Method and device for determining model parameters and internal battery state of electrochemical battery model of device battery of technical device

Abstract

The invention relates to a computer-implemented method for parameterizing an electrochemical battery model of a battery in a technical plant using updated model parameters for determining an internal battery state, in particular for determining a charging profile, for monitoring for anomalies and / or for determining an aging state, the method comprises the following steps: recording a time change process of operation variables of a large number of equipment battery packs; for a specific device battery pack, selecting a subset of similar device battery packs having a similar overall state with the specific device battery pack from the plurality of device battery packs; for the model parameters of the battery pack model of the device battery pack subset, respectively determining the parameter lower limit and the parameter upper limit of the corresponding model parameters according to the distribution of the model parameter values in the device battery pack subset under the calendar age condition of the specific device battery pack; a fitting method is performed for a battery pack model for a specific device battery pack according to parameter limits of the model parameters based on a temporal variation of the operating variable to obtain updated model parameters.

Description

Technical Field

[0001] The present invention relates to the determination of the internal battery pack state and model parameters of an electrochemical battery pack model of a device battery pack, and the electrochemical battery pack model can be designed as, for example, a "continuum model (Kontinuum-Modell)" by means of a fitting method. For example, the model parameters can be the basis for state monitoring, determining the aging state, and calculating the charging curve of the device battery pack. Background Art

[0002] The energy supply for electrical devices and machines that operate independently of the power grid, such as electrically driven motor vehicles, is usually carried out using a device battery pack or a vehicle battery pack. They provide electrical energy to operate the device.

[0003] The device battery pack degrades over its service life and depending on its load or usage. This so-called aging leads to a continuous decrease in the maximum performance or storage capacity. The aging state corresponds to a measure for describing the aging of the energy storage device. Conventionally, a new device battery pack can have an aging state of 100% (in terms of its capacity, i.e., SOH-C), and this aging state gradually decreases during its service life. The measure of device battery pack aging (the change in the aging state over time) depends on the individual load of the device battery pack, i.e., in the case of a vehicle battery pack of a motor vehicle, it depends on the driver's usage behavior, external environmental conditions, and the type of vehicle battery pack.

[0004] Therefore, the device battery packs in electrical devices operating independently of the power grid, especially in mobile devices, are continuously monitored in order to promptly identify sudden device battery pack failures (Sudden Death), thermal error events, rapid cycle aging, or other error events (collectively referred to herein as anomalies) and to warn the user of the occurrence of such anomalies. Here, such anomalies include identifying current or future anomalies based on the temporal course of the measured operating variables of the device battery pack.

[0005] Depending on the type of anomaly found, different measures can be taken, ranging from simply warning the user to immediately stopping the operation of the technical device. Usually, such anomaly identification is performed based on the cloud.

[0006] In order to perform anomaly identification as independent of the cloud as possible, anomaly identification models can be implemented in the technical device, and these anomaly identification models evaluate the course of the operating variable changes and signal anomalies in an appropriate manner when deviating from the traditional operating mode. The anomaly identification models determine and evaluate the battery pack state with the help of an electrochemical battery pack model.

[0007] The behavior of a device battery pack can generally be modeled with the aid of an electrochemical battery pack model. Such models are characterized in that they establish a connection between externally measurable battery pack characteristic variables, in particular the terminal voltage, and the internal battery pack state, such as the concentration of lithium in the electrolyte and the solid state. The internal battery pack state is derived here from the assumed model parameters. A typical representative of this model class is the reduced-order electrochemical continuum model, such as the Newman model. Electrochemical battery pack models can be used to: determine an optimized charging current profile (charging curve) to charge the battery cells of the battery pack as quickly and gently as possible and maintain an as-constant as possible expected lifetime; monitor the state of the battery cells of the battery pack; determine the aging state, etc. Summary of the Invention

[0008] According to the present invention, there is provided a method for parameterizing an electrochemical battery pack model of a device battery pack in a technical device by using updated model parameters for determining the internal battery pack state and by means of the device according to the co-pending independent claims.

[0009] Further design options are described in the dependent claims.

[0010] According to a first aspect, there is provided a computer-implemented method, in particular for determining a charging curve, for monitoring anomalies and / or for determining an aging state, the method having the following steps:

[0011] - Recording the temporal course of the operating variables of a large number of device battery packs;

[0012] - For a specific device battery pack, selecting a subset of similar device battery packs from the plurality of device battery packs that have a similar overall state to the specific device battery pack;

[0013] - For the model parameters of the battery pack model of the subset of the device battery packs, respectively determining a parameter lower limit and a parameter upper limit of the corresponding model parameter according to the distribution of the model parameter values of the subset of the device battery packs in the case of the calendar age of the specific device battery pack;

[0014] - Performing a fitting method for the battery pack model for the specific device battery pack based on the temporal course of the operating variables according to the parameter limits of the model parameters in order to obtain updated model parameters.

[0015] Traditional electrochemical battery models are mathematical models and have multiple model parameters. The battery model is used to mathematically describe the terminal voltage behavior and internal electrochemical processes in the battery cells of the battery pack. The model parameters of such battery models are typically fitted by comparing (Abgleich) the measured and simulated operating variable profiles with the aid of traditional optimization methods. This fitting is performed regularly at recalibration time points because the model parameters must be adapted to the degradation of the device battery pack in order to handle the gradual degradation of the battery pack reasonably.

[0016] Fitting methods are used to adapt a mathematical model to a given data set. Here, the model parameters are chosen such that the discrepancy between the observed data and the values predicted by the model is minimized. The goal is to estimate the best model parameters of the model, which result in the "best" agreement with the data. An error function that quantifies the difference between the observed data and the modeled data is usually defined. The most commonly used methods are the least squares method and the maximum likelihood method.

[0017] The model parameters of the battery model are adapted during the life cycle of the battery pack to reflect the gradual aging or degradation of the device battery pack. For this purpose, the model parameters are adapted via a non-linear fitting method based on the time-dependent operating variable profiles such that the battery model depicts (abbilden) the battery voltage profile as accurately as possible according to the profiles of the battery current, temperature, and state of charge, where the time-dependent operating variable profiles can in particular include the time-dependent profiles of the battery voltage, battery current, battery temperature, and state of charge. In particular, the least squares method is applied here, where the number of parameters to be adapted is usually very large, between 20 and 30.

[0018] Applying traditional fitting methods to adapt the model parameters to the current measurement sequence of the operating variable profiles has the following disadvantages especially in the case of the recalibration interval applicable between recalibration time points: A large number of model parameter combinations result in comparable terminal voltage behavior (Klemmspannungsverhalten). In other words, fitting methods based on minimizing the squared error usually lead to convergence within a local minimum, which, however, does not necessarily correspond to the global minimum of the cost function. Therefore, the internal battery pack state as the model state obtained from the evaluation of the battery model parameterized by the model parameters may deviate from the true battery pack state to be described by the modeled internal battery pack state. This effect is enhanced due to inaccurate measurements or also due to insufficient quality of the provided operating variable profiles in the case where the optimal operating conditions required for performing the recalibration, i.e., the reparameterization of the battery model, do not exist.

[0019] Thus, especially in the case of a large re-calibration interval, the following problem exists: The parameter values determined during re-calibration may jump because, due to a large number of model parameters, multiple local minima can usually be achieved through optimization. All these combinations of model parameters are characterized by the same or very similar terminal voltage behavior with respect to the calibration data on which they are based, i.e., the measured course of the operating variables over time. This can result in combinations of model parameters with significant deviations from each other with each re-calibration. This leads to: During the course of time of the model parameters that should be assigned to the internal battery state and that change monotonically (decrease or increase) as the device battery ages, a change in the sign of the gradient may occur. However, this contradicts the basic expectation that the model parameter values should change monotonically during the aging process and thus indicates at least a temporarily implausible description of the battery state expected at the corresponding time point.

[0020] The central concept of the above method lies in adapting the allowable parameter limits of the model parameters or the restriction of the value space to achieve an optimized setting of the model parameters. This helps to minimize the inherent uncertainty of the fitting method due to the possibility of converging to different local minima.

[0021] For this purpose, the above method provides for the parameterization of the battery model of a specific device battery: The parameter limit values for the model parameters are determined based on an evaluation of the model parameters or the aging state derived from the model parameters. For this, first, a subset of device batteries with an overall state similar to that of the specific device battery for which the model parameters are to be re-calibrated is selected from a large number of other device batteries. For this purpose, the aging state and the course of time of the model parameters of a large number of other device batteries are compared with the aging state and the course of time of the model parameters of the specific device battery, and several other device batteries are selected as a subset by means of a similarity analysis.

[0022] In particular, the overall state of a device battery can be determined by a characteristic point that is determined by the aging state and / or one or more of the model parameters of the associated battery model and the calendar age of the device battery.

[0023] Similar device battery packs can be determined by means of a clustering method for feature points, where the feature points are defined by the internal battery pack state, the aging state, and the calendar age and / or operating history of the battery pack. Thus, if the Euclidean distance between similar device battery packs relative to their feature points to each other or to the centroid of the cluster they form is less than a pre-given threshold, then the similar device battery packs themselves can be identified. For this purpose, the elements of the feature points can be normalized and / or weighted in a suitable manner. In particular, battery packs from a large number of other device battery packs of the same type are understood to be similar if they have the greatest consistency (= minimum Euclidean distance) with the examined device battery pack in terms of the temporal course of the aging state calculated by an aging model since the start of the service life of the device battery pack. Other features suitable for determining similar battery packs are the cumulative energy throughput and the load histogram since the start of the service life, which provide an explanation of how strongly the battery pack is loaded in terms of power consumption at what state of charge and temperature range.

[0024] If the course of change of the aging state and / or the course of change of the model parameters of the selected device battery pack is / are unreasonable, the device battery pack can be removed from the subset of the selected device battery packs. For this purpose, an aging and model parameter trajectory can be created by means of a polynomial or other regression method in order to exclude device battery packs with an unreasonable course of change of the aging state and / or one or more model parameters, or to remove from the subset those device battery packs whose trajectories of the aging state and / or model parameters do not correspond to the assumed functional curve in the regression. For example, the aging state and certain specific model parameters or the internal battery pack state can only be monotonically increasing or decreasing. If within the course of change of the aging state and / or the model parameters not only periods with a positive gradient are determined but also periods with a negative gradient, then an unreasonable course of change can be assumed.

[0025] It can be stipulated that the minimum and maximum values of the relevant model parameters of the battery pack model of the subset of device battery packs interpolated with respect to the calendar age of a specific device battery pack are assumed as the parameter lower limit and the parameter upper limit.

[0026] The historical model parameters determined at the recalibration intervals for the other device battery packs of the subset can be interpolated in order to determine the distribution of each of the model parameters of the battery pack model for the subset of selected device battery packs for the current calendar age of a specific device battery pack. Thereby, the corresponding parameter limits for the relevant model parameters can be determined in order to apply them in a fitting method for the model parameters for recalibrating a specific device battery pack and to predefine a search space for suitable combinations of model parameters. The lowest and highest values of the interpolations (Interpolationswert) of the relevant model parameters for the considered calendar age can be assumed as the lower and upper parameter limits, respectively.

[0027] It can be provided that the updated model parameters are used to directly determine or, after applying a calculation model, to determine the aging state and / or the recalibrated internal battery state of a specific device battery pack. The updated internal battery state can be used, in particular, to perform monitoring of anomalies of a specific device battery pack by means of a rule-based or data-based anomaly recognition model and / or to generate a charging curve. For example, a charging curve adapted to the updated model parameters can be calculated by applying a current profile (Stromprofil) to the recalibrated model, where, in the case of applying this current profile, the battery state associated with excessive battery aging assigned to the charging process (e.g., the lithium concentration at the electrode / electrolyte interface) does not exceed a specific threshold. This can ensure that the device battery pack is exactly loaded depending on its aging or state such that the excessive aging caused by the charging process is minimized.

[0028] According to another aspect, a device for performing the above method is provided. Description of the Drawings

[0029] The embodiments will be explained in more detail below with reference to the drawings. Among them:

[0030] Figure 1 A schematic diagram of a system with a large number of vehicles in a fleet and a central unit for performing recalibration of an electrochemical battery pack model, in particular for charging curve optimization, is shown; and

[0031] Figure 2 A flowchart for explaining a method for recalibrating model parameters of an electrochemical battery pack model, which method is in particular for monitoring the battery pack state and for determining a charging curve, is shown. Detailed Description of the Embodiments

[0032] The method according to the invention will now be described in terms of a vehicle battery pack as an equipment battery pack in a large number of motor vehicles as similar equipment. For this purpose, an electrochemical battery pack model is parameterized in a central unit and used to monitor the overall state of the vehicle battery pack, generate a charging curve and calculate the aging state. In the central unit, for each of the vehicle battery packs, the battery pack model is periodically updated or re-parameterized based on the operating variables of the corresponding vehicle battery packs from the fleet.

[0033] The above examples represent a large number of fixed or mobile devices with an energy supply independent of the power grid, such as vehicles (electric vehicles, scooters, etc.), facilities, machine tools, household appliances, Internet of Things devices, etc., which are connected to a central unit (cloud) outside the device via a corresponding communication connection (such as LAN, Internet).

[0034] Figure 1 A system 1 for collecting fleet data in a central unit 2 to create and run an electrochemical battery pack model is shown. The model parameters and the battery pack model parameterized with them are used to determine the aging state of the battery pack cells, monitor anomalies of the vehicle battery pack and determine the charging curve. Figure 1 A fleet 3 with a plurality of motor vehicles 4 is shown. The electrochemical battery pack model is used to depict the terminal voltage in a model-based manner based on the state of charge, battery pack current and battery pack temperature at the cell level, module level and / or pack level (Packebene).

[0035] Figure 1 One of the motor vehicles 4 is shown in more detail. These motor vehicles 4 each have a vehicle battery pack 41, 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 corresponding motor vehicle 4 and the central unit 2 (so-called cloud). The vehicle battery pack 43 has a large number of battery pack cells 45.

[0036] The control unit 43 is specifically designed to record the operating variables with a high time resolution, for example, between 1 Hz and 50 Hz, for example, 10 Hz, and transmit them to the central unit 2 via the communication device 44. In the case of a vehicle battery pack, these operating variables F can represent the instantaneous battery pack current, instantaneous battery pack voltage, instantaneous battery pack temperature and instantaneous state of charge (SOC: State of Charge), not only at the pack level, module level but also / or at the cell level. These operating variables F can be transmitted to the central unit 2 periodically in an uncompressed and / or compressed form. For example, for the purpose of minimizing the data traffic to the central unit 2, the time series can be transmitted to the central unit 2 in blocks at intervals of 10 minutes to several hours by using a compression algorithm.

[0037] The motor vehicle 4 sends the operating variable F to the central unit 2, the operating variable at least describing variables which influence the ageing state of the vehicle battery pack 41 and which are required for parameterizing the battery pack model. The model parameters correspond to the internal battery pack state and / or can be calculated to determine the internal battery pack state of the battery cells 45 of the battery pack.

[0038] The model parameters of the electrochemical battery pack model can be fitted or parameterized in the central unit within a limited period of time based on the course of the operating variables recorded during an idle phase over a short period (from a few minutes to a few hours), wherein electrochemical equilibrium parameters and kinetic model parameters can be derived which can be interpreted as the internal battery pack state, such as electrolyte concentration, reaction rate, layer thickness, porosity etc.

[0039] The central unit 2 has a data processing unit 21 and a database 22, in which the method described below can be carried out in the data processing unit 21, and the database 22 is used for storing data points, model parameters, states etc.

[0040] In the central unit 2, the battery pack model is parameterized for each vehicle battery pack 41. The battery pack model can be described in the form of the known Newman model per se. The Newman model is an electrochemical model of a lithium-ion battery pack which describes the transport and intercalation of lithium ions in a porous electrode and from which conclusions about the expected terminal voltage of the battery pack are drawn based on the electrochemical potential and the lithium concentration in the electrode / electrolyte.

[0041] The battery pack model can be used regularly, i.e. for example after a pre-given recalibration interval, to carry out a recalibration of the model parameters of the battery pack model based on the course of the operating variables over time. The model parameters enable the ageing state of the vehicle battery pack 41 to be determined by combination.

[0042] The ageing state (SOH: State of Health) is a key variable for describing the remaining battery pack capacity or the remaining battery pack charge. The ageing state represents a measure of the ageing of the battery pack module or the battery cells or the vehicle battery pack and can be specified as the capacity retention rate (SOH-C) or the internal resistance increase (SOH-R). The capacity retention rate SOH-C, i.e. the ageing state related to the capacity, is specified as the ratio of the measured instantaneous capacity to the initial capacity of a fully charged battery pack and decreases with increasing ageing. Alternatively, the ageing state can be specified as the internal resistance increase (SOH-R) relative to the internal resistance at the start of the service life of the device battery pack. The relative change in the internal resistance SOH-R increases with increasing battery pack ageing.

[0043] In the central unit, the course of the operating variables is generally evaluated in various ways. In particular, for each of the vehicle battery assemblies, an electrochemical battery pack model is modeled, which describes the associated vehicle battery pack 41 in the best possible way and in particular the state of the battery pack inside it.

[0044] The battery pack model is generally designed as a reduced-order electrochemical continuum model, such as the Newman model, and enables the description of the state of the battery pack and the terminal voltage behavior as well as the internal electrochemical processes of the vehicle battery pack 41 or the battery cells contained therein. Due to the degradation of the vehicle battery pack 41, the associated battery pack model must be recalibrated regularly. For this purpose, the time period of the course of the operating variables is used as calibration data in order to adapt the model parameters by means of a fitting method known per se, wherein the fitting method can be designed for optimization based on the least squares method.

[0045] The method described below makes use of the possibility of improving the optimization method by adapting the model parameters in such a way that parameter limits are pre-given in order to thus avoid false parameterization due to the diverging behavior of the optimization caused by the large number of model parameters to be parameterized.

[0046] Figure 2 The method carried out in the central unit 2 is shown schematically.

[0047] In step S1, the course of the operating variables is received from a large number of vehicle battery packs 41, and a specific one of the vehicle battery packs is selected for further processing for performing the recalibration. Generally, the recalibration of the battery pack model for the vehicle battery pack 41 is carried out after a regular recalibration interval, wherein the recalibration of all vehicle battery packs 41 does not have to be carried out simultaneously.

[0048] In step S2, first a subset of vehicle battery packs 41 is selected from the large number of vehicle battery packs 41 that have an overall state similar to that of a specific vehicle battery pack 41. The overall state of the vehicle battery pack 41 can be determined based on one or more of the model parameters of the associated battery pack model, the calendar age of the vehicle battery pack 41, and the characteristic points of the aging state.

[0049] For this purpose, for each vehicle battery pack 41, the aging state and the change process of the model parameters can be formed via its calendar age with the aid of a corresponding regression model. For example, by determining the aging state of other vehicle battery packs 41 for the corresponding calendar age and the model parameters of the battery pack model for the calendar age of a specific vehicle battery pack 41 with the aid of a regression model, and thus determining the characteristic points of all vehicle battery packs 41 for the corresponding calendar age, vehicle battery packs 41 similar to the specific vehicle battery pack 41 can be found. The similar vehicle battery packs 41 in the subset can be determined by means of cluster analysis of the characteristic points, or can be determined by comparing the Euclidean distances of these evaluation characteristic points to the characteristic points of the specific vehicle battery pack 41 with a pre-given threshold value.

[0050] In step S3, if the change process of the aging state and / or the change process of the model parameters of the selected vehicle battery pack is / are unreasonable, the selected vehicle battery pack can be removed from the subset. Unreasonableness is essentially characterized here by a change in the positive or negative signs of the aging state and the model parameter gradients (both the aging state and the model parameters should have a uniformly increasing or decreasing change process) or the general positive or negative signs of the aging state and the model parameter gradients (for many model parameters, it is known a priori, for example, based on domain knowledge: whether the aging state and the model parameters increase or decrease during the aging process).

[0051] In step S4, now the parameter limits are determined based on the distribution of the values of each model parameter of the subset of vehicle battery packs. In particular, the minimum and maximum values of the relevant model parameters of the battery pack model of the subset of vehicle battery packs can be assumed as the lower parameter limit and the upper parameter limit. For this purpose, for example, the change process of the model parameters of the selected device battery packs of the subset can be interpolated via its calendar age to determine the distribution of the values of each model parameter for the current calendar age of a specific device battery pack. Thereby, the corresponding parameter limits for the relevant model parameters can be determined in order to apply them to the fitting method for recalibrating the model parameters of a specific device battery pack 41 and to pre-give the search space for a suitable combination of model parameters. The corresponding lowest and highest values of the interpolated values of the relevant model parameters for the considered calendar age can be assumed as the lower parameter limit and the upper parameter limit.

[0052] In the subsequent step S5, now a fitting method can be performed for the battery pack model of a specific vehicle battery pack based on the change process of the operating variables recorded for the specific vehicle battery pack, wherein the parameter limits of the model parameters are correspondingly pre-given to define the value space of the model parameters for the fitting method.

[0053] As a result of the fitting method, recalibrated model parameters of the battery pack model are obtained, where the aging state and the internal battery pack state of a specific vehicle battery pack can be determined directly based on the model parameters or after applying a calculation model for the current calendar age of the vehicle battery pack 41.

[0054] These aging states and internal battery pack states are stored in relation to the specific vehicle battery pack 41, for example to provide a digital twin in a central unit. The method can now be re-executed for another vehicle battery pack, since the previously recalibrated battery pack model of the vehicle battery pack can be considered in a subset of the selected vehicle battery packs.

[0055] The model parameters of the battery pack model can be used, in particular, with the aid of a rule-based or data-based anomaly recognition model to monitor anomalies of the vehicle battery pack 41 and / or to generate a charging curve.

Claims

1. A computer-implemented method for parameterizing an electrochemical battery model of a device battery (41) in a technical device (4) with updated model parameters, wherein the model parameters are used to determine an internal battery state, in particular for determining a charging curve, for monitoring anomalies and / or for determining an aging state, the method comprising the following steps: - recording (S1) the temporal course of operating variables of a plurality of device batteries (41); - for a specific device battery pack (41), selecting (S2) from a plurality of device batteries a subset of similar device batteries (41) having a similar overall state as the specific device battery pack (41); - for model parameters of the battery model of the subset of the device batteries (41), respectively determining (S4) lower parameter limits and upper parameter limits of the corresponding model parameters according to the distribution of model parameter values ​​in the subset of the device batteries (41) at the calendar age of the specific device batteries (41); - performing (S5) a fitting method for a battery model for a specific device battery (41) according to the parameter limits of the model parameters based on the temporal course of the operating variables in order to obtain updated model parameters.

2. The method according to claim 1, wherein: Minimum and maximum values ​​of relevant model parameters of the battery model of the subset of device batteries (41) interpolated with respect to the calendar age of the particular device battery (41) are assumed as lower and upper parameter limits.

3. A method according to any one of claims 1 to 2, wherein the overall state of the device battery pack (41) is determined by characteristic points, wherein the characteristic points are determined by one or more model parameters of the aging state and / or the associated battery pack model and the calendar age of the device battery pack (41).

4. The method according to any one of claims 1 to 3, wherein: Similar device battery groups (41) are determined in particular by the Euclidean distance between one another or, after a cluster analysis, by the Euclidean distance to the center of mass of a specific cluster, wherein the Euclidean distance is less than a predefined threshold value.

5. The method according to any one of claims 1 to 4, wherein: If the aging state change process and / or the model parameter change process of the selected device battery pack (41) is unreasonable, the device battery pack (41) can be removed from the subset of the selected device battery packs (41).

6. The method according to any one of claims 1 to 5, wherein: The updated model parameters are used to determine the state of aging of the specific device battery (41) and / or the internal battery state either directly or after applying a computational model.

7. The method according to claim 6, wherein: The updated model parameters are used to perform monitoring of a specific device battery pack (41) for anomalies, in particular by means of a rule-based or data-based anomaly recognition model, and / or to generate a charging curve.

8. Apparatus for carrying out the method according to any one of claims 1 to 7.

9. 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 7.

10. 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 7.

Images