Method for detecting fault events in a battery cell arrangement of an electrical energy storage system, and energy storage system designed for this purpose

Abstract

The invention relates to a method for detecting fault events in a battery cell arrangement (20) of an electrical energy storage system (10), wherein the battery cell arrangement (20) is formed from a plurality of electrically interconnected battery cells (22). The method according to the invention comprises the steps: (a) performing electrochemical impedance spectroscopy on the individual battery cells (22) in order to record for each of the battery cells (22) a number n of complex impedance values ​​for a number n different excitation frequencies, wherein n is in the range of 2 to 10; and (b) detecting a fault event if for at least one battery cell (22) n ) in the battery cell arrangement (20) that is for this battery cell (22) n) recorded complex impedance values ​​differ significantly from those complex impedance values ​​obtained for several adjacent battery cells in the battery cell arrangement (20) (22) n-1 , 22 n+1 ) were recorded. Furthermore, the invention proposes an energy storage system (10) suitable for carrying out the method.

Description

[0001] The present invention relates to a method for detecting fault events in a battery cell arrangement of an electrical energy storage system. The invention further relates to a correspondingly designed energy storage system and its use.

[0002] The field of the invention relates to energy storage. This is important, for example, in the generation of electrical energy from renewable energy sources such as photovoltaics, etc., in order to enable the demand-driven provision of energy regardless of when it is generated. Furthermore, energy storage plays a central role, for example, in the automotive sector for the energy supply of an electric drive system of a vehicle such as a battery electric vehicle (BEV) or a hybrid electric vehicle (HEV).

[0003] Particularly in the aforementioned applications, electrical energy storage devices (rechargeable batteries) are used, in which energy is stored in a large number of rechargeable battery cells, e.g., electrochemical battery cells, connected in parallel and / or series (battery cell arrangement). A series connection of battery cells (or of several battery cells connected in parallel) allows the energy storage device to operate with an electrical voltage at one of its electrical terminals that can be many times higher than the voltage of the individual battery cells. Conversely, a parallel connection of battery cells (or of several battery cells connected in series) allows for an increase in the storage capacity with respect to energy or electrical charge, as well as an increase in the load capacity with respect to maximum electrical power.a maximum electrical current during charging or discharging of the energy storage device.

[0004] It is known from the prior art to equip energy storage systems with a battery management system (BMS) to control the operation (charging and discharging) of individual battery cells and / or subunits of the battery cell assembly formed from them, such as, in particular, individually switchable battery cell modules within the battery cell assembly. Furthermore, the BMS enables the monitoring of the aforementioned components of the battery cell assembly or the battery cell assembly as a whole (e.g., recording charge levels, temperatures, etc.) and thus, if necessary, also protecting against overload (e.g., deep discharge protection, overcharge protection, etc.).

[0005] In energy storage systems of the type of interest here, various fault events can occur in the battery cell array. These include, for example, defects in individual battery cells, whether due to aging (e.g., dendrite formation in electrochemical cells) and / or due to thermal or mechanical stress (e.g., in a vehicle accident involving the battery cell array). Such defects can lead to internal short circuits and, consequently, to excessive and often rapid heating of the affected battery cells, especially if the heating triggers an exothermic chemical reaction within the battery cell.

[0006] Excessive heating of a defective battery cell can lead to a safety-critical problem within the battery cell array, known as "thermal runaway." This occurs when excessive heating of one battery cell spreads to neighboring cells, causing the fault to propagate through a chain reaction due to the self-reinforcing process within the battery cell array. Such thermal runaway can lead to the destruction of the battery cell array through fire or explosion, thus posing a significant hazard to the environment containing the energy storage system (e.g., the vehicle and its occupants).

[0007] It is known from the prior art to monitor energy storage systems for the occurrence of fault events in the battery cell arrangement. However, a disadvantage of the known monitoring methods and devices is that they usually detect a fault event relatively late and often do not provide a more precise diagnosis of fault events. Therefore, with the known monitoring methods, it is generally not possible to detect the danger of an impending thermal runaway early enough to warn users of the relevant technical equipment (e.g., drivers / occupants of a vehicle) in a timely manner and / or to potentially prevent the thermal runaway through appropriate countermeasures.One monitoring method known from the prior art involves measuring the hydrogen gas produced and escaping from a battery cell as a result of the exothermic chemical reaction using a gas sensor. However, since such a gas release is only detectable very late, i.e., shortly before the onset of thermal runaway, this method is unsuitable for an "early prediction" of the danger of an impending thermal runaway.

[0008] Exemplary methods and devices are known from DE 10 2009 000 337 A1 and DE 102021 129 351 A1.

[0009] It is therefore an object of the present invention to provide a novel method for enabling simple, reliable, and early detection of fault events in a battery cell arrangement of an electrical energy storage system. Advantageously, such a detection of a fault event can then also classify, for example, whether it is only a defect that is likely to be localized within the battery cell arrangement (e.g., of a single battery cell), or whether the fault event is likely to lead to a "thermal runaway" in which excessive heating, starting from, for example, only a single defective battery cell, spreads through the battery cell arrangement as a result of a chain reaction and destroys further battery cells.

[0010] According to a first aspect of the invention, this problem is solved by a method for detecting fault events in a battery cell arrangement of an electrical energy storage system, wherein the battery cell arrangement is formed from a plurality of electrically interconnected battery cells, and wherein the method comprises the following steps: a) Performing electrochemical impedance spectroscopy on the individual battery cells to record for each of the battery cells a number n of complex impedance values ​​for a number n different excitation frequencies, where n is in the range of 2 to 10, b) Determination of a fault event if, for at least one battery cell in the battery cell arrangement, hereinafter also referred to as the “defective battery cell”, the complex impedance values ​​recorded for that battery cell differ significantly from those complex impedance values ​​recorded for several adjacent battery cells in the battery cell arrangement and c) Classifying a fault event identified in step b), in particular e.g. depending on the extent of the significant deviation, either as a defect of the battery cell in question that is expected to remain locally confined or as a defect of the battery cell in question that is expected to lead to thermal runaway propagating in the battery cell arrangement (at least if no countermeasures are taken).

[0011] The invention advantageously enables simple, reliable and early detection of fault events in a battery cell arrangement, whereby, within the framework of a diagnosis of a detected fault event (e.g. by means of a "classification algorithm"), it is particularly possible to classify whether the fault event associated with the battery cell identified as defective is likely to remain locally limited (in the area of ​​the defective battery cell) or will lead to a "thermal runaway".

[0012] Advantageously, based on the result of such a classification within the scope of the invention, a relatively early prediction of an impending thermal runaway can be made, for example, to generate a corresponding warning for a user of the energy storage system.

[0013] In the event that the energy storage system is used on board a vehicle, for example to supply energy to an electrical system and / or electric drive system of the vehicle, the aforementioned warning may be issued to a user of the vehicle, such as the driver of a battery electric vehicle (BEV) or a hybrid electric vehicle (HEV).

[0014] However, based on the result of the aforementioned classification, it is also advantageous to react, for example, with measures that are particularly well adapted to the situation (countermeasures to avoid thermal runaway) in order to ensure the availability or operational capability of the energy storage system is maintained for as long as possible while reliably preventing damage to property and persons.

[0015] In a preferred embodiment, for the decision to be made in step b) (regarding the question of "significant deviation"), average values ​​are first calculated from the complex impedance values ​​recorded for the several adjacent battery cells for each of the excitation frequencies (e.g., a complex average, or, for example, a real-part average and an imaginary-part average), in order to then make the decision based on these average values. In other words, in this embodiment, the fault event is detected if, for at least one battery cell, the complex impedance values ​​recorded for that battery cell deviate significantly from the complex impedance values ​​averaged over the adjacent battery cells.

[0016] However, for the decision to be made in step b) (regarding the question of the “significant deviation”), the complex impedances recorded for the neighboring battery cells for a specific excitation frequency could also be compared individually with the complex impedance value recorded for the battery cell under consideration for the same frequency.

[0017] In one embodiment of the invention, it is provided that the existence of a “significant deviation” between the relevant complex impedance values ​​is assumed if, for at least one of the excitation frequencies used in electrochemical impedance spectroscopy, the complex impedance values ​​recorded for that excitation frequency differ significantly from each other.

[0018] In this embodiment, a criterion can be defined for each of the excitation frequencies used, the fulfillment of which is considered equivalent to the existence of a significant difference between the relevant complex impedance values. In one embodiment, the existence of a significant difference at one (single) of the excitation frequencies used is sufficient to detect the fault event. In another embodiment, it can be provided, for example, that a significant difference must exist for at least two excitation frequencies used in order to detect the fault event.

[0019] In one embodiment of step b), a criterion to be met for determining a "significant deviation" takes into account all differences between the relevant complex impedance values ​​recorded, i.e., the impedance value differences recorded for all excitation frequencies between, on the one hand, the battery cell under consideration and, on the other hand, the neighboring battery cells (or, for example, the aforementioned mean values ​​of the neighboring battery cells). In this case, the criterion can also be more complex, for example, by determining a significant deviation when a sum, a weighted sum, or some other mathematical combination of the recorded impedance value differences exceeds a certain threshold.

[0020] In one embodiment of the invention, for example, it is provided that the existence of a significant difference between the complex impedance values ​​is assumed if a difference between the real parts and / or a difference between the imaginary parts exceeds a respective predetermined threshold.

[0021] Alternatively or additionally, a criterion for the existence of a significant difference between the complex impedance values ​​can, for example, also require that a difference between the magnitudes and / or a difference between the phase angles (according to a polar representation of the complex impedance values) exceeds a predefined threshold.

[0022] In one embodiment, it is provided that the existence of a significant difference between the complex impedance values ​​is assumed upon fulfillment of a criterion which takes into account the temperature and / or the state of charge of the battery cells in question.

[0023] Here is an example: As mentioned above, a criterion for the existence of a significant difference between the complex impedance values ​​could, for example, require that a difference between similar "real-valued components" of these complex impedance values ​​(e.g., real parts, imaginary parts, magnitudes, phase angles, etc., depending on the representation of the complex impedance values) exceeds a predefined threshold. In this case, the above embodiment could, for example, be implemented by specifying one or more of these thresholds as a function of the temperature and / or the state of charge of the battery cells in question.When the method according to the invention is applied to an energy storage system equipped with a battery management system (BMS), the BMS typically already possesses the functionality to detect such temperatures (of individual battery cells or groups of battery cells) and / or states of charge (of individual battery cells or groups of battery cells). In this case, the BMS's sensor capabilities can advantageously be used to implement the aforementioned embodiment (i.e., to define threshold values ​​depending on the temperature and / or state of charge).

[0024] In one embodiment of step b), instead of, for example, a conventional mathematical calculation, it may be advantageously provided, for example, that the determination of the significant deviation in step b) is carried out using a regression model previously determined empirically. The regression model describes, in a sense, the relationship between, on the one hand, the presence or absence of a battery cell defect in a battery cell under consideration (failure event) and, on the other hand, the acquisition data to be provided to the model as "input data" according to the invention, i.e., the complex impedance values ​​acquired in step a) (where, if necessary, preprocessing, such as averaging the complex impedance values ​​of the neighboring cells as mentioned above, may be provided in order to input, for example, only the mean values ​​for the neighboring cells into the model). In the (e.g.,In the empirical determination of the regression model, as well as in the later use of the regression model, it may be provided that the real-valued components of the complex impedance values, i.e., a real part and an imaginary part, are treated as two separate (real-valued) data in their respective assignment to the relevant excitation frequency.

[0025] In one embodiment of the method, step b) involves determining the significant deviation using a regression model implemented as a trained neural network. In this embodiment (as with a parameterized model), an empirical determination of the regression model can be provided, whereby a regression analysis can be carried out by training the neural network using experimentally recorded training data. The trained neural network then represents a regression model to be used in step b).

[0026] For example, differences between similar real-valued components (e.g., real parts, imaginary parts, magnitudes, or phase angles) can be used for such classification to determine whether a significant difference exists between the complex impedance values ​​being compared. If such a difference is particularly large, e.g., exceeding a predefined "second threshold" for the component in question (which is greater than a "first threshold" for this component used to detect the fault event), the fault event can be classified as a defect likely to lead to thermal runaway propagating throughout the battery cell array, whereas otherwise, the fault event is classified as a defect likely to remain locally confined.

[0027] For such a classification, analogous to a criterion for determining the fault event itself, as mentioned above, it may be provided, for example, that the fault event is classified as a defect likely to lead to thermal runaway propagating in the battery cell arrangement if the complex impedance values ​​recorded for at least one of the excitation frequencies used in electrochemical impedance spectroscopy differ significantly from one another (e.g., exceeding at least one predefined "second threshold"). In another embodiment, it may be provided, for example, that a particularly large difference must exist for at least two of the excitation frequencies (e.g., exceeding at least one predefined "second threshold") in order to determine this classification result.

[0028] Regarding all the aforementioned “thresholds” for certain differences between real-valued components (e.g., between real parts, imaginary parts, etc.), whether within the framework of step b) or step c), it should be noted that one or more of these thresholds may also be specified depending on the sign of the difference in question.

[0029] In one embodiment of step c), instead of, for example, a conventional mathematical calculation, it may be advantageous, for example, that the classification in step c) is carried out using a regression model determined empirically beforehand. The regression model describes, in a sense, the relationship between, on the one hand, the presence or absence of a battery cell defect in the identified battery cell that is likely to lead to thermal runaway, and, on the other hand, the acquisition data to be provided to the model as "input data" according to the invention. With regard to the determination of such a model, as well as its use in step c), the embodiments and details already described above for the model usable in step b) can be provided analogously. In particular, according to one embodiment of the method, for example,In step c), the error event is classified using a regression model implemented as a trained neural network. In this embodiment (as with a parameterized model), an empirical determination of the regression model can be provided, whereby a regression analysis can be realized by training the neural network using experimentally recorded training data.

[0030] In one embodiment, it is provided that in step c) depending on the result of the classification of the fault event, one or more of the following measures are taken: - Control of reduced electrical power of the battery cell arrangement, - Control of increased cooling power to cool the battery cell assembly, and - Shutdown of a battery cell module containing the battery cell identified as defective in step b).

[0031] In one embodiment, the initiation of such measures depends on a previous detection history, whereby, for example, repeated detection of a defect in the same battery cell (by each finding of a significant deviation of the cell impedance from the (possibly averaged) impedance values ​​of the neighboring cells) can be evaluated differently than the first detection of a defect in a cell.

[0032] When the invention is used on board a vehicle (e.g. BEV, HEV), the initiation of countermeasures can be carried out in stages, e.g. as follows: Stage 1: Reduction of drive power (during driving) or charging power depending on the operating situation, as well as intensification of cooling power to the maximum, in order to prevent or stop an exothermic reaction within the cell identified as defective. Furthermore, the driver can be informed about the detection of the anomaly and the initiation of the relevant countermeasure(s). Stage 2: greater reduction of electrical power (during driving) or termination of a charging process. Stage 3: Forced shutdown of the vehicle (while driving) and automatic placement of an emergency call.

[0033] According to another aspect of the present invention, the problem stated at the outset is solved by an electrical energy storage system comprising: - a battery cell arrangement consisting of a large number of electrically interconnected battery cells, and - a detection device designed to detect fault events in the battery cell assembly, the detection device comprising: - an impedance spectroscopy device for performing electrochemical impedance spectroscopy on the individual battery cells in order to record for each of the battery cells a number n complex impedance values ​​for a number n different excitation frequencies, where n is in the range of 2 to 10, - an evaluation device designed to detect an error event if, for at least one battery cell in the battery cell arrangement, the complex impedance values ​​recorded for that battery cell deviate significantly from those complex impedance values ​​recorded for several adjacent battery cells in the battery cell arrangement.

[0034] The embodiments and special configurations described here for the detection method according to the invention can, individually or in any combination, also be provided in an analogous manner as embodiments or special configurations of the energy storage system according to the invention, and vice versa.

[0035] In one embodiment of the energy storage system according to the invention, the detection device further comprises a classification device which is configured to classify a detected fault event depending on the extent of the significant deviation either as a defect of the battery cell in question (identified as defective) that is expected to remain locally limited or as a defect that is expected to lead to thermal runaway.

[0036] Further advantageous embodiments of the detection method or the energy storage system result from the following description of possible configurations of the individual components of the energy storage system. Battery cell arrangement

[0037] The battery cell arrangement can consist of, for example, more than 100, and in particular more than 200, battery cells. A smaller number of battery cells can be electrically connected in parallel and / or series to form individually switchable battery cell modules. Several such (e.g., identically designed) battery cell modules can, in turn, be electrically connected to each other in a controllable manner (e.g., by means of a battery management system) to enable their individual switching on and off within the energy storage system formed by the system. In one embodiment, several (e.g., 2 to 20) battery cell modules are provided, each containing at least 10 (e.g., 10 to 100) battery cells. Battery cells

[0038] The battery cells can be, in particular, electrochemical battery cells. Examples include lithium-ion cells based on lithium transition metal mixed oxides, which can be based on lithium nickel cobalt manganese oxide (NCM), lithium titanate (LTO), lithium ferrophosphate (LFP), or lithium nickel manganese oxide (LNMO). Furthermore, ultracapacitors can also be used as battery cells.

[0039] For the purposes of the invention, any battery cell that is directly adjacent to a particular battery cell in an electrical series connection can be considered "adjacent" to that battery cell. Alternatively or additionally, however, one or more battery cells can also be considered adjacent to a particular battery cell within the scope of the invention, even if they are not direct neighbors of the battery cell in question in its electrical series connection, but are arranged at a relatively short distance from that battery cell due to their spatial arrangement. This includes, in particular, "nearest neighbors" of the battery cell in question in a geometric (e.g., square or hexagonal) grid arrangement of the battery cells. Furthermore, it can be provided that battery cells which are, for example,The "next-next neighbors" (and possibly also, for example, the "next-next-neighbors") of the battery cell in question are represented. In one embodiment of the invention, at least 2, in particular at least 4, neighboring battery cells are taken into account in step b). On the other hand, it is usually advantageous to take into account at most 20, in particular at most 10, neighboring battery cells in this step. Detection device

[0040] The detection device serves to detect fault events in the battery cell arrangement and for this purpose has an impedance spectroscopy device to perform electrochemical impedance spectroscopy on the individual battery cells during the operation of the energy storage system.

[0041] The method of electrochemical impedance spectroscopy (EIS) is known in the art for characterizing electrochemical systems and has previously been used on electric batteries, for example, to determine the internal resistance in the form of a frequency-dependent complex impedance, such as to assess age-related degradation phenomena. It has been found that, within the scope of the invention, the results (complex impedance values) of impedance spectroscopy performed at the cell level during operation of the energy storage system, i.e., on the individual battery cells, can be evaluated in order to detect and, if necessary, classify fault events in individual battery cells at an early stage. In the specific design of the electrochemical impedance spectroscopy, or rather...For the realization of individual components, it is advantageous within the scope of the invention to make use of established methods and details in this field.

[0042] In the invention, battery cells can be subjected to a generated excitation signal (e.g., sinusoidal or rectangular) using the impedance spectroscopy device, and the resulting "response signal" (e.g., voltage signal) of the cell under investigation can be measured. In particular, a resulting phase shift between the excitation and response signal, which depends on the state of the cell in question, can then be investigated, e.g., using a Fourier analysis (e.g., FFT), and thus, e.g., the real and imaginary parts of the cell's intrinsic impedance can be determined (calculated) for each excitation frequency. For each battery cell, a number n of complex impedance values ​​are recorded for a number n different excitation frequencies, where n is in the range of 2 to 10. In one embodiment, n is in the range of 3 to 6.

[0043] In electrochemical impedance spectroscopy, the invention allows for individual acquisition processes, each performed or initiated at a specific time, in which the respective n complex impedance values ​​are acquired for a plurality of battery cells, preferably all battery cells of the battery cell arrangement. Due to the limitation of the number n of different excitation frequencies and thus complex impedance values ​​to a maximum of 10, and in particular, for example, a maximum of 6, these acquisition processes can be carried out very quickly. In one embodiment, for example, the duration of each such acquisition process is in the range of 0.1 to 2 s.

[0044] The impedance spectroscopy system can be configured to successively adjust the excitation signal frequency to the various excitation frequencies during each acquisition process, in order to record the complex impedance values ​​sequentially based on an evaluation of the respective response signal. Alternatively, it can also be configured to use a single excitation signal containing several or all of the different excitation frequencies during the acquisition process and to simultaneously record the complex impedance values ​​for these multiple excitation frequencies based on an evaluation of the response signal resulting during such a time phase.

[0045] A further advantage of the aforementioned limitation of the number n is that it simplifies the evaluation of the recorded complex impedance values ​​for the purpose of detecting and, if necessary, classifying a detected fault event. Such an evaluation can therefore be carried out, for example, in a resource-efficient yet rapid manner using a dedicated, program-controlled computer system (e.g., on board a vehicle).

[0046] Such rapid evaluation advantageously allows for relatively rapid succession of such acquisition processes on each battery cell during electrochemical impedance spectroscopy performed during operation of the energy storage system, e.g. at predetermined times with mutual intervals in a range of 10 s to 1 min.

[0047] In this context, it has been advantageously shown that the relatively small number of complex impedance values ​​(and corresponding excitation frequencies) is sufficient in practice to enable sufficiently reliable fault event detection. This is particularly true if the excitation frequencies are chosen appropriately (adapted to the specific application). In many cases, for example, it is advantageous if all excitation frequencies lie within a range whose lower limit is at least 1 Hz, in particular at least 10 Hz, and / or whose upper limit is at most 1 kHz, in particular at most 0.5 kHz. Furthermore, in many cases it is advantageous if the excitation frequencies are at least approximately equidistantly distributed on a logarithmic scale.

[0048] Regarding the excitation of the battery cells within the framework of electrochemical impedance spectroscopy, the excitation signal can, for example, advantageously be applied to a group of several battery cells interconnected in the battery cell arrangement, such as a battery cell module of the aforementioned type, or a so-called "cell cluster." Alternatively, the excitation signal can also be applied across the entire battery cell arrangement, i.e., between a positive and a negative pole of the overall arrangement (battery cell arrangement), for example, at an energy storage terminal of the energy storage system. To implement the latter embodiment in particular, it is possible, for example, to equip the impedance spectroscopy device with a sufficiently powerful driver (amplifier) ​​to provide the excitation signal, which in this case requires a higher power output.For performing impedance spectroscopy "on the individual battery cells," it is therefore by no means absolutely necessary to apply a cell-specific excitation signal directly to each battery cell (although this is also possible within the scope of the invention). Rather, it is sufficient to tap off the "response signal" individually (at the individual battery cells). The latter can advantageously be implemented using so-called cell supervising circuits (CSCs).

[0049] The detection device also includes an evaluation unit to identify fault events using the aforementioned evaluation of the recorded complex impedance values. For this purpose, a "detection algorithm" running on a program-controlled computer (e.g., on board a vehicle) can be provided to determine whether the complex impedance values ​​recorded for at least one of the battery cells deviate significantly from those complex impedance values ​​(or their mean values, if applicable) recorded for several adjacent battery cells in the battery cell array. The detection algorithm can, in particular, also employ a regression model, e.g., in the form of a parameterized mathematical model or, e.g., in the form of a model implemented as a trained neural network.It may also be advantageous to provide a regression model that is used jointly for steps b) and c) of the detection method according to the invention, which in this case combines the functionality of determining the error event (based on the determination of the “significant deviation”) and the functionality of classification.

[0050] In one embodiment, the detection device further includes a voltage measuring device for measuring the individual cell voltages of the battery cells. By evaluating the results of the cell voltage measurements, the plausibility of a fault event determined based on the impedance spectroscopy results can be verified. Alternatively or additionally, the cell voltage measurements can be used to determine the state of charge of the battery cells and, based on this, to dynamically adjust a criterion that verifies the presence of a "significant deviation" between complex impedance values ​​when detecting fault events (as already mentioned, such a criterion can, for example, take into account the state of charge and / or the temperature of the battery cells in question).When using a regression model (as described above) to determine the fault event and / or to classify it, measured cell voltages (of the battery cell under consideration and of the neighboring battery cells) can advantageously also be provided as input variables for the regression model.

[0051] The detection device may further include a classification device that serves to classify a detected fault event either as a defect expected to remain locally confined or as a defect expected to lead to thermal runaway propagating throughout the battery cell assembly. This classification may, in particular, be based on, for example, the (quantitatively determined) magnitude of the significant deviation between the complex impedance values ​​that led to the detection of the fault event.

[0052] If a voltage measuring device is available for measuring the individual cell voltages, the result of a classification of a fault event based on impedance spectroscopy can be validated by evaluating the measured cell voltages. Alternatively or additionally, the cell voltage measurement can be used for dynamically adjusting a criterion by which the classification is implemented. When using a regression model (as described above) to classify a detected fault event, the measured cell voltages (of the battery cell identified as defective and of adjacent battery cells) can advantageously also be used as input variables for the regression model.

[0053] Preferably, the detection device is implemented at least partially as a program-controlled computer device or as a functional component (software and / or hardware) of such a program-controlled computer device (e.g., battery management system of the energy storage system in question).

[0054] According to another aspect of the invention, the use of a method of the type described herein for detecting fault events in a battery cell arrangement of an electrical energy storage system for supplying energy to a vehicle is proposed.

[0055] According to another aspect of the invention, the use of an energy storage system of the type described herein for supplying energy to a vehicle is proposed.

[0056] According to another aspect of the invention, a vehicle is proposed which is equipped with an energy storage system of the type described herein.

[0057] The vehicle may in particular be a vehicle equipped with an electric drive system such as a battery electric vehicle (BEV) or hybrid electric vehicle (HEV), whereby the energy storage system is intended for the electrical supply of this drive system.

[0058] The invention is further described below with reference to exemplary embodiments and the accompanying drawings. These schematically depict: Fig. 1 a block diagram of an electrical energy storage system according to an exemplary embodiment, Fig. 2, Fig. 3, Fig. 4 to Fig. 5 exemplary time courses of some quantities measurable in an energy storage system during the occurrence of a fault event classifiable as thermal runaway, and Fig. 6 A flowchart of a procedure for detecting fault events carried out by means of a battery management system of an energy storage system according to an exemplary embodiment.

[0059] Fig. Figure 1 shows an electrical energy storage system 10 for supplying an electrical on-board network of a vehicle equipped with an electric drive (e.g. BEV, HEV).

[0060] The energy storage system 10 comprises a battery cell arrangement 20, which is formed from a plurality of electrically interconnected battery cells 22, and a battery management system (BMS) 30 for controlling and monitoring the battery cell arrangement 20.

[0061] In this example, the battery cells 22 are designed as electrochemical battery cells (e.g., lithium-ion cells).

[0062] For charging and discharging the electrical energy storage formed by the battery cell arrangement 20, the energy storage system 10 is equipped with a two-pole energy storage connection 40.

[0063] In the illustrated example, the battery cell arrangement 20 is divided into a plurality of "battery cell modules" 24, each containing a portion of the battery cells 22, which can be individually switched on and off by means of the battery management system 30 during operation of the energy storage system 10. For example, the illustrated battery cell arrangement 20 could be formed from 4 such battery cell modules 24, in each of which 100 battery cells 22 of the same battery cell type are connected in series.

[0064] The battery management system 30, implemented by means of a program-controlled computer device, is located as in Fig. 1 is connected to the battery cell arrangement 20 or its modules 24 via a connection (data interface) 31, symbolized by a communication link, in order to monitor the operation (charging and discharging) of the energy storage system in a program-controlled manner (e.g., charging / discharging current measurement, etc.) and, in the example, can communicate operating parameters recorded in the area of ​​the energy storage system 10, such as state of charge, charging / discharging current, temperature, etc. (e.g., also individually for the individual modules 24), to an external device (e.g., a central control unit in a vehicle) via a connection (data interface) symbolized by 32.

[0065] The battery management system 30 protects the components of the battery cell assembly 20 from overload by preventing further charging (e.g., for overcharge protection) or further discharging (e.g., for deep discharge protection) as needed (e.g., also for individual modules 24). In this example, the battery management system 30 can activate or deactivate individual battery cell modules 24 as needed (e.g., due to a malfunction). To perform the monitoring tasks of the battery management system 30, the battery cell assembly 20 is equipped with sensors (not shown in the figure) to acquire a range of operating parameters (e.g., measured by sensors) from the battery cell assembly 20 and to supply corresponding signals or data to the battery management system 30.From a purely functional point of view, the components provided in the battery cell arrangement 20 and serving this purpose could also be considered as part of the battery management system 30.

[0066] In the example shown, the following operating parameters from the area of ​​the battery cell arrangement 20 are recorded, for example: - State of charge (e.g. SOC) of the battery cell arrangement 20 and / or of the individual modules 24 and / or of the individual battery cells 22, - a temperature, or preferably several temperatures at correspondingly several different respective measuring points, within the battery cell arrangement 20, - respective cell voltages at all battery cells 22.

[0067] A special feature of the energy storage system 10 is that it has a detection device 33, by means of which fault events in the battery cell arrangement 20 can be detected and classified at an early stage.

[0068] The detection device 33 has an impedance spectroscopy device 34 for this purpose, by means of which an electrochemical impedance spectroscopy is carried out on the individual battery cells 22 during the operation of the energy storage system 10, in order to detect a number of n complex impedance values ​​for a number of n different excitation frequencies for each of the battery cells 22 by means of temporally (e.g. periodically) successive detection processes, where n is in the range of 2 to 10.

[0069] For this purpose, an excitation signal in the form of an alternating voltage or an alternating current is generated for each of the battery cells 22 and applied between the anode terminal and the cathode terminal of the respective cell 22 (i.e., "superimposed" on the cell voltage). In the example shown, the impedance spectroscopy device 34 communicates with (in) the connection (data interface) 31 via the connection (in Fig. 1 cell supervising circuits (CSCs) within the battery cell arrangement 20 (not shown) which, in addition to, for example, the aforementioned cell voltage measurement, also have the functionality to generate and apply the excitation signal to the cells 22. Frequency generators integrated into the cell supervising circuits can be used for this excitation to initiate individual acquisition processes when appropriately controlled by the impedance spectroscopy device 34 (via interface 31). In each of these processes, for example, n different excitation signals (with different frequencies) are generated sequentially and applied to the battery cells 22. The individual excitation signals can be, for example, sinusoidal or rectangular.As an alternative to applying the excitation signal using the cell monitoring units, it may also be possible to generate an excitation signal common to several or, in particular, all of the battery cells 22 and to apply it between a positive and a negative terminal of the battery cell arrangement (e.g., at terminal 40). The impedance spectroscopy device 34 may include a sufficiently powerful driver for generating the excitation signal.

[0070] Simultaneously with the excitation, a measurement signal (response signal) resulting from the excitation is recorded for each individual cell, which in the example is tapped off as a voltage signal between the respective anode and cathode terminals of the battery cells 22 by means of the (not shown) cell monitoring units.

[0071] A resulting phase shift between the excitation and response signals is analyzed separately for each battery cell 22, e.g., using a Fourier analysis performed in the cell monitoring units, and the real and imaginary parts of the battery cell's intrinsic impedance are determined (e.g., calculated) for each of the n excitation frequencies. In this example, the result of this impedance determination is communicated from the cell monitoring units to the impedance spectroscopy device 34 via interface 31.

[0072] It should be noted that electrical energy storage systems of the type of interest within the scope of the invention often already have a cell monitoring device designed using such cell monitoring units, by means of which operating states of individual battery cells and / or at least respective battery cell clusters are monitored (e.g. by tapping and evaluating and / or communicating the cell voltage(s)), so that to realize the present invention, this cell monitoring device or its cell monitoring units only need to be equipped with means for applying the excitation signal with the required excitation frequencies and for recording the response signals or complex impedance values ​​resulting from these excitations.

[0073] The n complex impedance values ​​(for the n different excitation frequencies) obtained by the impedance spectroscopy device 34 for each of the battery cells 22 during each acquisition process are evaluated in an evaluation device 35 also implemented in the detection device 33.

[0074] The evaluation device 35 serves to evaluate the complex impedance values ​​or their real-valued impedance components (here, for example, real and imaginary parts) recorded by the impedance spectroscopy device 34 in order to identify any defective battery cell 22 by means of this evaluation of the result of the impedance spectroscopy.

[0075] In the example shown, this evaluation can also take into account other quantities recorded in the area of ​​the battery cell arrangement 20, such as the aforementioned cell voltages of the individual battery cells 22 and / or one or more temperatures measured in the battery cell arrangement 20.

[0076] With regard to the implementation of the battery management system 30 by means of a program-controlled computer device, the impedance spectroscopy device 34 and the evaluation device 35 represent functional components of the battery management system 30, which can therefore each be implemented at least partially by corresponding parts of software or a control program running in the battery management system 30.

[0077] The same applies to one in Fig. 1 Classification device 36 shown as a functional component of the detection device 33 or the battery management system 30, which serves to classify a detected fault event as a locally limited defect or as a spreading thermal runaway.

[0078] The result of the evaluation and, if applicable, classification can be advantageously used by the BMS 30 within the framework of controlling the operation of the battery cell arrangement 20 (e.g., also to initiate countermeasures after a fault event has been detected). Alternatively or additionally, the result can be communicated further, e.g., via interface 32 (e.g., to a central control unit of the vehicle in question).

[0079] With regard to the detection and subsequent classification of error events in the Fig. The energy storage system 10 shown in section 1 is described below with reference to the Fig. 2, Fig. 3, Fig. 4 to Fig. 5. First, a possible error event scenario is explained.

[0080] In this scenario, it is assumed that a defect in one battery cell, initially only local, develops into 22 n (cf.) Fig. 1) In the battery cell arrangement 20, a spreading thermal runaway develops after a certain period of time if the energy storage system 10 continues to operate unchanged during this period. Fig. 2, Fig. 3, Fig. 4 to Fig. Figure 5 shows exemplary time courses of some quantities measurable in the energy storage system 10.

[0081] Fig. Figure 2 shows the temperature T at the center of the battery cell arrangement 20 as a function of time t. In this example, it is assumed that the initially only local defect of battery cell 22 n occurs at time t = 0 and the battery cell 22 nIt begins to warm up approximately 100 seconds later (t = 100 s). This warming causes a phenomenon in cell 22 some time later. n an exothermic chemical reaction was triggered, which led to the destruction of the cell 22 n This leads to further problems. Furthermore, within a period of approximately 100-200 seconds, the excessive heating of cell 22 spreads. n also on adjacent battery cells 22 n-1 , 22 n+1 , so that the fault event spreads in a chain reaction through the battery cell arrangement 20. From t = 200 s, the defect in cell 22 becomes apparent. n (from t = 0) finally resulting increase in the temperature T of the entire battery cell arrangement 20.

[0082] Fig. Figure 3 shows an example of a corresponding time course of a cell voltage U, specifically for cell 22, which is initially to be assessed as defective (approximately at t = 100 s). n , as well as (for example at t = 200 s) for their neighboring cells 22 n-1 , 22n+1 , and finally (at approximately t = 250 s) for neighboring cells 22 n-2 , 22 n+2 of the cells 22 n-1 , 22 n+1 The cell voltage U of cell 22 n will be in Fig. 3 as “U n “ denoted, and the cell voltages U of the neighboring cells 22 n-1 , 22 n+1 accordingly as "U n-1 “ or “U n+1 “etc. In Fig. 3. One can recognize the clear and rapid drop in cell voltage U associated with each occurrence of a defect. Fig. Figure 3 also shows an average value M of the instantaneous cell voltages U of all other (still intact) cells 22 of the battery cell arrangement 20.

[0083] Fig. Figure 4 shows an example of a corresponding time course of the H2 concentration C. H2 in battery cell arrangement 20. Here, an increase in the concentration C associated with the cell defects can be seen. H2 approximately at t = 100 s, t = 200 s and t = 250 s.

[0084] Fig. Figure 5 shows, by way of example, the corresponding time course of a difference DIFF between the imaginary parts of complex impedance values ​​of the battery cell 22. n and on the other hand the neighboring cells 22 n-2 , 22 n+2 for a specific of the various excitation frequencies used in electrochemical impedance spectroscopy. In Fig. 5. It can be seen that the difference DIFF already exists shortly after t = 0 (i.e., the beginning of the defect in cell 22). n ) begins to rise and then successively exceeds threshold values ​​TH1 and TH2. The respective times at which the threshold values ​​TH1 and TH2 are exceeded are shown in Fig. 5 is labelled with t1 and t2 respectively.

[0085] As part of the procedure for detecting such error events, the evaluation unit 35 is used ( Fig. 1) the evaluation of the complex impedance values ​​recorded by means of the impedance spectroscopy device 34 in order to detect any defect in one of the battery cells 22 at an early stage.

[0086] The evaluation unit 35 detects an error event if the complex impedance values ​​recorded for a battery cell 22 in the battery cell arrangement 20 deviate significantly from the complex impedance values ​​recorded for neighboring battery cells 22.

[0087] In this example, it is assumed that for the determination of such a significant deviation, it suffices if, for at least one of the various excitation frequencies, the complex impedance values ​​recorded for that excitation frequency differ significantly from one another, on the one hand for the battery cell under consideration and on the other hand for the neighboring battery cells. The existence of a significant difference between the complex impedance values ​​is assumed here if the differences between the real parts and / or imaginary parts calculated by the evaluation unit 35 fulfill a criterion (error case criterion) specified by an evaluation algorithm. For example, it may be stipulated that the significant difference is assumed if the difference between the real parts and / or the difference between the imaginary parts exceeds at least one (alternatively: several) of several (e.g.,exceeds the threshold values ​​specified individually for each of the excitation frequencies.

[0088] Thus, in the scenario according to the example of the Fig. 2, Fig. 3, Fig. 4 to Fig. 5 the fault event (defect of battery cell 22) n ) can be detected very early, namely when the difference DIFF exceeds the threshold TH1 at time t1. The threshold TH1 is an example of various thresholds (defined individually for each of the excitation frequencies) according to a fault case criterion, where each exceedance of such a threshold triggers the detection of the fault event. Fig. In section 5, the indicated threshold TH1 refers to a specific excitation frequency and to the difference DIFF between the imaginary parts of the compared complex impedance values. Fig. Although not shown in section 5 for clarity, the evaluation algorithm provides for further threshold values, each relating to a different excitation frequency and / or a difference in the real parts of the compared complex impedance values. Therefore, for a number "n" of excitation frequencies, the algorithm can provide a number of "2 x n" threshold values ​​(e.g., with 4 excitation frequencies: 4 threshold values ​​relating to real part differences and 4 threshold values ​​relating to imaginary part differences).

[0089] If, in this manner, i.e., due to a significant deviation between recorded complex impedance values, an error event is detected by the evaluation unit 35 after a recording process of electrochemical impedance spectroscopy, the evaluation unit 35 classifies this error event. - either as a defect of the battery cell in question that is expected to remain localized (22 n in Fig. 1) - or as a defect of the battery cell in question (22 n in Fig. 1), which is expected to lead to thermal runaway propagating in the battery cell arrangement 20.

[0090] This classification is based on the magnitude of the previously determined significant deviation. In the example, further threshold values ​​("secondary thresholds") are provided according to a classification algorithm.

[0091] In the scenario according to the example of the Fig. 2, Fig. 3, Fig. 4 to Fig. 5 is determined after the fault event (defect of battery cell 22). n) underlying detection process, this error event is classified by the evaluation unit 35 as likely to lead to thermal runaway if (at least) one of the detected differences exceeds not only the assigned and already explained "first threshold" (threshold TH1, etc.) but also a comparatively higher "second threshold". For example, in a detection process during the one described in Fig. At the plotted time t1, this is not yet the case, so that after such an early recording process, the error event is still classified by the evaluation unit 35 as likely to remain locally limited.

[0092] In Fig. Figure 5 shows an example of the second threshold TH2, which refers to the same excitation frequency and the same difference (DIFF) as the threshold TH1 and which is taken into account in the classification.

[0093] During a data collection process, for example, around or after the time in Fig. At the plotted time t2, the evaluation unit 35 will determine that the difference DIFF has now also exceeded the second threshold TH2 and will therefore classify the fault event as likely to lead to a thermal runaway.

[0094] The TH2 threshold (analogous to the explanation for TH1) is an example of various second thresholds (defined individually for each excitation frequency) according to a classification criterion, where each exceedance of such a second threshold triggers the classification of the fault event as likely to lead to thermal runaway. Fig. In section 5, the indicated threshold TH2 (as well as TH1) refers to a specific excitation frequency and the difference DIFF of the imaginary parts of the compared complex impedance values. Fig. For the sake of clarity, further such "second thresholds" are omitted. The classification algorithm used provides for additional second thresholds, each relating to a different excitation frequency and / or a difference in the real parts of the compared complex impedance values. Therefore, for a number "n" of excitation frequencies, the algorithm can provide for a number "2 x n" of second thresholds (analogous to the explanation above for TH1).

[0095] It should be noted that, although in Fig. 5 not shown, depending on the sign of a relevant difference (e.g. DIFF in the example) different (i.e. “sign-dependent”) thresholds may also be provided.

[0096] As an alternative to using, for example, a parameterized detection or classification algorithm, these algorithms can advantageously also be implemented as a regression model determined in advance (e.g., empirically) or include such a regression model.

[0097] Fig. 6 illustrates, using a flowchart as an example, the steps of an energy storage system 10 of Fig. 1 feasible detection method.

[0098] In step S1, an electrochemical impedance spectroscopy acquisition process is carried out to acquire for each of the battery cells (22) a number n of complex impedance values ​​for a number n different excitation frequencies, where n is in the range of 2 to 10 (e.g. n=5).

[0099] In a subsequent step S2, it is checked whether the complex impedance values ​​recorded for at least one battery cell (22) in the battery cell arrangement (20) deviate significantly from those complex impedance values ​​recorded for several adjacent battery cells in the battery cell arrangement. For this check, for example, impedance values ​​averaged over the adjacent battery cells can be used. The check is performed sequentially or in parallel for all battery cells of the battery cell arrangement in question.

[0100] If this is not the case, the processing proceeds back to step S1 to initiate the next data collection process.

[0101] However, if this is the case, the presence of an error event is detected (assumed) in step S2 and the processing proceeds to step S3.

[0102] In step S3, the complex impedance values ​​recorded in step S1 are further evaluated in order to determine the extent of the significant deviation identified in step S2 according to a predefined criterion and to classify the error event depending on this extent of the significant deviation.

[0103] If the error event in step S3 is likely to remain locally limited to a defect in an affected battery cell (e.g. 22 n Once classified, the processing proceeds to step S4.

[0104] In step S4, countermeasures are initiated (e.g., forcibly reducing the electrical power of the battery cell array, etc.), which may also depend on a previous detection history (for example, accessing stored data regarding previous detections of fault events may be considered). The processing then proceeds back to step S1.

[0105] However, if in step S3 the error event is a defect of the battery cell in question (22 n If the processing has been classified as a thermal runaway that is likely to spread throughout the battery cell assembly, the process proceeds to step S5.

[0106] In step S5, more extensive countermeasures are initiated (e.g., forced shutdown of the energy storage system).

[0107] To implement steps S2 and S3, it may be possible to use a regression model determined empirically beforehand, into which the complex impedance values ​​recorded in step S1 (possibly after preprocessing such as averaging the complex impedance values ​​of neighboring cells) are fed as "input data" and which provides as "output data" the decision on the presence of an error event and, if applicable, its classification result. Reference symbol list 10 Energy storage system 20 battery cell arrangement 22 battery cells 22 n Battery cell identified as defective 22 n-1 adjacent battery cell 22 n+1 adjacent battery cell 24 battery cell modules 30 Battery Management System 31 Connection (data interface) 32 Connection (data interface) 33 Detection device 34 Impedance spectroscopy device 35 Evaluation unit 36 Classification institution 40 Energy storage connection t time Temperature (in battery cell arrangement) Cell voltages M Mean (of cell voltages) C H2 H2 concentration (in battery cell arrangement) DIFF Difference between imaginary parts TH1, TH2 thresholds

Claims

[1] Method for detecting fault events in a battery cell arrangement (20) of an electrical energy storage system (10), wherein the battery cell arrangement (20) is formed from a plurality of electrically interconnected battery cells (22), the method comprising the following steps: a) Performing electrochemical impedance spectroscopy on the individual battery cells (22) to record for each of the battery cells (22) a number of n complex impedance values ​​for a number of n different excitation frequencies, where n is in the range of 2 to 10, b) Detection of a fault event if for at least one battery cell (22 n ) in the battery cell arrangement (20) that is for this battery cell (22) n ) recorded complex impedance values ​​differ significantly from those complex impedance values ​​obtained for several adjacent battery cells in the battery cell arrangement (20) (22)n-1 , 22 n+1 ) were recorded, and c) Classifying a fault event identified in step b) depending on the extent of the significant deviation either as a defect of the battery cell in question that is expected to remain locally confined (22 n ) or as a defect in the battery cell concerned (22 n ), which is expected to lead to thermal runaway propagating in the battery cell arrangement (20). [2] Method according to claim 1, wherein in step b) the presence of the significant deviation is assumed if for at least one of the n excitation frequencies the complex impedance values ​​recorded for that excitation frequency differ significantly from each other. [3] Method according to claim 2, wherein the existence of a significant difference between the complex impedance values ​​is assumed if a difference between the real parts and / or a difference between the imaginary parts exceeds a respective predetermined threshold. [4] Method according to one of claims 2 or 3, wherein the existence of the significant difference between the complex impedance values ​​is assumed upon fulfillment of a criterion which is the temperature and / or the state of charge of the battery cells concerned (22 n-1 , 22 n , 22 n+1 ) taken into account. [5] Method according to any of the preceding claims, wherein in step c) depending on the result of the classification of the fault event, one or more of the following measures are taken: - Control of a reduced electrical power of the battery cell arrangement (22), - Control of increased cooling power to cool the battery cell assembly (22), and - Shutdown of a battery cell module (24) which replaces the battery cell (22) identified as defective in step b). n ) contains. [6] Electrical energy storage system (10), comprising: - a battery cell arrangement (20) formed from a plurality of electrically interconnected battery cells (22), and - a detection device (33) configured to detect fault events in the battery cell arrangement (20), the detection device (33) comprising: - an impedance spectroscopy device (34) for performing electrochemical impedance spectroscopy on the individual battery cells (22) in order to record for each of the battery cells (22) a number n complex impedance values ​​for a number n different excitation frequencies, where n is in the range of 2 to 10, - an evaluation unit (35) designed to detect a fault event if at least one battery cell (22) n ) in the battery cell arrangement (20) that is for this battery cell (22) n ) recorded complex impedance values ​​differ significantly from those complex impedance values ​​obtained for several adjacent battery cells in the battery cell arrangement (20) (22) n-1 , 22 n+1 ) were detected, and wherein the detection device (33) optionally further comprises: - a classification device (36) designed to classify a detected fault event depending on the extent of the significant deviation either as a defect of the battery cell in question that is expected to remain localized (22 n ) or as a defect of the battery cell in question (22 n ), which is expected to lead to thermal runaway propagating in the battery cell arrangement (20). [7] Use of a method according to any one of claims 1 to 5 for detecting fault events in a battery cell arrangement (20) of an electrical energy storage system (10) for supplying energy to a vehicle.

Images