A method for model-based estimation of the impedance of galvanic cells in secondary batteries and its use, as well as a battery cell monitoring device and vehicle.

Abstract

The invention relates to a method for model-based estimation of the impedance of a galvanic cell (1) of a secondary battery (2) by means of an estimation model (4) implemented in a calculation unit (3). The invention relates to a method for model-based estimation of the impedance of a galvanic cell (1) of a secondary battery (2), the method comprising the following method steps, which are carried out before the actual use of the cell (1): - an initial parameterization of the cell model (6); - forming a reference database (7); - fitting each model parameter reference value from the reference database (7) using a polynomial fit and storing the fitting coefficients obtained in the process in a data memory; Including, and the following subsequent method steps carried out during the actual use of the cell (1): -Measured cell voltage (U mes ) and the calculated cell voltage (U rec ) step to find the difference; - setting an amplification factor and multiplying the voltage difference determined in the previous method step by the amplification factor; and - a step of incrementally determining the impedance of cell (1), calculating the next increment of impedance by adding the current increment of impedance to the product of the amplification factor and the voltage difference calculated in the previous method step, followed by a step of updating the variable model parameters to the next increment of impedance thus calculated, taking into account the fitting coefficients.

Description

[Technical Field]

[0001] The present invention relates to a method for model-based estimation of the impedance of galvanic cells of a secondary battery using an estimation model performed in a computing unit, as defined in more detail by the broader concept of claim 1; a battery cell monitoring device, as defined in more detail by the broader concept of claim 6; the use of a method for determining the service life of battery cells in a vehicle's traction battery, as defined in more detail by claim 7; and a vehicle, as defined in more detail by the broader concept of claim 8. [Background technology]

[0002] Due to their high energy density and power density, lithium-ion batteries are essential for the development and operation of electromobility concepts in electric vehicles. Battery management systems (BMS) control battery packs used in mobile applications and ensure their optimal performance.

[0003] Accurate estimation of the health status, also known as SOH (State-of-Health), is critical to the safety of electric vehicle operation. The increase in internal resistance, also known as SOHR (State-of-Health Resistance), during the aging process is considered one of the most important degradation phenomena limiting the performance and lifespan of battery cells. Furthermore, internal resistance is extremely important for thermal analysis and temperature monitoring of cells. The internal resistance of a cell is highly dependent on temperature and the charge state, also known as SOC (State-of-Charge).

[0004] Accurate prediction of cell internal resistance prevents the design and use of batteries from unnecessarily oversized for backups in case of failure, thus ensuring that batteries are fully utilized to their actual performance and lifespan limits in real-world operation. Consequently, premature and unnecessary battery replacements are prevented, significantly improving the sustainability of electric vehicles.

[0005] Impedance, or the AC resistance of a battery cell, can be measured using electrochemical impedance spectroscopy. However, this measurement is relatively expensive and time-consuming due to the required measurement techniques and the effort involved. Therefore, measuring the impedance of battery cells in a vehicle is not practical.

[0006] Instead, the impedance curve of the battery cells in a vehicle's traction battery is estimated using a computational model during operation. The basis of such computational models is an electrical equivalent model of the battery, which in turn is based on an electronic equivalent circuit. Components included in such equivalent circuits, such as resistors, coils, and capacitors, are also affected by aging and thus influence the impedance. In other words, in order to calculate an impedance that is close to reality, it is important to accurately estimate the parameters of these electronic components over their lifespan, taking aging into account.

[0007] For example, Patent Document 1 discloses a method for estimating parameters of an impedance model for a lithium-ion battery. The parameters found by measurement are implemented in a vehicle's computing unit for use in a computational model to determine impedance. Here, the measurements are performed in the vehicle. A comprehensive overview of battery parameter estimation is also known from Non-Patent Document 1. Furthermore, the authors also introduce their own estimation model, which is described in the second paper along with its calculation results and evaluation of prediction accuracy. For details, please refer to Non-Patent Document 2. To find the relevant model parameters, a computationally complex closed-form optimization problem is solved online on the vehicle while it is in operation. For this purpose, a recursive estimator based on the least squares method is used. This is achieved solely through the linearity of the simplified computational model, because the computational resources on the vehicle are limited, and solving the optimization problem is computationally demanding and time-consuming; otherwise, a solution cannot be obtained on the vehicle within a reasonable timeframe. The temperature dependence of the model parameters is also estimated onboard. Estimating temperature dependence degrades prediction quality. Furthermore, a computationally intensive iterative genetic algorithm is used to estimate current dependence. [Prior art documents] [Patent Documents]

[0008] [Patent Document 1] DE102019127384A1 [Non-patent literature]

[0009] [Non-Patent Document 1] Y: C. Fleischer, W. Waag, H.-M. Heyn and DUSauer, On-line adaptive battery impedance parameter and state estimation considering physical principles in reduced order equivalent circuit battery models: Part 1. requirements, critical review of methods and modeling. Journal of Power Sources, 260:276-291, 2014 [Non-Patent Document 2] Y:C.Fleischer, W.Waag, H.-M.Heyn and DUSauer, On-line adaptive battery impedance parameter and state estimation considering physical principles in reduced order equivalent circuit battery models part 2.parameter and state estimation.Journal of Power Sources, 262:457-482, 2014 [Overview of the project] [Problems that the invention aims to solve]

[0010] The problem on which the present invention is based is to provide an improved model-based method for estimating the impedance of a galvanic cell in a secondary battery, which is suitable for use in a vehicle, even while the vehicle is in operation, and which offers improved predictive quality. [Means for solving the problem]

[0011] According to the present invention, this problem is solved by a model-based method for estimating the impedance of a galvanic cell of a secondary battery having the features of claim 1. Advantageous configurations and developments, as well as the use of such a method for determining the service life of battery cells in a vehicle's traction battery, a battery cell monitoring device for carrying out the method, and a vehicle equipped with such a battery monitoring device, become apparent from the claims dependent on claim 1.

[0012] A method for model-based estimation of the impedance of a galvanic cell of a secondary battery using an estimation model executed in a computing unit, wherein at least some of the model parameters of the cell's equivalent circuit are adapted over the lifespan of the cell in the cell model included in the estimation model, and the equivalent circuit includes at least resistive and capacitive elements, wherein, according to the present invention, the following method steps are taken before the cell is actually put into use: - A step of performing initial parameterization of the cell model based on a series of electrochemical impedance spectroscopy measurements of cells with the same structure; - A reference database containing the mapping between model parameter reference values ​​and corresponding impedance reference values ​​across life-life characteristic maps of cells with the same structure. - Multiple temperature-specific charge and discharge profiles, each containing the voltage progression of the cell over time, are supplied to the estimation model. -In this process, a nonlinear optimizer is used to update the model parameters to be fitted to the equivalent circuit, the difference between the measured cell voltage and the cell voltage calculated by the cell model is formed and minimized, and the model parameters found at each minimum value are used to form the model parameter reference values, and -Calculate the impedance reference value corresponding to the model parameter reference value. Steps formed by; - A step in which polynomial fitting is used to fit each model parameter reference value from the reference database, and the resulting fitting coefficients are stored in data memory. It was implemented; Furthermore, while the cell is actually being used, the following subsequent steps: - A step to determine the difference between the measured cell voltage and the cell voltage calculated using the cell model; - A step of setting an amplification factor and multiplying the voltage difference obtained in the preceding step by the amplification factor; and - A step of incrementally determining the impedance of a cell, wherein the next increment of the impedance is calculated by adding the current increment of the impedance to the product of the amplification factor and the voltage difference calculated in the previous method step, and subsequently, considering the fitting factor, updating the variable model parameters according to the next increment of the impedance thus calculated. is further developed by being implemented.

[0013] The method according to the present invention enables a particularly accurate and thus highly reliable estimation of the impedance of a battery cell, and also requires a relatively small amount of calculation. Therefore, this method is particularly suitable for applications during the operation of a vehicle. That is, in order to obtain model parameters adapted to the operating state of the battery each time, it is not necessary to perform optimization in the vehicle itself. Instead, those model parameters are first estimated outside the vehicle, in a laboratory or a test vehicle, i.e., a commercially available vehicle, and determined for all possible states using fitting. Those reference values are used for each actual application case in the (mass-produced) vehicle. The voltage difference between the measured terminal voltage and the estimated terminal voltage considered for calculating the impedance is evaluated for each sampling period, which enables a very high time resolution.

[0014] In the method according to the present invention, results regarding changes in the electrical cell characteristics over the service life are collected offline or off-board, i.e., not during the operation of the vehicle itself, for example, in a laboratory or a test vehicle. These cell characteristics are typical for cells of a specific structural type and can thus be transferred to cells of the same structure. Therefore, the method steps performed before the cell is actually used can be performed in the laboratory on any cell with the same structure, and the method steps performed while the cell is actually being used can be performed on another secondary battery with a corresponding cell of the same structure. Here, the secondary battery can be used as a voltage source for operating the machine in any machine such as a vehicle.

[0015] It is common practice to create an equivalent circuit of a cell and implement it in a computing unit. However, the problem is how to adapt model parameters that depend on aging to their physically correct values over the service life of a secondary battery. In this regard, for the type of cell under consideration, a complete parameterization of the electrical part of the estimation model or equivalent circuit and, by extension, the cell model is carried out based on the electrochemical impedance spectroscopy (EIS measurement) of a new cell, whereby the physical characteristics of this “initial” (BOL: beginning-of-life) cell are reproduced relatively physically correctly. The physical characteristics of the cell depend on temperature and state of charge. When performing EIS measurements, various sets of measurements are carried out for various states of charge between 0 and 100% and, for example, at temperatures between -20°C and +45°C. The temperature dependence is fitted based on the Arrhenius function. The Arrhenius function describes the relationship between the reaction kinetics within the battery cell for each temperature. The Arrhenius function thus found is stored and fixedly incorporated into the estimation model for later use. The dependence for various states of charge is stored in the estimation model in the form of elements of a look-up table. In addition to the temperature dependence, the dependence for various states of charge (SOC) also remains constant over the service life of the galvanic cell.

[0016] Here, the EIS measurement may be carried out on one or more individual cells or on various cells integrated in one battery module.

[0017] In the next step, the "start" parameterization found in those (one or more) new cells (BOL cells) is adapted over the lifespan of the cells. The estimation model is introduced into a battery cell monitoring device to find the transitions of all relevant model parameters, which take values ​​that change over the aging of the cells. Such a battery cell monitoring device includes detection means for detecting the output current, terminal voltage, and battery temperature at any point, in particular the cell temperature of the cell being monitored, from at least one cell of the secondary battery. Furthermore, the battery cell monitoring device includes a data memory and a computing unit on which the estimation model is implemented. Here, an already aged secondary battery containing multiple cells of the same structure is used, and emulated or simulated, or actually known, discharge and charge profiles for that secondary battery are recorded.

[0018] Such discharge and charge profiles are generated, for example, through test runs of a vehicle. For example, discharge occurs during a relatively long period of vehicle operation, such as two hours. During this time, the vehicle can also drive a corresponding electric motor in generator mode, and thus, based on regeneration, can supply electrical energy to the secondary battery to charge the battery cells. The corresponding charging process lasts for a relatively short period, on the order of a few seconds. The charge profile is, for example, the charging process actually carried out at a charging station. For example, the secondary battery is charged from a 20% charge state to an 85% charge state within a few minutes, for example, within 40 minutes. Recording of the discharge and charge profiles may be done in the vehicle itself or in a traction battery that has been removed and placed on a test bench. Subsequently, the discharge and charge profiles that have been actually found, or simulated or emulated, are recorded on this test bench.

[0019] Each cell in a secondary battery exhibits various aging degradation states over its service life. Accordingly, individual battery cells in the traction battery under consideration exhibit various impedances. A reference database is formed using an estimated model and measured characteristic quantities. Model parameters that change over the service life and through corresponding aging degradation are determined using a nonlinear optimizer. To this end, the estimated model compares measured voltage values ​​with estimated voltage values ​​and minimizes the deviation until it is smallest, ideally zero. Subsequently, the model parameters found at each minimum value are used to form model parameter reference values. These are specific values ​​(corresponding to a series of measurements performed). However, in subsequent vehicle operation, cells typically take on arbitrary impedance values, and these arbitrary impedance values ​​are not necessarily included in their series of measurements. Nevertheless, for these impedance values, in order to include the model parameters in the reference database, a fitting function corresponding to the variable model parameter over time with respect to impedance is determined for each model parameter.

[0020] For this purpose, a general polynomial fitting method is used. An n-th degree polynomial can be used. However, the use of a 2-degree polynomial, i.e., a quadratic function, has been found to be particularly advantageous due to its good accuracy-to-computation ratio.

[0021] This provides a knowledge base on what values ​​each model parameter, which is specific to aging, takes through each impedance value. This is the core idea of ​​the present invention because it allows computationally intensive optimization to be moved from the vehicle to the laboratory or test vehicle, and such optimization only needs to be performed once for each of the various types of cells.

[0022] The calculated fitting coefficients are stored in data memory, which is part of the computing device or test bench. The fitting coefficients can then be distributed from data memory to any other computing unit and implemented there. For example, this could be a battery management system for a vehicle's traction battery.

[0023] Next, an estimation model can be used to estimate the impedance of the battery cells in the vehicle's traction battery while the vehicle is in operation. This enables the estimation of the impedance of the vehicle's traction battery cells while the vehicle is running. For this purpose, the estimation model is implemented in the vehicle's computing unit, such as a separate computing unit, BMS, or battery cell monitoring device.

[0024] In an advantageous development of the method according to the present invention, the equivalent circuit of the cell includes at least one of the following additional elements: -ZARC element; - Coil element; -Finite-length Warburg elements; and / or -finite space Warburg elements.

[0025] By complementing the cell's equivalent circuit with the aforementioned elements, a particularly realistic equivalent circuit, and thus an electrical cell model, can be formed. This allows the physical processes of individual cells to be described mathematically as much as possible. Here, individual resistances represent the electrical resistance of conductive materials such as conductors in electrodes and cables in batteries, and the limited conductivity of the electrolyte.

[0026] To represent the inductive properties, an inductor can be included in the equivalent circuit. The inductive properties arise from metal-to-metal contact within the cell.

[0027] A ZARC element consists of an ohm resistor connected in parallel with a capacitance. The ZARC element replicates the effects of the double layer and charge transport, allowing for the consideration of nonlinear voltage dependence. The high-frequency portion of cell impedance can be described using the ZARC element.

[0028] The low-frequency portion of cell impedance can be described using Warburg elements. In other words, Warburg elements detect a process that describes relatively slow ion diffusion within the electrode. Finite-space Warburg elements represent a non-conductive boundary layer at maximum diffusion length, while finite-length Warburg elements represent a fully conductive boundary layer at maximum diffusion length.

[0029] In another advantageous configuration of this method, the voltage difference between the measured cell voltage and the calculated cell voltage is low-pass filtered before being multiplied by the amplification coefficient. This stabilizes the parameter simulation of the model parameters and allows for the distinction of short-term model errors from parameter learning errors as much as possible.

[0030] In another advantageous configuration of this method, the cell model considers one term each for overvoltage, hysteresis voltage, and no-load voltage to calculate the cell voltage. Here, the no-load voltage is approximated by a weighted average of the low-pass filtered difference between the terminal voltage and overvoltage, and the no-load voltage calculated by current integration. For example, the hysteresis stress is provided by a parallel software module, following the modified Plett model published by D. Wycisk, M. Oldenburger, MGStoye, T. Mrkonjic, and A. Latz, Modified Plett-model for modeling voltage hysteresis in lithium-ion cells. Journal of Energy Storage, 52:105016, 2022. The no-load voltage is approximated by a weighted average of two different voltage estimates. The first voltage estimate is obtained from the low-pass filtered difference between the terminal voltage and overvoltage, and the second voltage estimate is obtained from the no-load voltage calculated by current integration.

[0031] According to another advantageous configuration of the method according to the present invention, the amplification coefficient is determined depending on the following four terms: -Basic amplification term; - Sign term of battery current; - Power term of the absolute value of the cell current; and - Exponential decay term.

[0032] The amplification factor may be greater than 1, less than 1, or 1, and may take positive or negative values.

[0033] The sign term of the battery current can be used to determine whether the cell is currently being charged or discharged.

[0034] The power of the absolute value of the cell current controls the order of the cell overvoltage at which the residual error is smallest. Higher values ​​improve the degree of agreement at high overvoltages and increase fault tolerance at low overvoltages. The inverse relationship also holds.

[0035] The exponential decay term enables faster learning of model parameters after a parameter reset. A decay constant can be used to accelerate the learning of the reset parameters. This term can further include a coefficient that allows for slower learning from immediately after the control unit is started until the voltages of the model's RC elements are learned through dynamic charging and discharging.

[0036] A battery cell monitoring device, comprising detection means for detecting the output current, terminal voltage, and battery temperature of at least one cell of a secondary battery, as well as a data memory and a computing unit, is configured according to the present invention such that the detection means, data memory, and computing unit are designed to perform method steps carried out while the cell is actually in use. The battery cell monitoring device may be part of a battery management system or may form such a battery management system. As detection means, the battery cell monitoring device may include or be connected to current sensors, voltage sensors, and temperature sensors. The data memory is non-volatile memory and may be supplemented by volatile memory as needed. The computing unit may be a computer, a system on a chip (SoC), etc. The data memory may be integrated into the computing unit.

[0037] The method described above is used, according to the present invention, to determine the service life of battery cells in a traction battery of a vehicle equipped with at least a partially electrified powertrain, and the method steps performed while the cells are actually in use are carried out in the vehicle. As described above, the method according to the present invention can be used to estimate the aging degradation state of battery cells in a vehicle's traction battery with particular accuracy, and therefore with particular reliability. This is synonymous with the impedance of each cell. The cells are, for example, lithium-ion batteries. However, the method according to the present invention is also suitable for different types of battery cells, such as sodium-ion batteries.

[0038] In a vehicle equipped with a powertrain that is at least partially electrified, including a traction battery and a battery cell aging detection device, according to the present invention, the battery cell aging detection device has the battery cell monitoring device described above. The battery cell aging detection device may be a higher-level computing unit of the battery cell monitoring device. The battery cell monitoring device may be an integrated component of the battery cell aging detection device. For example, the battery cell aging detection device may be a battery management system into which the battery cell monitoring device is incorporated.

[0039] Another advantageous configuration of the method according to the present invention for model-based estimation of the impedance of galvanic cells in secondary batteries will become apparent from the embodiments described below in more detail with reference to the drawings. [Brief explanation of the drawing]

[0040] [Figure 1] A schematic diagram of the battery cell monitoring device according to the present invention is shown. [Figure 2] Figure 1 shows a schematic diagram of the equivalent circuit of a cell implemented in the cell model of the estimation model performed in the battery cell monitoring device shown in the diagram, and the impedance spectrum belonging to that cell. [Figure 3] This graph shows the changes in the model parameters of the components included in the equivalent circuit over the cell impedance. [Figure 4] This shows the set of equations stored in the estimation model. [Figure 5] A flowchart of the method according to the present invention for estimating the impedance of a galvanic cell in a secondary battery is shown. [Modes for carrying out the invention]

[0041] The battery cell monitoring device 8 shown in Figure 1 includes a computing unit 3 used to run estimation model 4. Using the battery cell monitoring device 8, cell 1 of the secondary battery 2 is monitored, thereby estimating the impedance of cell 1, which depends on the aging of the secondary battery 2 over its service life. The estimated impedance can be used to provide information about the aging of the secondary battery 2.

[0042] Each cell 1 of the secondary battery 2 may be provided with its own battery cell monitoring device 8, however, multiple cells 1 may be monitored by a single identical battery cell monitoring device 8. In the estimation model 4, the input quantities are the current I flowing through each cell 1 and the measured voltage U. mes The terminal voltage U, also known as the terminal voltage, the secondary battery 2, preferably the temperature T of the cell 1 to be monitored, and the state of charge (SOC) estimated by calculation of each cell 1 are used. The state of charge (SOC) is provided by a general method and is obtained, for example, by a battery management system (BMS). The battery management system (BMS) may be part of the calculation unit 3, or it may be provided separately from the calculation unit 3, as shown in the figure.

[0043] Estimated model 4 is the measured voltage U mes It may have a phase matching module 9 for matching the phase. The cell model 6 has an electrical cell model 6.1 and a thermal cell model 6.2. The electrical cell model 6.1 has the resistance or power loss P of the cell 1 under consideration as an output quantity. loss The output is, and the thermal cell model 6.2 is, the temperature T of each cell 1. abg The output is as follows. Furthermore, the estimation model 4 has a reference database 7 which contains the changes in model parameters included in the electrical cell model 6.1 of the equivalent circuit over the service life.

[0044] Using cell model 6, calculate voltage U ber The following is calculated: Measured voltage U mes and calculated voltage U berThese characteristics are coupled in the subtractor 10 and subtracted from each other, thereby identifying the difference between them. Here, it is important to minimize this difference, called the error ε. That is, the error ε is 0 when the model parameters of the equivalent circuit 5 or the electrical cell model 6.1 shown in Figure 2, which depend on the lifespan of cell 1, match the actual physical values. The error ε can optionally pass through the filter module 11 for smoothing, for example, using a low-pass filter. Subsequently, the error ε is supplied to the control system 12, for example, a PID controller. As a result, the control system 12 supplies the desired aging state of cell 1 in the form of estimated internal resistance, i.e., impedance. Impedance is also referred to as SOHR (State-of-Health-Resistance). The impedance is fed back in the estimation model 4 and used as an input quantity for cell model 6. Similarly, impedance can be derived from the estimation model 4, or even from the calculation unit 3, to be provided as an input quantity for another calculation model, for example, as an input quantity for a driver assistance system.

[0045] In the structure of Estimation Model 4 illustrated here, the impedance is calculated iteratively, i.e., incrementally. Therefore, the most recent increment of the impedance, which has been fed back, is used to calculate the next increment.

[0046] Figure 2 shows the structure of the electrical equivalent circuit 5 for each cell 1. This equivalent circuit 5 includes a resistive element R, two consecutive ZARC elements ZARC, and two subsequent consecutive Warburg elements. Here, the two Warburg elements are a finite-space Warburg element FSW and a finite-length Warburg element FLW. The ZARC element itself includes a capacitive element C connected in parallel with the resistive element R.

[0047] In Figure 2, below the equivalent circuit 5, the impedance spectra of each element, as well as the impedance spectrum of the complete model, are shown. The horizontal axis plots the real part of the impedance, and the vertical axis plots the imaginary part of the impedance in ohms.

[0048] In order to execute the estimated model 4 in a vehicle at high speed and with computational efficiency, a correlation is needed that describes how each model parameter of the equivalent circuit 5, namely impedance or SOHR (State-Of-Health-Resistance), takes value depending on the aging degradation state of cell 1. These correlations are stored in the reference database 7 shown in Figure 1.

[0049] Figure 3 illustrates the process for retrieving information stored in the reference database 7, using a graph. The provided calculation model is used to analyze an actual battery in a test environment. The actual battery contains multiple cells 1 with varying degrees of aging. The calculation model is pre-parameterized based on EIS measurements of new cells. Here, the progression of the impedance model parameter is identified. Exemplarily, Figure 3 shows the progression for individual resistance R. In this regard, simulated or emulated actual discharge and charge profiles for an aged secondary battery 2 are recorded, for example, in a laboratory, test bench, or test vehicle. An optimization algorithm is used to identify the progression of individual model parameters related to impedance or SOHR (State-Of-Health-Resistance). The optimization algorithm has the task of minimizing the error ε in the estimated model 4. In doing so, each value of the model parameter is changed. This yields the corresponding model parameter for each measurement point in the characteristic maps of the discharge and charge profiles for each individual cell 1 of the battery (black circles in the graph).

[0050] However, in actual operation, values for all impedance values are required, that is, not only the locations where the black dots are located in the graph but also continuous progression is required, so fitting is performed, thereby obtaining the best-fit curve 13. The fitting coefficients obtained to determine the best-fit curve 13 are stored in the reference database 7. Therefore, for all values of impedance, that is, SOHR (State-Of-Health-Resistance), it is realized to obtain the adapted model parameter values. In this case, in the calculation unit 3, that is, in each vehicle, there is no need to perform optimization again during later operations, so this significantly reduces the calculation cost, and thus enables a very fast implementation of the method according to the present invention and, consequently, the calculation of impedance.

[0051] Figure 4 shows a set of equations that can be stored in the calculation model.

[0052] Equation 1 G-1 describes an equation for defining various model parameters. Term 1-0 represents each model parameter. Term 1-1 describes the order of each model parameter and is used to scale all model parameters to the same order. Term 1-2 describes the coefficient. Term 1-3 represents the original impedance. The impedance indicates the exponent j. Ideally, j = 2. This is a compromise between sufficiently fast calculation and sufficiently accurate results. This equation is a quadratic polynomial.

[0053] Equation 2 G-2 is used to describe the voltage difference in the subtractor 10. Term 2-0 describes the voltage difference between the measured voltage U mes and the calculated voltage U ber The index "n" refers to the current increment or iteration. That is, during the sampling step or the calculation step, multiple iterations are performed. Term 2-1 describes a filter constant that can take a value between 0 and 1, where α is much smaller than 1. Term 2-2 corresponds to the measured voltage U mes for the current increment, and term 3-0 corresponds to the calculated voltage U for the current incrementber This corresponds to the voltage difference for the preceding increment "n-1".

[0054] Equation 3 G-3 is the calculated voltage U ber It is used to describe the following. Term 3-1 corresponds to the overvoltage. Term 3-2 corresponds to the no-load voltage. Term 3-4 corresponds to the hysteresis voltage. Term 4-0 corresponds to the low-pass filtered difference between the clamp voltage and the overvoltage, and term 3-3 corresponds to the no-load voltage calculated by current integration. The characteristic quantity β is a weighting constant and a filter constant, and can take values ​​between 0 and 1.

[0055] Equation 4 G-4 describes the calculation of term 4-0. Term 4-1 is the measured voltage U mes This describes the difference between this value and the overvoltage. Item 4-2 describes the preceding increment of item 4-0.

[0056] Equation 5 G-5 is used to describe the impedance at the next increment "n+1", i.e., SOHR (State-Of-Health-Resistance). Term 5-0 corresponds to the impedance of cell 1. This consists of the product of terms 6-0 and 2-0 added to term 5-1, i.e., the impedance for the current increment "n", where term 6-0 corresponds to the amplification factor and term 2-0 corresponds to the voltage difference mentioned above.

[0057] Equation 6 G-6 represents the calculation rule for determining the amplification coefficient. Term 6-1 shows the basic amplification for feedback of the voltage difference. Term 6-2 describes the sign of the battery current. Term 6-3 describes the power term of the absolute value of the cell current and controls the order of the cell overvoltage where the residual error is smallest. The characteristic quantity v can take values ​​between 0 and 1. Higher v values ​​improve the degree of agreement when the overvoltage is high and improve fault tolerance when the overvoltage is low. The inverse relationship also holds. Term 6-4 describes the exponential decay term. Term 6-4.1 is the time t after parameter reset. absThis enables K0''-faster parameter learning. The damping constant λ determines the temporal decay rate of this additional amplification to accelerate the learning of the reset parameters. Section 6-4.2 enables K0''-faster learning from immediately after the startup of the computing unit 3 until the voltages in the model's RC elements are learned by dynamic charging and discharging.

[0058] Figure 5 shows an implementation of the estimated model 4 in an embedded system in the form of a flowchart, illustrating the method steps of the method according to the present invention performed in a vehicle. The embedded system is formed by a computing unit 3 representing a discrete-time control device. Preferably, this is a fixed-point implementation, which improves performance while reducing memory consumption compared to a floating-point implementation in the embedded system. In Figure 5, preferably, the computing unit 3 may be a vehicle battery management system. For example, the tick rate of the battery management system is set to a frequency of 50 Hz.

[0059] The real-time operating system (RTOS) of the computing unit 3 is triggered by an initialization event to execute an initialization call 550, which in turn initializes the cell model 6. For this purpose, during the last execution time, i.e., before the shutdown of the computing unit 3, the stored values ​​for the model parameters and their corresponding voltages are read from the non-volatile memory (EEPROM) and subsequently written to the volatile memory (RAM). In step 501, the model parameter curves dependent on aging are read from the reference database 7. In step 502, the last processed voltage values ​​are read. In step 503, the voltage distribution for the RC elements of the cell model 6 is read. In step 504, the RC voltages are newly calculated before the computing unit 4 enters sleep mode.

[0060] Next, the computing unit 3 or the real-time operating system RTOS performs a cyclic call to the executable main function of the estimation model 4 in method step 560. In method step 505, the measured cell voltage, current, and estimated cell temperature are detected. At this time, if necessary, phase matching of the measured voltage values ​​is performed to synchronize the current and voltage signals. Subsequent method steps 506 to 515 can be performed sequentially for multiple cells 1 of the secondary battery 2, preferably all cells 1. For this purpose, method steps 506 to 515 can be implemented, for example, as a FOR loop repeated across individual cells 1. Here, in method step 506, the cell repetitions are counted.

[0061] In step 507 of the method, model parameters that depend on aging for the current impedance are calculated, and in doing so, values ​​written to volatile memory RAM from reference database 7 are processed.

[0062] The Arrhenius equation and its correlation with temperature, as mentioned earlier, are also taken into consideration. This is done in method step 508, along with the calculation step that considers the charge state (SOC). In method step 509, the sleep duration of the computing unit 3 is read and passed to the model. This is necessary after the restart in method step 511 to determine the voltage values ​​of the RC elements. Therefore, a corresponding voltage value needs to be read from the volatile RAM memory before the computing unit 3 enters sleep mode. Here, in processing step 510, it is checked whether this is the initialization after the computing unit 3 has started up.

[0063] In step 512 of the method, forward calculations are performed to calculate terminal voltage, cell voltage, and heat loss as output signals. The aging factor to be considered, i.e., impedance, is tracked via voltage value feedback. The updated aging coefficient, i.e., the impedance calculated for subsequent increments, is fed back to fit the calculation of model parameters. In step 513 of the method, it is checked whether the current is above the current minimum, and at that time, whether the change in current is above the limit value corresponding to that change in current. If affirmative, in step 514 the impedance is updated and written to volatile memory RAM. In step 515 of the method, it is checked whether the cell is the last cell 1 to be monitored.

[0064] In step 570 of the method, the real-time operating system (RTOS) triggers a shutdown sequence. For this purpose, in step 516, the voltages of the Warburg elements and the current model parameters (i.e., model parameters adapted for aging) are written to the volatile memory EEPROM for each cell 1. In step 517, the voltage distribution of each Warburg element for the RC voltage is similarly written to the non-volatile memory EEPROM, but only for a single cell 1. In step 518, the impedance of each cell 1, i.e., the aging coefficient, is stored. This completes the method.

Claims

1. A method for model-based estimation of the impedance of a galvanic cell (1) of a secondary battery (2) using an estimation model (4) executed in a computing unit (3), At least some of the model parameters of the equivalent circuit (5) of the galvanic cell (1) implemented in the cell model (6) included in the estimation model (4) are adapted over the service life of the galvanic cell (1), and the equivalent circuit (5) includes at least a resistive element (R) and a capacitive element (C) in the method, The following method steps are performed before actually using the galvanic cell (1): - A step of initializing the cell model (6) based on a series of measurements of a galvanic cell (1) with the same structure at various charge states and temperatures using electrochemical impedance spectroscopy; - A reference database (7) is provided, which includes the correspondence between model parameter reference values ​​and impedance reference values ​​corresponding to the model parameter reference values, across the service life characteristic maps of galvanic cells (1) having the same structure. - Multiple temperature-specific charge and discharge profiles, each including the voltage progression of the galvanic cell (1) over time, are supplied to the estimation model (4). ・At that time, a nonlinear optimizer is used to update the model parameters to be adapted to the equivalent circuit (5), and the measured cell voltage (U) obtained from the terminal voltage of the galvanic cell (1) is determined. ｍｅｓ ) and the cell voltage (U) calculated by the cell model (6) obtained from the current flowing through the galvanic cell (1) and the impedance of the equivalent circuit (5). ｒｅｃ The difference between ) is formed and minimized, and the model parameters found at each minimum value are used to form the model parameter reference values, and - Calculate the impedance reference value corresponding to the model parameter reference value. Steps formed by; - A step of fitting each model parameter reference value from the reference database (7) using polynomial fitting, and storing the fitting coefficients obtained in order to determine the best-fit curve (13) in data memory; Furthermore, the following method steps are performed while the galvanic cell (1) is actually in use: - The measured cell voltage (U) obtained from the terminal voltage of the galvanic cell (1) ｍｅｓ ) and the cell voltage (U) calculated by the cell model (6) obtained from the current flowing through the galvanic cell (1) and the impedance of the equivalent circuit (5). ｒｅｃ The step of finding the difference between ) and ); - A step of setting an amplification coefficient and multiplying the difference in voltage obtained in the method step of determining the difference by the amplification coefficient; and - A step of incrementally determining the impedance of the galvanic cell (1), comprising: calculating the next increment of the impedance by adding the current increment of the impedance to the product of the amplification coefficient and the difference of the voltage calculated in the multiplication method step; and then updating the variable model parameter to match the next increment of the impedance thus calculated, taking into account the fitting coefficient. The method characterized by the above.

2. The measured cell voltage (U ｍｅｓ ) and the calculated cell voltage (U ｒｅｃ The method according to claim 1, characterized in that the voltage difference between ) is low-pass filtered before being multiplied by the amplification coefficient.

3. A battery cell monitoring device (8) comprising detection means for detecting outputtable current, terminal voltage and battery temperature from at least one galvanic cell (1) of a secondary battery (2), and a data memory and a calculation unit (3), in the battery cell monitoring device (8), The battery cell monitoring device (8) is characterized in that the detection means, the data memory, and the calculation unit (3) are designed to perform a method step of the method according to claim 1 or 2, which is performed while the galvanic cell (1) is actually in use.

4. A method according to claim 1 or 2 for determining the service life of battery cells in a traction battery of a vehicle having at least a partially electrified powertrain by estimating the aging degradation state of the battery cells, The use wherein the method steps performed while the galvanic cell (1) is actually in use are performed in the vehicle.

5. In a vehicle equipped with a powertrain that is at least partially electrified, including a traction battery and a device for identifying the aging degradation of battery cells, The vehicle is characterized in that the battery cell aging deterioration identification device has the battery cell monitoring device (8) described in claim 3.

Images