Methods and apparatuses providing a relaxation model for predicting voltage during relaxation
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ROBERT BOSCH GMBH
- Filing Date
- 2026-01-07
- Publication Date
- 2026-07-07
Smart Images
Figure CN122345792A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a device battery, and more particularly to a method for determining model values of the terminal voltage of the device battery. The invention also relates to a method for determining precise values of the state of charge using a relaxation model. Technical Background
[0002] In many battery-powered applications, such as electric vehicles, understanding the current state of charge (SOC) of the battery is crucial. However, most devices do not operate continuously but rather experience operational pauses, during which no information about the device's battery is available to provide a precise indication of the current SOC value. Therefore, it is often difficult to determine the current SOC when the device is powered on.
[0003] Several methods are known in the prior art to determine the state of charge (SOC) during or after the relaxation phase, enabling the most accurate determination of the startup SOC when starting / turning on a technical device. These methods utilize the voltage values of individual battery cells measured when the technical device is shut down / shut down and determine the startup SOC at device startup based on a model. However, discrepancies may arise between the measured voltage values and the modeled voltage values at device startup / startup, indicating a mismatch in the underlying model and potentially requiring model correction.
[0004] In principle, during the operational phase, the change in state of charge (SOC) can be tracked very accurately by accumulating or integrating the transmitted and extracted electrical energy. However, it is important to know the startup SOC after the equipment has been shut down for a period of time and then restarted. Summary of the Invention
[0005] According to the present invention, a method for providing the state of charge of a device battery according to claim 1 is provided, particularly after a shutdown phase followed by a restart, and a corresponding apparatus according to the parallel claims.
[0006] Further design options are described in the dependent claims.
[0007] According to a first aspect, a computer-implemented method is provided for determining at least a portion of the startup state of charge of a device battery after a shutdown phase, comprising the following steps: -After determining that the technical equipment is shut down, provide the last detected operating variable change curve of the equipment battery's operating variables; - Using a parameterized electrochemical cell model, the terminal voltage change curve and state of charge change curve are modeled for a period of time after the technical equipment is turned off; - Provide terminal voltage variation curves and state of charge variation curves in the battery control device of the device battery; -Measure the terminal voltage at the moment of activation when the technical equipment is determined to be turned on; -Based on the comparison between the measured terminal voltage at the start-up time and the modeled terminal voltage, the state of charge at the start-up time in the modeled state of charge change curve is provided as the starting state of charge.
[0008] A device battery is an electrochemical energy storage device with specific electrochemical properties. The substances present in a device battery will change even in the absence of current, and even if chemical equilibrium has been established within the device battery, its internal state and composition will still change.
[0009] The above method, in principle, is based on determining the state of charge (SOC) of a device's battery by utilizing operational variables detected until the device is turned off, based on the startup value present at the moment the device is turned on. This startup SOC determination is model-based, and the method also allows for the modification of the corresponding electrochemical cell model.
[0010] The absence of operational variables related to the battery's operation outside of the actual operation of the technical equipment leads to inaccuracies in model calculations, such as inaccurate voltage predictions and the determination of related state of charge (SOC) values. Conversely, monitoring operational variables after the technical equipment is shut down provides advantages for model calculations and the determination of SOC at the time the equipment is turned on.
[0011] After the equipment begins operation, the battery enters a relaxation phase, during which the terminal voltage changes. Specifically, the terminal voltage increases after discharge and decreases after charging. This also relates to changes in the state of charge (SOC) value determined based on the terminal voltage. Until now, the SOC value has not been determined during the relaxation phase; that is, changes in the SOC during this period have not been considered.
[0012] The battery voltage is measured at the moment the device is powered on / off. This value almost corresponds to the open-circuit voltage (U_ocv). When powered on during the active relaxation phase, the deviation from the actual open-circuit voltage is caused by the fact that during the relaxation phase—that is, even when no external voltage is applied to the battery and no current flows—lithium ions are either inserted into the graphite of the anode (which increases the battery voltage and SoC after discharge) or removed from the anode (which decreases the battery voltage and SoC after charging). Therefore, there is no defined steady-state open-circuit state of the battery that allows for accurate determination of the state of charge (SOC) based on the measured open-circuit terminal voltage. At the beginning of the relaxation phase, charge flow is greater than in subsequent processes; that is, charge flow decreases during the relaxation phase until saturation (equilibrium state) is reached at the end of the relaxation phase, and the charge balance between the anode and cathode (or vice versa) becomes zero. Only from this moment (which is also strongly dependent on temperature) can the SOC be accurately determined based on the terminal voltage.
[0013] This inaccuracy in determining the state of charge (SOC) is traditionally compensated for using a weighting factor based on battery temperature and the last stored SOC value before device shutdown / shutdown. Furthermore, since load behavior before device shutdown / shutdown also affects the voltage change curve during the relaxation phase, thus influencing the SOC value during that phase, this method leads to errors in determining the SOC upon startup after shutdown.
[0014] If the device starts up after the relaxation phase has completely ended, the state of charge (SOC) value is determined based on the terminal voltage measured at startup using the open-circuit voltage characteristic curve (OCV-SOC characteristic curve). This SOC value determination is very accurate because it is performed under steady-state conditions and determined by the battery cell manufacturer before SOC through extensive measurements, taking into account relevant correlations, and stored in the battery control device in the form of a characteristic curve or lookup table. However, since the OCV-SOC characteristic curve also depends on the aging conditions, which can only be determined imprecisely in the battery control device, the startup SOC determination using this method can only be imprecise as well.
[0015] By utilizing a central unit located away from the equipment to assess operational variable data, it may be possible to more accurately determine the state of charge at startup.
[0016] According to the present invention, a method for determining the state of charge of a battery operating device after a shutdown phase is provided according to claim 1, and a corresponding apparatus according to the parallel claims.
[0017] According to a first aspect, a computer-implemented method is provided for determining the startup state of charge of a device battery after a shutdown phase, comprising the following steps: -After determining that the technical equipment is shut down, provide the last detected operating variable change curve of the equipment battery's operating variables; - Using a parameterized electrochemical cell model, the terminal voltage change curve and state of charge change curve are modeled for a period of time after the technical equipment is turned off; - Provide terminal voltage variation curves and state of charge variation curves in the battery control device of the device battery; -Measure the terminal voltage at the moment of activation when the technical equipment is determined to be turned on; -Based on the comparison between the measured terminal voltage at the start-up moment and the modeled terminal voltage, the state of charge at the start-up moment is provided as the starting state of charge in the modeled state of charge change curve.
[0018] In addition, the operating variable change curves can be provided in a central unit remote from the equipment, where the terminal voltage change curves and state of charge change curves are modeled for a period of time after the shutdown moment and then transmitted back to the technical equipment.
[0019] The aforementioned method aims to determine the initial state of charge based on an electrochemical cell model. If a discrepancy is found between the electrochemical cell model and the actual device battery, the model can be continuously readjusted. To this end, technical devices can be connected to a central unit that is connected to a large number of technical devices and detects the changing curves of operating variables related to the device battery from these devices. Based on these changing curves, the underlying electrochemical cell model is continuously readjusted.
[0020] The above method now aims to utilize the off-time or non-operation time in the device battery when there is no current flow and no data transmission to the central unit, in order to determine, on the one hand, the starting state of charge and the modeled starting terminal voltage of the device battery at the moment when the technical equipment is put back into use or operation, and on the other hand, based on the determination of the deviation between the modeled starting terminal voltage and the measured starting terminal voltage, to readjust the electrochemical cell model that can be used to determine the starting state of charge if necessary.
[0021] Preferably, the method is executed in a central unit. The method proposes that, after determining that the technical equipment is shut down, the operating variable change curves that have not yet been transmitted to the central unit are transmitted to the central unit. These operating variable change curves are used to run an electrochemical cell model, and the model parameters can be reparameterized using the operating variable change curves, or the change curves of one or more operating variables can be predicted.
[0022] Such electrochemical cell models may include a system of differential equations, which, based on differential equations parameterized by model parameters, model the internal cell states (especially equilibrium states and, where necessary, kinetic states) using time integration and provide the relationship between the device cell's operating variables over time (i.e., cell current, cell voltage, cell temperature, and the device cell's state of charge) and the internal cell states. Such electrochemical cell models are disclosed, for example, by US20220179009A1, US20220334191A1, US20220099743A1, US2016 / 023,566, US2016 / 023,567, and US2020 / 150,185. Furthermore, aging states can be derived from the internal cell states in ways known per se.
[0023] It is possible to propose modeling the terminal voltage change curve and the state of charge change curve only during the relaxation phase, and to assume that the values of the terminal voltage and state of charge reached after the relaxation phase are constant until the turn-on time; or to model the terminal voltage change curve and the state of charge change curve for a period of time beyond the relaxation phase.
[0024] Then, using an electrochemical cell model, starting from the moment the device is turned off, the terminal voltage and state of charge of the device's battery can be modeled over time based on the model parameters, the operating variable curve of the 0A battery current, and the assumed temperature. This time-varying curve reflects the relaxation phase and the subsequent equilibrium phase.
[0025] The battery model can use the final determined state of charge (SOC) value, the final determined battery temperature, and the aging state at the shutdown time as input variables, and then extrapolate the parameterized electrochemical battery model over a certain period of time. This period of time may include the expected duration of the technical equipment's resumption of operation.
[0026] Alternatively, this time period may consist only of a relaxation phase, which lasts until voltage saturation is reached, i.e., the voltage change of the modeled terminal voltage is less than a preset threshold (e.g., 0.05 mV / min). This criterion determines the end of the relaxation phase. The terminal voltage reached at this point and the state of charge reached at this point are assumed to be constant in the subsequent time period until the turn-on time.
[0027] Therefore, based on the temperature and aging state change curves, with a battery current of 0A, the terminal voltage change curve and state of charge change curve of the device battery during the relaxation stage were obtained.
[0028] The voltage and state of charge (SOC) change curves modeled during the relaxation phase are transmitted to the technical equipment as a relaxation model and temporarily stored there. The relaxation model can indicate the modeled terminal voltage and modeled SOC based on the duration of battery power failure and battery temperature.
[0029] Because the battery control unit remains laging for several minutes after the technical equipment is shut down (Nachlauf), the battery control equipment is still ready to receive and store these change curve data, so that these change curve data are available at the time of startup and can be used to indicate the startup state of charge.
[0030] If the technical equipment restarts, the battery control device performs a measurement of the terminal voltage and compares the measured terminal voltage with the terminal voltage modeled based on the modeled voltage change curve. If the measured voltage and the modeled terminal voltage are the same or have only a deviation within tolerance, the corresponding modeled state of charge value is determined as the startup state of charge.
[0031] Conversely, if there is a discrepancy between the two terminal voltage values, the measured terminal voltage value is trusted more than the modeled voltage value, assuming the voltage sensor device is not faulty. A state of charge (SOC) value is assigned to the measured voltage value, corresponding to the modeled voltage value on the voltage change curve during the relaxation phase.
[0032] When a voltage difference exceeding a preset threshold occurs between the measured terminal voltage and the modeled terminal voltage, not only can the measured terminal voltage value be transmitted to the central unit, but also the modeled terminal voltage value at the time of device startup can be transmitted. This can be used to readjust the electrochemical cell model through parameter changes (e.g., using the least squares method), so that all devices using this central unit can benefit from the improved electrochemical cell model. The improved cell model will then be used as a basis for calculating the terminal voltage change curves of other devices after their shutdown time.
[0033] Furthermore, when there is a deviation between the measured terminal voltage at the turn-on time and the modeled terminal voltage, it can be assumed that the state of charge derived from the terminal voltage measured at the turn-on time is the state of charge, wherein the following times during the relaxation phase are allocated to the terminal voltage measured at the turn-on time, at which time the modeled terminal voltage corresponds to the terminal voltage measured at the turn-on time, wherein the equivalent state of charge corresponds to the state of charge at such determined times in the state of charge change curve.
[0034] It can be proposed that, after the technical equipment is turned off and before it is turned on, one or more terminal voltages measured at different times are detected, wherein the electrochemical cell model is reparameterized or corrected using the measured terminal voltages, wherein the measured terminal voltages deviate from the voltage values of the corresponding voltage change curves modeled by the measured terminal voltages by more than a preset threshold.
[0035] According to one implementation, a data-based correction model can be created that provides correction values based on aging conditions and battery temperature, and the correction values are used to determine the state of charge. Attached Figure Description
[0036] The embodiments will now be described in more detail with reference to the accompanying drawings.
[0037] Figure 1 A schematic diagram of a battery-powered vehicle is shown, in which the vehicle's battery control unit is connected to a central unit to determine the state of charge at startup; Figure 2 A flowchart illustrating a method for operating a vehicle, and in particular for determining the starting state of charge, is shown. Figure 3 The curves showing the changes in the terminal voltage of the vehicle battery before, during, and after the shutdown phase are illustrated. Figure 4 The time curves of the terminal voltage modeled during the relaxation phase and the measurements of the terminal voltage at specific moments are shown; and Figure 5 The modeled correction value K is shown based on battery temperature and aging characteristics. Detailed Implementation
[0038] In the following description, the method according to the present invention will be illustrated using vehicle batteries in multiple motor vehicles (as similar devices) as an example. Here, an electrochemical battery model is implemented in a central cell outside the vehicle and is used to determine the terminal voltage change curves and state of charge change curves during the vehicle's off-state phase. The electrochemical battery model can be reparameterized or retrained in the central cell.
[0039] The examples above represent a variety of stationary or mobile devices with grid-independent energy supply systems, such as vehicles (electric vehicles, electric bicycles, etc.), facilities, machine tools, home appliances, IoT devices, etc., which communicate with a central unit (cloud) outside the device via corresponding communication connections (such as LAN, Internet).
[0040] Figure 1A system 1 is shown for collecting fleet data in central unit 2 to create, run, and evaluate an electrochemical battery model. The electrochemical battery model is used to determine the terminal voltage variation curves and state-of-charge variation curves of the vehicle battery within the motor vehicle based on provided operating variable variation curves 4 over time (especially battery current and battery temperature). Figure 1 A fleet 3 consisting of multiple vehicles 4 is shown, all of which are connected to a central unit 2.
[0041] Figure 1 The diagram shows a vehicle 4 in more detail. Vehicle 4 has a vehicle battery 41 as a rechargeable energy storage device, an electric drive motor 42, and a control unit 43. Control unit 43 is connected to a communication module 44, which is adapted to transmit data between the respective vehicle 4 and the central unit 2 (i.e., the so-called cloud). Furthermore, a battery control device 45 may be provided, which monitors the function of the vehicle battery 41 and continuously detects operating variables.
[0042] The vehicle 4 transmits operating variables F to the central unit 2 via communication module 44. These operating variables at least indicate variables characterizing the battery state of the vehicle battery 41. In the case of the vehicle battery 41, the operating variables F may include time series of battery current, battery voltage, battery temperature, and state of charge (SOC) at the pack level, module level, and / or single-cell level. The operating variables F are detected using a fast time grid from 1 Hz to 100 Hz and may be periodically transmitted to the central unit 2 in uncompressed and / or compressed form.
[0043] In addition, to minimize the data traffic to central unit 2, compression algorithms can be used to transmit the time series in blocks at intervals of several hours to several days to central unit 2.
[0044] The central unit 2 has a data processing unit 21 and a database 22 for storing data points, model parameters, states, etc. The following methods can be executed in the data processing unit 21.
[0045] An electrochemical battery model is implemented in central unit 2, which is based on data as a hybrid or semi-hybrid model. The battery model can be periodically evaluated to determine the current internal state of the vehicle battery 41 involved, based on the time-varying curves of operating variables (especially since the corresponding vehicle battery was put into service or since the last known battery state). This electrochemical battery model may include a system of differential equations that, based on differential equations parameterized by model parameters, model the internal battery state (especially the equilibrium state and, where necessary, the kinetic state) using time integration, and provide the relationship between the operating variable curves of the device battery (i.e., battery current, battery voltage, battery temperature, and the state of charge of the device battery) and the internal battery state. Such electrochemical battery models are disclosed, for example, by US2016 / 023,566, US 2016 / 023,567, and US 2020 / 150,185.
[0046] This electrochemical cell model is particularly suitable for and designed to model the corresponding or related changes in terminal voltage and state of charge based on the time-varying curves of cell current and cell temperature (depending on the preset model parameters of the cell model).
[0047] Figure 2 A schematic diagram illustrating a flowchart for a method of operating a vehicle is shown. This method is executed in conjunction with a battery control device and a central unit.
[0048] In step S1, check whether vehicle 4 is closed. If yes (option: Yes), the method continues to step S2; otherwise (option: No), it jumps back to step S1.
[0049] In step S2, the final detected change curve of the running variable is transmitted to the central unit 2.
[0050] In central unit 2, in step S3, voltage modeling is performed based on the last detected operating variable change curve, the assumed battery current of 0A, and the battery temperature (which may correspond to the last detected battery temperature or the ambient temperature derived from weather information). Voltage modeling is based on a preset electrochemical cell model implemented in central unit 2.
[0051] The voltage and state-of-charge (SOC) curves can be modeled continuously until the relaxation phase ends. This is true when the voltage gradient is less than 0.05 mV / min. Subsequently, it is assumed that the reached terminal voltage and SOC values remain constant until the start-up time. Alternatively, an electrochemical cell model can be used to model the voltage and SOC curves over a longer time period (longer than the relaxation phase).
[0052] As a result of the modeling, we obtain the voltage change curves over time during and after the relaxation phase, as well as the time-varying curves of the equivalent state of charge of the modeled voltage, starting from the turn-off moment.
[0053] In other words, if the vehicle stops (terminal 15 is disconnected), data that has not yet been transmitted to the central unit will be transmitted to the central unit in the form of data packets so that the battery model used for voltage prediction has all the necessary input data, such as the final determined state of charge, terminal voltage, battery temperature, and aging state. The electrochemical battery model generates a voltage change curve for the relaxation phase, which ends when voltage saturation is reached (standard: voltage change < threshold).
[0054] exist Figure 3 The diagram illustrates, for example, the terminal voltage variation curves before, during, and after the shutdown phase. It can be seen that after the operation phase, starting from shutdown time A, the terminal voltage rises during the relaxation phase R until it reaches an equilibrium value, and then remains essentially constant. At startup time E, the terminal voltage subsequently decreases again due to battery current flow.
[0055] In step S4, the voltage change curve during the relaxation phase, along with its equivalent state of charge (SOC) change curve, can be transmitted back to the battery control device 43. This transmission typically occurs substantially immediately after calculations performed following the vehicle's shutdown at the time of shutdown. Furthermore, a relaxation model created in the central cell, indicating the relationship between SOC and terminal voltage (potentially temperature-dependent), can be transmitted, allowing an equivalent SOC to be assigned to the measured terminal voltage (potentially depending on battery temperature).
[0056] Therefore, regardless of when the vehicle is restarted, there is a modeled terminal voltage, a measured terminal voltage, and a state of charge value derived from the comparison of the two voltage values at the start-up moment.
[0057] In step S5, check whether vehicle 4 is turned on. If it is determined that the vehicle is turned on (option: yes), the method continues to step S6; otherwise, it jumps back to step S5.
[0058] In step S6, the current terminal voltage of the vehicle battery 41 is measured using the battery control device 45.
[0059] In step S7, the modeled terminal voltage is determined based on the modeled voltage change curves during and after the relaxation phase. Furthermore, the modeled onset state of charge is determined at the start-up time based on the modeled state of charge change curves. This can be based on an evaluation of the electrochemical cell model.
[0060] In step S8, the modeled terminal voltage is compared with the measured terminal voltage. If they are consistent (considering tolerance) (option: Yes), then in step S9, the modeled starting state of charge (SOC) is determined as the actual starting SOC. If it is determined that the modeled terminal voltage deviates from the measured terminal voltage by more than a threshold amount (option: No), then in step S10, the starting SOC is determined based on the measured terminal voltage. This can be based on the SOC change curve over time during the relaxation phase and the terminal voltage change curve over time. An equivalent SOC change curve can be derived from the measured terminal voltage change curve. The voltage at startup determines the SOC at that moment.
[0061] Furthermore, in step S11, the electrochemical cell model can be reparameterized, specifically based on least squares error. Reparameterization is performed by transmitting the terminal voltage measured at the start-up moment to the central cell 2, where the reparameterized electrochemical cell model can be used to determine the next terminal voltage and state-of-charge change curve during the relaxation phase. Reparameterization should only be performed if the current sensor is functioning correctly and the cell exhibits no other abnormalities.
[0062] Furthermore, the reparameterization of the electrochemical cell model can be based on the deviation between the terminal voltage measured at the turn-on moment and the modeled terminal voltage.
[0063] Reparameterization can also be based on multiple terminal voltages measured and modeled from multiple vehicles at their respective start-up times, thereby improving the data basis for determining the parameters used in electrochemical battery modeling.
[0064] This method enables precise determination of the initial state of charge and allows for simultaneous reparameterization of the electrochemical cell model. If there are errors in the terminal voltage measurement or uncertainties in the measured terminal voltage value, the electrochemical cell model is not readjusted or reparameterized.
[0065] If data detected during the inactivity of vehicle 4 is also provided to central unit 2, this data can be used to compensate for inaccuracies in model calculations, such as inaccuracies in voltage prediction and its associated state of charge. If data is also detected after vehicle 4 has stopped or the equipment has been turned off—and, if necessary, at a higher sampling rate—and provided to central unit 2, advantages in model calculations are derived.
[0066] If the terminal voltage during the relaxation phase is used even when the vehicle is off, reparameterization can be performed in an improved manner because a larger amount of data is available to fit the electrochemical cell model.
[0067] Therefore, measurement data can still be determined during downtime, but using a higher sampling rate. To ensure that the time required until the data packet reaches its preset size and is ready for transmission is in a reasonable proportion to the downtime or relaxation time, the data packets during this period should be selected to be correspondingly smaller than the data packets during the vehicle's running time (modified data packets). If this size is adjusted so that one data packet is transmitted every 10-15 minutes, at least one data packet containing the current value can be provided to the central unit for evaluation, even in the case of short downtime.
[0068] These data packets may, among other things, include measured terminal voltage values (time-varying curves) for all individual cells. Before determining the predicted voltage variation curves during the relaxation phase, diagnostics can be performed, analyzing the measured terminal voltages to determine if: 1) The voltage sensor is faulty, and this fault leads to the exclusion of the voltage signal in all subsequent considerations; 2) The measured voltage values of each individual cell from which the time-varying curves are generated are abnormal, so these individual cells are no longer considered for evaluation and reparameterization of the electrochemical cell model. 3) The modeled terminal voltage value of the single cell with the lowest terminal voltage at the same sampling time has too large a difference from the measured terminal voltage, so the electrochemical cell model needs to be corrected.
[0069] In scenario 3), the electrochemical cell model is modified / parameterized based on the measured voltage value and optionally based on the temperature value, so that the modeled voltage change curve and the equivalent state of charge change curve can be determined more accurately subsequently. Both curves can be transmitted to the battery control device 45 so that the accurately predicted voltage trajectory during the relaxation phase can always be determined there.
[0070] The battery model can be adjusted during the shutdown phase as each data point is determined, and this process is repeated until the vehicle leaves the inactive state or is started.
[0071] Figure 4 The time-varying curve of the modeled terminal voltage U_modell (dashed line) during the relaxation phase and the measurement of the terminal voltage U_mess (cross) at a specific moment are shown. The equivalent state of charge is derived from the voltage value of the terminal voltage measured at the turn-on moment, wherein the following moments during the relaxation phase are assigned to the terminal voltage measured at the turn-on moment t_Start, at which moment the modeled terminal voltage corresponds to the terminal voltage measured at the turn-on moment t_Start, and the equivalent state of charge corresponds to the state of charge at the moment thus determined in the state of charge change curve.
[0072] It can be proposed that the terminal voltage be transmitted to the central unit only when there is a difference between the voltage value of the last updated voltage change curve and the currently measured voltage value.
[0073] Therefore, the transmission voltage variation curve and the matched equivalent state of charge are necessary so that the matched equivalent state of charge can be determined based on the measured voltage when there is a difference between the modeled voltage and the measured voltage.
[0074] In central cell 2, the electrochemical cell model can be reparameterized using the voltage values of a large number of similar cells during the relaxation phase.
[0075] A global correction model can be proposed and derived, and the training of this global correction model is as follows: - Input / features, assigned to the moment of the last time-series data measured in the vehicle before it stopped: ○ Aging state; ○ Electrochemical characteristics or parameters, such as recyclable lithium; ○ Environmental conditions, especially battery temperature; ○ State of charge; ○ and other electrochemical or physical variables, especially cumulative variables such as Ah throughput.
[0076] - Labels / Target variable for regression: ○ SoC correction value (or alternatively, absolute / relative SoC correction, or correction factor) This correction model can be trained under supervision, assigning correction values to all features present in the vehicle at the moment the last signal was measured, with respect to the measured state of charge. Advantageously, the model is constructed as a probabilistic model, particularly a sparse Gaussian process, and utilizes training data from all previously executed correction processes across all fleet members in our method, including all historical fleet data from the cloud. Other possible supervised learning models include neural networks, random forests, AdaBoost, or other data-based regression models.
[0077] Figure 5 The correction value K is visually shown based on the characteristics—battery temperature T and aging state SOH.
[0078] Once the corrected model has achieved a certain quality in central unit 2, it can be embedded for use in the vehicle. Advantageously, only the model parameters are updated here. Therefore, while the vehicle is running, just before stopping / shutting down, or in the event of a connection loss, the model can be queried and its output stored in an EPROM (Electrically Erasable Programmable Read-Only Memory), ensuring an optimal estimate of the state of charge when the vehicle is activated.
[0079] In a favorable design, the Gaussian process prediction is evaluated not only by the expected value but also by quantiles, such as the 5th percentile. Therefore, model uncertainties that depend solely on the input space of available training data points can be directly incorporated. This helps to present the driver with a realistic, but not overly optimistic, range prediction based on state of charge derivation.
Claims
1. A computer-implemented method for determining at least a portion of the startup state of a device battery (41) of a technical device (4) after a shutdown phase, comprising the following steps: - After determining that the technical device is turned off (S1), provide (S2) the last detected change curve of the operating variable of the device battery; - Using a parameterized electrochemical cell model, the terminal voltage change curve and state of charge change curve are modeled for the time period after the shutdown time of the technical device (S3). - The terminal voltage change curve and the state of charge change curve are provided (S4) in the battery control device of the device battery; - When the technical device is determined to be turned on (S5), the terminal voltage is measured at the time of turn-on (S6); - Based on the comparison between the terminal voltage measured at the start-up time and the modeled terminal voltage, a state of charge change curve modeled in (S8, S9) is provided for the state of charge at the start-up time as the starting state of charge.
2. The method according to claim 1, wherein, The terminal voltage change curve and the state of charge change curve are modeled based on the last detected operating variable change curve, the assumed battery current of 0A, and the battery temperature, which corresponds to the last detected battery temperature or the ambient temperature derived from weather information.
3. The method according to claim 2, wherein, The terminal voltage change curve and the state of charge change curve are modeled only during the relaxation phase, and it is assumed that the reached terminal voltage value and state of charge value are constant until the turn-on time; or the terminal voltage change curve and state of charge change curve are modeled for a period of time beyond the relaxation phase.
4. The method according to any one of claims 1 to 3, wherein, The operating variable change curves are provided in a central unit (2) remote from the device, wherein the terminal voltage change curves and the state of charge change curves are modeled in the central unit (2) for a time period after the shutdown moment and are subsequently transmitted back to the technical device.
5. The method according to any one of claims 1 to 4, wherein, When the terminal voltage measured at the turn-on time substantially corresponds to the terminal voltage modeled at the turn-on time, a modeled state of charge change curve is provided for the state of charge at the turn-on time as the starting state of charge, wherein a state of charge derived from the measured terminal voltage is provided as the starting state of charge.
6. The method according to any one of claims 1 to 5, wherein, When the deviation between the terminal voltage measured at the start-up time and the terminal voltage modeled at the start-up time exceeds a preset threshold, the electrochemical cell model is reparameterized or corrected.
7. The method according to any one of claims 1 to 6, wherein, In the event that there is a deviation between the terminal voltage measured at the turn-on time and the terminal voltage modeled at the turn-on time, it is assumed that the state of charge derived from the voltage value of the terminal voltage measured at the turn-on time is taken as the state of charge, wherein the terminal voltage measured at the turn-on time is assigned to the following times during the relaxation phase, at which time the modeled terminal voltage corresponds to the terminal voltage measured at the turn-on time, wherein the equivalent state of charge corresponds to the state of charge at such determined times in the state of charge change curve.
8. The method according to any one of claims 1 to 7, wherein, After the technical device is turned off and before it is turned on, one or more terminal voltages measured at different times are detected, wherein the electrochemical cell model is reparameterized or corrected using the measured terminal voltages, wherein the measured terminal voltages deviate from the voltage values of the corresponding voltage change curves modeled by the measured terminal voltages by more than a preset threshold.
9. The method according to any one of claims 1 to 8, wherein, A data-based correction model is created, which provides correction values based on aging conditions and battery temperature, and the state of charge is determined using the correction values.
10. An apparatus for performing the method according to any one of claims 1 to 9.
11. A computer program product comprising instructions that, 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 9.
12. A machine-readable storage medium comprising instructions that, 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 9.