Method for estimating a health status of a battery cell of an electrical energy store, computer program product and electronic computer unit

Abstract

The invention relates to a method for estimating a health status (18) of a battery cell of an electrical energy store of an at least partially electrically operated motor vehicle (12) by means of an electronic computer unit (10) of the motor vehicle, comprising the following steps: providing a mathematical model for estimating the health status (18); determining at least one differential voltage curve (34) of the battery cell during a charging process (36) of the electrical energy store by means of current filtering, voltage filtering and mathematical derivation; determining modal coordinates (44) of the at least one differential voltage curve (34), and estimating the health status (18) by evaluating the modal coordinates (44) by means of the electronic computer unit (10). The invention also relates to a computer program product and to an electronic computer unit (10).

Description

[0001] Method for estimating the health status of a battery cell of an electrical energy storage device, computer program product and electronic computing device

[0002] The invention relates to a method for estimating a state of health of a battery cell of an electrical energy storage device of an at least partially electrically operated motor vehicle by means of an electronic computing device of the motor vehicle according to the preamble of patent claim 1. Furthermore, the invention relates to a computer program product and an electronic computing device.

[0003] So-called on-board methods, in other words, methods that are performed inside the vehicle, such as differential voltage analysis (DVA), are already known from the state of the art. However, these only consider a so-called peak position and peak height and thus fail in cases of inhomogeneous aging, for example, when no peaks are identifiable. Previous methods that identify so-called peak vanishing are based on electrochemical models. Therefore, they are not suitable for on-board use.

[0004] DE 10 2015 016 987 A1 relates to a method for detecting degradation of a rechargeable battery cell, comprising the steps of: detecting the terminal voltage of the rechargeable battery cell during at least one charging and / or discharging process of the rechargeable battery cell; mathematically deriving the terminal voltage according to the state of charge or the electrical charge flowing during the at least one charging and / or discharging process of the rechargeable battery cell and forming a first corresponding data set; determining the height of at least one peak in the first data set;Comparing the height of the at least one peak determined in the first data set with the height of the at least one corresponding peak in the corresponding second data set for the same rechargeable battery cell type or the rechargeable battery cell in a state with fewer charging and / or discharging processes than in the first step, preferably in the new state; and determining a degradation status and / or calculating or estimating a degree of degradation of the rechargeable battery cell based on a lower height of the at least one peak determined in the first data set compared to the height of the at least one corresponding peak in the second data set, as determined by the comparison;

[0005] DE 10 2018 132 083 A1 relates to a method for determining properties of a cell of a lithium-ion battery, in particular a traction battery of a motor vehicle, wherein the cell has a first half-cell and a second half-cell. The method comprises, among other things, the following method steps: determining a first initial average open-circuit voltage profile of the first half-cell by detecting a first initial open-circuit voltage of the first half-cell over the duration of several initial charging and / or discharging processes of the battery at the beginning of the battery's service life, wherein the first initial average open-circuit voltage profile represents the average course of the open-circuit voltage of the first half-cell over the several initial charging or discharging processes of the battery at the beginning of the battery's service life.

[0006] The object of the present invention is to provide a method, a computer program product and an electronic computing device by means of which the health status of a battery cell can be determined in an improved manner.

[0007] This object is achieved by a method, a computer program product, and an electronic computing device according to the independent patent claims. Advantageous embodiments are specified in the subclaims.

[0008] One aspect of the invention relates to a method for estimating the health status of a battery cell of an electrical energy storage device of an at least partially electrically powered motor vehicle using an electronic computing device of the motor vehicle. A mathematical model is provided for estimating the health status. At least one differential voltage curve of the battery cell is determined during a charging process of the motor vehicle using current filtering, voltage filtering, and mathematical derivation. Modal coordinates of the at least one differential voltage curve are determined, and the health status is estimated by evaluating the modal coordinates using the electronic computing device. In particular, an improved method for estimating the health status of the battery cell can thus be provided.In particular, the mathematical model or the applicable algorithm utilizes effects from the electrochemical model and transmodal effects from measurements, yet is provided model-free. The advantage of the proposed method is that the cell capacity is estimated from only one partial charge, specifically from approximately 40 to 80 percent of the so-called State of Charge (SOC) relative to the so-called Beginning-of-Life (BOL) cell capacity. This represents a clear advantage over classical capacity estimation algorithms based on a linear regression of charge and voltage. The introduced algorithm offers greater stability and robustness compared to conventional DVA-based estimation methods that only consider peak position and height. The proposed method utilizes all the information contained in the differential voltage curve of the partial charge.The most striking advantage of the proposed method is that it is reliable even for inhomogeneously aged battery cells with vanished peaks. Based on such a fast, accurate capacity estimation, both a better determination of the state of charge, especially the SOC, and a reduction in state of charge jumps are possible compared to the state of the art. Furthermore, this multidimensional degradation prediction achieves an improved estimation of the aging state and thus of the battery replacement criterion. This leads to greater sustainability in battery replacement.

[0009] In particular, the invention proposes an accurate low-rank estimate of cell capacity by observing charging processes from, in particular, approximately 40 percent SOC to approximately 80 percent SOC. This can provide, in particular, an accurate low-rank estimate of the loss of active cyclable lithium, particularly in a battery cell designed as a lithium cell, and of the loss of active material at the anode and cathode. Furthermore, an automated evaluation of differential voltage curves with vanishing peaks in inhomogeneous, aged cells can be enabled.

[0010] In particular, the invention relates to an on-board method for estimating the capacity and multidimensional aging of lithium-ion battery cells, in particular, based on differential voltage analysis (DVA). The proposed algorithm is based on data-driven modal reduction through orthogonal eigenmode decomposition. Insights from measured capacities, differential voltage curves (DV curves), and electrochemical simulations are encoded offline in low-rank patterns suitable for on-board application. A routine is introduced that can represent each differential voltage curve with five modal coordinates of the dominant aging dynamics.This particularly concerns cell capacity, aging homogeneity, loss of active lithium, and loss of active material at the anode and cathode, which is estimated on-board using a parameterized polynomial function of the modal coordinates. Such fast and accurate capacity estimation enables both a better determination of the state of charge and a reduction in state of charge jumps compared to the state of the art.

[0011] According to an advantageous embodiment, the mathematical model is provided historically based on defective battery cells and / or on non-defective battery cells. In particular, a one-time parameterization can be provided, which can be performed offline, for example. The corresponding differential voltage curves can be provided from cycling data of homogeneously and inhomogeneously aged battery cells.

[0012] It is also advantageous if differential voltage curves are decomposed into orthogonal modes in the mathematical model using singular value decomposition. For this purpose, it can be predefined, for example, that the differential voltage curves are normalized with normalized charge axes. The charge axis of the differential voltage curve can be normalized, for example, using a so-called beginning-of-life cell capacity, and then the differential voltage curves can be truncated, for example, at QNorm.min to ONO™,max. Q stands for the charge. The orthogonal modes are decomposed using singular value decomposition, also known as Singular Value Decomposition (SVD).

[0013] Furthermore, it has proven advantageous to determine a termination order for a modal reduction. In particular, the corresponding termination order is determined from the orthogonal modes, which are then determined for the modal reduction.

[0014] A further advantageous embodiment provides for modes to be normalized depending on the determined termination order. In particular, the termination order of the first dominant modes can be normalized, and the remaining modes can be omitted. It is also advantageous if the method is carried out at a battery cell charge level between 40 percent and 80 percent. This allows the method to be applied advantageously.

[0015] It is also advantageous if battery cell degradation is determined from orthogonal eigenmode decomposition and reconstruction of differential voltage curves. In particular, this allows cell voltage and current to be filtered during a charging process from approximately 40 percent SOC to 80 percent SOC, i.e., a charging cycle of approximately 40 percent. The current can then be integrated, and the voltage after charging can be mathematically derived.

[0016] Furthermore, it has proven advantageous to determine battery cell degradation from the orthogonal eigenmode decomposition and reconstruction of differential voltage curves. In particular, the modal coordinates can be calculated, and the reconstruction error can optionally be output as an indicator of the quality of the estimate.

[0017] It is also advantageous if an offset in a charge axis of a differential voltage curve is eliminated by minimizing the reconstruction error with an orthogonal eigenmode decomposition.

[0018] The method presented is, in particular, a computer-implemented method. Therefore, a further aspect of the invention relates to a computer program product with program code means that cause an electronic computing device, when the program code means are processed by the electronic computing device, to perform a method according to the preceding aspect. The computer program product can also be referred to as a computer program. A further aspect of the invention therefore also relates to a computer-readable storage medium containing the computer program product.

[0019] The invention also relates to an electronic computing device for estimating the health status of a battery cell of an electrical energy storage device of an at least partially electrically powered motor vehicle, wherein the electronic computing device is designed to carry out a method according to the preceding aspect. In particular, the method is carried out by means of the electronic computing device. The electronic computing device comprises, for example, processors, circuits, in particular integrated circuits, and other electronic components in order to be able to carry out corresponding method steps.

[0020] The invention also relates to a motor vehicle with an electronic computing device according to the preceding aspect. The motor vehicle is designed, in particular, as an at least partially electrically powered or fully electrically powered motor vehicle.

[0021] Advantageous embodiments of the method are to be regarded as advantageous embodiments of the computer program product, the electronic computing device, and the motor vehicle. The motor vehicle and the electronic computing device, in particular, have physical features for this purpose in order to be able to carry out corresponding method steps.

[0022] Further advantages, features, and details of the invention will become apparent from the following description of a preferred embodiment and from the drawings. The features and combinations of features mentioned above in the description, as well as the features and combinations of features mentioned below in the description of the figures and / or shown alone in the single figure, can be used not only in the respective combinations specified, but also in other combinations or on their own, without departing from the scope of the invention.

[0023] The single figure shows a schematic block diagram according to an embodiment of the method.

[0024] In the figure, identical or functionally identical elements are provided with the same reference numerals.

[0025] The figure shows a schematic block diagram according to one embodiment of the method. In particular, an electronic computing device 10 is shown for a purely schematically illustrated motor vehicle 12, wherein the motor vehicle 12 can be designed in particular as an at least partially electrically operated motor vehicle 12 or also as a fully electrically operated motor vehicle 12. The figure shows in particular a first block 14, which is carried out in particular offline, i.e. only relates to a one-time preparation for the method according to the invention. Furthermore, a further block 16 is shown, which is carried out in particular on-board, in other words in real time, within the motor vehicle 12. In particular, the principle of a modal differential voltage analysis is thus shown in the figure.

[0026] The figure shows, particularly in block 14, that measured cell capacities and an electrochemical simulation are provided, for example, to estimate the state of health 18, which can also be referred to as SOHx (State of Health). This is illustrated in particular by block 20. This model relates in particular to the cell capacities 22, the reference loss of lithium 24, the reference loss of active material at the anode 26, and the reference loss of active material at the cathode 28. This is then summarized in particular as the so-called state of health parameterization 30. In a further block 32, it is specified in particular that the differential voltage curves of homogeneously and inhomogeneously aged cells are provided there. In this case, a singular value decomposition is then also carried out in block 32.This is followed by the generation of orthogonal modes in a first step S1, the generation of truncated modes in a second step S2, the normalization of the modes in a third step S3, the determination of the modal coordinates 44 in a fourth step S4, and the function parameterization in the fifth step S5, which is also represented by block 30. The corresponding differential voltage curves 34 are transferred to both block 20 and the fourth step S4.

[0027] In block 16, i.e., in particular, in the on-board evaluation, a so-called AC charging process 36 takes place, whereby the AC charging process 36 enables so-called alternating charging of the battery cell. In particular, a voltage 38, a current 40, and a time 42 are determined. The differential voltage curve 34 for the current charging process 36 is then determined. Furthermore, a charge quantity is also determined. The modal coordinates 44 are then determined, and subsequently, the state of health 18 is estimated.

[0028] In particular, the figure thus illustrates a method for estimating the state of health 18 of a battery cell (not shown) of an electrical energy storage device (not shown) of an at least partially electrically powered motor vehicle 12. The mathematical model for estimating the state of health 18 is provided, which is represented in particular by block 14. This determines at least one differential voltage curve 34 of the battery cell during the charging process 36 of the motor vehicle 12 by means of current filtering, voltage filtering, and mathematical derivation. The modal coordinates 44 of the at least one differential voltage curve 34 are then determined, and the state of health 18 is estimated by evaluating the modal coordinates 44 using the electronic computing device 10.

[0029] In particular, it can be provided that the mathematical model is provided historically based on defective battery cells and / or on the basis of non-defective battery cells. Furthermore, in the mathematical model, differential voltage curves 34 can be decomposed into orthogonal modes using a singular value decomposition. Furthermore, a termination sequence r is determined for a modal reduction. Furthermore, depending on the determined termination sequence r, modes can be normalized. Furthermore, the method can be carried out at a battery cell state of charge between 40 percent and 80 percent.

[0030] It can also be provided that a degradation of the battery cell is determined from the orthogonal eigenmode decomposition and a reconstruction of differential voltage curves 34. It can also be provided that an offset in a charge axis of a differential voltage curve 34 is eliminated by minimizing the reconstruction error with an orthogonal eigenmode decomposition.

[0031] Lithium-ion batteries (LIBs) play a crucial role in the implementation and success of electromobility concepts due to their high energy and power density. The battery packs used in mobile applications are monitored by a battery management system (BMS), particularly the electronic computing device 10, to ensure their reliable and optimal use. Safe operation by the BMS requires compliance with the battery operating limits and an accurate condition assessment for various aging and temperature ranges.

[0032] The BMS relies on an accurate estimate of the state of charge (SOC) and the remaining energy available (SOE) to ensure the reliability of the vehicle system. The SOC estimate depends on the actual cell capacity, the internal resistance, and thus on the state of health (SOH). The SOH is known to have multiple dimensions related to at least eight underlying electrochemical aging mechanisms. The most important degradation mechanisms include the formation of the solid electrolyte interface (SEI), its thickness growth, and decomposition on the anode surface.Depending on the cell, other dominant effects include the formation and growth of the cathode-electrolyte interface (CEI), the metallic lithium coating on the graphite anode, electrolyte depletion, particle cracking, loss of electrical contact, and corrosion of the current collectors.

[0033] Precisely predicting a battery's remaining service life requires a multidimensional aging estimate that takes into account both cell capacity reduction and losses of cyclable lithium and active material at the anode and cathode. Only such deep insights into the battery's aging state prevent unnecessarily large failure margins in battery design and use. This allows the actual operating and service life limits to be more closely achieved in real-world operation. Consequently, the battery's service life can be optimally utilized. This prevents unnecessarily premature battery replacement, significantly increasing the efficiency and sustainability of electromobility.

[0034] Thus, a novel on-board DVA algorithm is proposed using proven signal processing algorithms. The new approach is based on proper orthogonal decomposition (POD). The data-driven method uses singular value decomposition (SVD) as a stable matrix decomposition that always exists.

[0035] The approach describes a method for encoding the dynamic, complex nonlinear behavior of a battery cell system into its dominant low-rank patterns. The charge range is limited to QNorm.min to QNorm.max to reliably avoid potential operational limitations of the cell over its entire lifetime. nOrm is the charge normalized to the BOL cell capacity. A set of orthogonal modes describing the progression of the aging dynamics is generated by online analysis of a dataset of cycled cells. The dataset contains DV curves of homogeneous and inhomogeneous aged cells. The eigenmode decomposition is performed using SVD in Eq. (1). The number of decomposed DV curves determines the maximum total number of generated modes m. The number of modes necessary to accurately represent the aging dynamics in the DV curves is called the truncation order r / termination order r. The optimal truncation order r results from the number of dominant modes with singular values ​​(diagonal values ​​of the matrix £) greater than a threshold value to be determined o min. The latter denotes the threshold of singular values ​​that distinguish important from negligible aging patterns or represent signal noise. The first mode dominates the aging evolution. The contribution of the remaining modes to the dynamics typically decreases exponentially. Therefore, only the first r modes are retained, and the remaining modes are truncated.

[0036] Since the modes participate to varying degrees in the aging dynamics, their magnitudes vary. The modal reconstruction of DV curves in the reduced r-dimensional subspace of the dynamics in Eq. (2) is based on the idea of ​​separation of variables. To enable comparability of the modal coordinates, mode normalization is performed. For this purpose, the calculated eigenmodes are divided by their Euclidean norms. The normalized modes qjj retain the dimension of a differential stress.

[0037] Shifts in the Q nO rm coordinates must be prevented during the modal reconstruction of a partial DV curve to eliminate model reduction errors. By exploiting the orthogonality of the modal basis, the range of a loading process of Q s to Q s +AQ, where AQ is the charge amount calculated by current integration and normalized with respect to the BOL cell capacity.jjQs+ Q denotes the segment of modes for the modal reconstruction of the partial charging process.

[0038] The squared error of the modal reconstruction increases when an uncertainty offset factor δQ is included in the Q nOrm-axis is introduced. This corresponds to the initial error in the capacity estimation. The modal reconstruction of DV curves with a shifted Qnorm axis increases the reconstruction error. The reconstruction error reaches its minimum at exactly zero shift öQ = 0 in the Q nO rm axis. The key for the initial capacity estimation is to minimize the reconstruction error of a charging process with respect to the offset in the ONO™ ZU axis, as shown in Eq. (3). The solution of the proposed optimization problem proceeds iteratively by making an initial estimate of the Q nOThe rm coordinate is varied incrementally to minimize the reconstruction error, as in the gradient method. In other words, a moving window of modes with a width of AQ is used for the modal reconstruction of the partial charging process while the reconstruction error is observed. Since the reconstruction error is a convex function, only one direction (charging or discharging direction) is iterated until the error nearly disappears. Once the starting point Qo of the charging process is estimated, the mode segment for modal reconstruction is identified. Thus, the modal coordinates and, consequently, the cell capacity can be calculated.

[0039] With: Q s > QNorm, min and Q s +AQ < QNorm, max, especially Q s > 0.1 and Q s +AQ < 0.8 (3)

[0040] The estimator in Eq. (4) calculates the modal coordinate vector after a (partial) loading process k. Here, i denotes the observation matrix of the estimator. The optimal value of the estimator variable k is the projection of the DV curve (dV / dQ) in the r-dimensional subspace of dynamic aging evolution spanned by the orthogonal modes qjj.

[0041] The aging variable State-of-Health-Inhomogeneity (SOHIH) is estimated, which quantifies the inhomogeneity of the distribution of degradation mechanisms in the cell from the modal coordinates. To identify the progression of inhomogeneous aging in the cell, the following features are defined for pattern recognition purposes: a) Beginning-of-Life (BOL) DV curves show homogeneous cells; b) End-of-Life (EOL) DV curves show inhomogeneously aged cells; c) Over the course of aging, the inhomogeneity tends to increase; the cell capacity exhibits a dominant, usually linear inverse proportionality to the first modal coordinate. The quantification of the reference losses due to degradation is calculated by matching the relative full-cell capacities and reconstructing the shape of the measured DV curves based on electrochemical simulation models using methods from the literature. This step is performed online.From the measured cell capacities and reference losses for cyclable lithium and active material at the anode and cathode, the SOHx function in Eq. (5) is parameterized online for the mentioned degradation mechanisms using an optimizer.

[0042] For on-board applications, after calculating the modal coordinates, only the equation with the appropriate parameterization is evaluated on the BMS.

[0043] List of reference symbols

[0044] 10 electronic computing device

[0045] 12 Motor vehicle

[0046] 14 blocks

[0047] 16 blocks

[0048] 18 State of health

[0049] 20 blocks

[0050] 22 cell capacity

[0051] 24 Reference loss of cyclable lithium

[0052] 26 Reference loss of active material at the anode

[0053] 28 Reference loss of active material at the cathode

[0054] 30 Function parameterization

[0055] 32 blocks

[0056] 34 differential voltage curves

[0057] 36 Charging process

[0058] 38 Tension

[0059] 40 electricity

[0060] 42 time

[0061] 44 modal coordinates

[0062] S1 to S5 steps of the procedure

Claims

Patent claims Method for estimating a state of health (18) of a battery cell of an electrical energy storage device of an at least partially electrically operated motor vehicle (12) by means of an electronic computing device (10) of the motor vehicle, comprising the steps: - Providing a mathematical model for estimating health status (18); - determining at least one differential voltage curve (34) of the battery cell during a charging process (36) of the electrical energy storage device by means of current filtering, voltage filtering and mathematical derivation; - determining modal coordinates (44) of the at least one differential voltage curve (34); and - Estimating the state of health (18) by evaluating the modal coordinates (44) using the electronic computing device (10). Method according to claim 1, characterized in that the mathematical model is provided based on defective battery cells and / or on the basis of non-defective battery cells historically. Method according to claim 1 or 2, characterized in that, in the mathematical model, differential voltage curves (34) are decomposed into orthogonal modes by means of singular value decomposition. Method according to claim 3, characterized in that a termination sequence for a modal reduction is determined. Method according to claim 4, characterized in that modes are normalized depending on the determined termination sequence. Method according to one of the preceding claims, characterized in that the method is carried out at a state of charge of the battery cell between 40% and 80%. Method according to one of the preceding claims, characterized in that a degradation of the battery cell is determined from the orthogonal eigenmode decomposition and reconstruction of differential voltage curves (34). Method according to one of the preceding claims, characterized in that an offset in a charge axis of a differential voltage curve (34) is eliminated by minimizing the reconstruction error with an orthogonal eigenmode decomposition.Computer program product with program code means which cause an electronic computing device (10) to carry out a method according to one of claims 1 to 8 when the program code means are processed by the electronic computing device (10). Electronic computing device (10) for estimating a state of health (18) of a battery cell of an electrical energy storage device of an at least partially electrically operated motor vehicle (12), wherein the electronic computing device (10) is designed to carry out a method according to one of claims 1 to 8.

Images