is arranged to be used in an optimization method, in particular in the alternative determination step 16, to determine the cell characteristic parameters or to determine the cell characteristic parameters directly. In particular, it is conceivable that the regression model is trained to determine at least one cell characteristic parameter of the cell system 30 directly from the operating characteristic parameters, preferably at least one input characteristic parameter and at least one output characteristic parameter, preferably in the alternative determination step 16, wherein preferably at least in the determination step 16, in particular, the inversion of the physical model is omitted.

[0053] Active learning, in particular secure active learning, is used to train the regression model. The statistical experimental design is generated with the aid of active learning. Alternatively, it is also conceivable to generate the statistical experimental design with the aid of Latin hypercube sampling or the like. Figure 3 A schematic process for training a regression model is shown.

[0054] Active learning is used to select the most informative training data to train the regression model. Safe active learning is used to avoid using training data that makes the physical model unable to converge, especially diverge. For example, when determining the cell characteristic parameters, the convergence characteristic parameters are determined. The convergence characteristic parameter is an indicator that shows how close the solution method, especially the numerical one, of the physical model is to non-convergence, especially divergence. The convergence characteristic parameter can be determined, for example, based on the required number of iterations of the optimization method for determining the cell characteristic parameters (especially the aging characteristic parameters). Alternatively, it is also conceivable to determine the convergence characteristic parameter based on the output variables of the physical model, especially based on at least one output characteristic parameter of the cell system 30 and / or based on internal characteristic parameters.

[0055] In the initialization step 20 of training the regression model, in order to initialize the training of the regression model, operating characteristic parameters of the battery cell system 30 are selected for the system state of the battery cell system 30, and the solution method for the physical model converges when the operating characteristic parameters are known. According to the physical model, in particular by inversion and optimization of the physical model, the battery cell characteristic parameters are determined for the operating characteristic parameters selected in the initialization step 20. According to the result of the initialization step 20, in particular according to the operating characteristic parameters selected in the initialization step 20 and the battery cell characteristic parameters determined in the initialization step 20, the value of the convergence characteristic parameter associated with the result is determined. Based on the result from the initialization, in particular based on the operating characteristic parameters selected in the initialization step 20, a preferably probabilistic regression model, preferably a Gaussian process model, is trained. Additionally, based on the value of the convergence characteristic parameter determined in the initialization step 20, a safety model, in particular a probabilistic one, preferably a Gaussian process model, is trained.

[0056] After the initialization step 20, a loop is executed to train the regression model, in particular by means of the control unit 34. The loop has a first loop step 22, in which an operating characteristic parameter is selected for another system state of the cell system 30, the operating characteristic parameter preferably having a maximum value of the information content, and the solution method for the physical model converges with a probability above the probability threshold in the case of the operating characteristic parameter. More particularly preferably, the probability threshold is 0.5. However, it is also conceivable as an alternative that the probability threshold is between 0.1 and 0.9, preferably between 0.2 and 0.8, preferably between 0.3 and 0.7, and particularly preferably between 0.4 and 0.6. In addition, it is also conceivable as an alternative that the probability threshold is outside the value range between 0.1 and 0.9. The information content is evaluated by entropy, which is proportional to the prediction variance, in particular in a Gaussian process. In particular, a restricted optimization problem is generated, in which an operating characteristic parameter is calculated, in the case of which the information content has a maximum value, and the probability of convergence of the solution method for the physical model is greater than the probability threshold.

[0057] The probability threshold can be set by an operator, for example, via an input unit (not shown here). The input unit is in particular data-connected or connectable to the control unit 34, preferably for wireless and / or wired data transmission. The input unit can be configured, for example, as a keyboard, a keyboard panel, a touch screen, a knob, a slide regulator, a button, etc. It is conceivable that the control unit 34 or an external unit has an input unit. As an alternative to the probability of convergence of the solution method for the physical model used to determine the cell characteristic parameters, it is also conceivable to have a criterion that the probability of interference-free simulation with the aid of the physical model is greater than a probability threshold. The probability of interference-free simulation with the aid of the physical model takes into account the convergence of the solution method for the physical model, storage requirements, processor requirements and simulation duration. Alternatively or in addition, the distance between the cell characteristic parameters determined by the physical model and the known physical limits can also be used as an indicator of the probability of interference-free simulation.

[0058] In the second loop step 24 of the loop, the values ​​of the battery characteristic parameters and the corresponding convergence characteristic parameters are determined using the operating characteristic parameters of another system state. In the third loop step 26 of the loop, the regression model and / or the safety model are updated according to the cell characteristic parameters determined in the second loop step 24 and / or according to the convergence characteristic parameters determined for another system state of the cell system 30. If the termination criterion is not met in the third loop step 26, the loop restarts the first loop step 22. The termination criterion can be, for example, the exhaustion of the calculation time, the exhaustion of the number of evaluated system states of the cell system 30, the information content falling below an information content threshold, or the maximum error on the validation data set falling below a threshold.

[0059] According to the determined cell characteristic parameters, in particular in the determination step 16, other cell characteristic parameters, in particular internal characteristic parameters, of the cell system 30 are determined. Based on the physical model, based on the aging characteristic parameters determined based on the physical model, in particular by inversion and optimization of the physical model for the aging characteristic parameters or by a regression model generated with the help of the physical model, other cell characteristic parameters are determined, in particular output variables of the physical model.

[0060] It is conceivable that other cell characteristic parameters, in particular internal characteristic parameters, are determined at least partially with spatial resolution, preferably in the determination step 16. In particular, it is preferred that in the determination step 16, respective local distributions of other cell characteristic parameters are determined. The respective local distributions of other cell characteristic parameters are local distributions of respective other cell characteristic parameters within the cell system 30, in particular within a stack formed by a plurality of electrochemical cells 32. The respective local distributions of other cell characteristic parameters are determined based on the physical model according to the operating characteristic parameters of the cell system 30, preferably by an optimization method of the physical model applied to the inversion or by a regression model generated with the aid of the physical model. The other cell characteristic parameters can be determined at a position or a list of positions in the cell system 30, respectively. It is also conceivable to determine a combination of other cell characteristic parameters of the cell system 30, in particular a combination of internal characteristic parameters, such as a product, a sum, a nonlinear function, etc., at the position or the list. In addition, it is also conceivable to determine the derived values ​​of other respective battery cell characteristic parameters (especially respective internal characteristic parameters) or the derived values ​​of combinations of other characteristic parameters (especially combinations of internal characteristic parameters), such as the maximum value, minimum value, average value in a partial volume or the total volume of the battery cell system 30, for describing the local distribution, such as the first-order area moment, the second-order area moment, the support points of the cumulative distribution density, etc., the gradient or the rate of change.

[0061] exist Figure 4 , a schematic process for training a fault recognition model and applying the fault recognition model to a cell system 30 is shown in FIG. The physical model is used to generate training data for the fault recognition model. In particular, a critical value range is at least partially preset for other cell characteristic parameters (in particular internal characteristic parameters), and the critical value range is stored, for example, on a storage element of the control unit 34. It is conceivable that the critical value range is adaptable, for example, automatically adapted by the control unit 34 or manually adapted by an operator via an input unit. The physical model can be used to determine the following operating characteristic parameters, for which at least one of the other cell characteristic parameters is within a critical value range. In the data generation step 28, a database is determined, which has operating characteristic parameters and values ​​of associated internal characteristic parameters determined according to the physical model. The database is stored, for example, on a storage element of the control unit 34. The fault recognition model can be trained in a targeted manner with the operating characteristic parameters by means of the database, wherein each associated value of the other cell characteristic parameters is within or outside the critical value range. The fault recognition model is trained, for example, by machine learning, in particular for performing a classification method to classify a determinable fault state of the cell system 30. The training of the fault identification model is based on, for example, random forest, support vector machine, neural network, etc.

[0062] During operation of the cell system 30 , the fault recognition model is used to identify fault states in the cell system 30 based on operating characteristic parameters of the cell system 30 detected during operation, and in particular to classify the fault states, for example, at least associate them with at least one of the other cell characteristic parameters.

Claims

1. A method for determining at least one cell characteristic parameter of a cell system (30), the cell characteristic parameter being in particular an aging characteristic parameter, the cell system having at least one electrochemical cell (32), characterized in that: The battery cell characteristic parameters are determined according to the operating characteristic parameters of the battery cell system (30) based on the physical model of the battery cell system (30).

2. The method according to claim 1, characterized in that The operating characteristic parameter includes at least one input characteristic parameter of the battery cell system (30).

3. The method according to claim 2, characterized in that The input characteristic parameter comprises at least one operating parameter of the battery cell system (30).

4. The method according to any one of the preceding claims, characterized in that The operating characteristic parameter includes at least one output characteristic parameter of the battery system (30).

5. The method according to claim 4, characterized in that The output characteristic parameter has at least one performance characteristic parameter of the battery cell system (30).

6. The method according to any one of the preceding claims, characterized in that The operating characteristic parameter is detected at least partially on the battery cell system (30).

7. The method according to any one of the preceding claims, characterized in that A time series is formed based on the operating characteristic parameter.

8. The method according to any one of the preceding claims, characterized in that The physical model is inverted and optimized to determine the cell characteristic parameters.

9. The method according to any one of the preceding claims, characterized in that The physical model is used to train a regression model.

10. The method according to claim 9, characterized in that Use active learning, especially secure active learning, to train regression models.

11. The method according to any one of the preceding claims, characterized in that Determine the time variation process of the aging characteristic parameters of the battery cell system (30).

12. The method according to any one of the preceding claims, characterized in that At least one other battery cell characteristic parameter of the battery cell system (30) is determined according to the determined battery cell characteristic parameter.

13. The method according to claim 12, characterized in that The at least one further cell characteristic parameter is determined in a spatially resolved manner.

14. The method according to claim 12 or 13, characterized in that The at least one other cell characteristic parameter comprises a temperature distribution in the cell system (30).

15. The method according to any one of claims 12 to 14, characterized in that The at least one other cell characteristic parameter comprises fluid distribution in the cell system (30), in particular oxygen distribution and / or hydrogen distribution.

16. The method according to any one of claims 12 to 15, characterized in that The at least one other battery cell characteristic parameter includes a current density distribution in the battery cell system (30).

17. The method according to any one of the preceding claims, characterized in that The physical model is used to generate training data for a fault identification model.

18. A cell system (30) having at least one electrochemical cell (32) and at least one control unit (34) for carrying out the method according to any one of the preceding claims.

19. A computer program (36) comprising instructions which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 17.

Images