Method for monitoring battery cells of battery of motor vehicle, computer program, data processing device, and motor vehicle

Abstract

The invention relates to a method (100) for monitoring battery cells (215) of a battery (210) of a motor vehicle (200), said method (100) comprising: detecting (110) a time sequence (260) of cell state variables (265) characterizing the battery cells (215); determining (120) a state vector (270) by inputting the time sequence (260) into the first neural network (280); determining (130) an impedance spectrum (275) of the battery cells (215) by means of impedance spectroscopy; and determining (140) state information (290) for monitoring the battery cells (215) by inputting the impedance spectrum (275) and the state vector (270) into a second neural network (285).

Description

Technical Field

[0001] The present disclosure relates to a method for monitoring battery cells of a battery of a motor vehicle and a data processing device designed to at least partially implement the method. In addition, a motor vehicle with a data processing device is provided. Additionally or alternatively, a computer program is provided, which includes instructions that, when the program is implemented by a computer, cause the computer to at least partially implement the method. Additionally or alternatively, a computer-readable medium is provided, which includes instructions that, when the instructions are implemented by a computer, cause the computer to at least partially implement the method.

[0002] The present disclosure relates to a rechargeable battery or storage battery, which can be used, for example, as a traction battery for an electrically driven motor vehicle. Such a battery usually has a plurality of interconnected battery cells. These battery cells should be monitored during their operation. For this purpose, it is known, for example, to monitor the cell voltage and temperature of the battery cells by means of voltage sensors and temperature sensors on the battery cells. However, it is also desirable to monitor the dynamic behavior of the battery cells, which describes, for example, the aging of the battery cells. When developing battery cells, batteries and electric vehicles, the state of the battery chemistry is crucial for operating the motor vehicle, in particular for calculating the remaining range of the motor vehicle. Background Art

[0003] In particular, in battery cells based on lithium iron phosphate (LiFePo, LFP), it is particularly difficult to calculate the state of charge (SOC) and / or the state of health (SOH). Typically, the SOC and / or SOH are determined by measuring the open circuit voltage (OCV). Here, the voltage is measured for each individual battery cell, and the state of charge corresponding to the measured voltage is determined in a lookup table including the association between the voltage and the state of charge. However, the OCV curve of the LFP battery cell is relatively flat, which makes it only difficult and / or not so reliable to determine the SOC. For example, in the SOC range of 30% to 55%, a voltage difference of about 8mV can occur, but in the SOC range of 65% to 95%, only a voltage difference of about 2mV can occur. The known algorithm for determining the SOC based on the development of battery cells based on lithium nickel manganese cobalt oxide (NMC) is inaccurate. Therefore, a relatively large buffer (Puffer) is required to maintain energy in order to avoid battery discharge and vehicle stagnation in order to operate the motor vehicle.

[0004] In addition to the OCV curve, so-called energy integrators are used, which integrate the current over time, for example to add up the consumed energy. Thus, the SOC can be determined reliably, but at a high cost. Current integrators are used in motor vehicles, but these have a lower accuracy. To determine the SOH, current pulse measurement methods are used in the laboratory and / or on a test bench.

[0005] To improve the prediction and calculation of SOC and / or SOH, for example in a vehicle, electrochemical impedance spectroscopy can be performed. This measurement method yields additional cell information that exhibits a higher sensitivity compared to battery state indicators such as SOC and SOH.

[0006] For this purpose, it is known from the prior art to monitor the battery cells by means of impedance spectroscopy. In constant current impedance spectroscopy, a current signal with currents of different frequencies is impressed as an excitation signal into the battery cells, and a frequency-dependent cell voltage signal of the battery cells is detected as a response signal. The impedance spectrum of the battery cells is determined from the correlation between the current signal and the cell voltage signal.

[0007] DE 10 2019 125 014 A1 discloses a method for monitoring a battery cell of a battery of a motor vehicle by means of a constant current impedance spectroscopy method. In this method, a current signal with currents of different frequencies is applied to the battery cell as an excitation signal. In addition, a cell voltage signal of the battery cell is detected as a frequency-dependent response signal to the excitation signal, and the impedance spectrum of the battery cell is determined based on the excitation signal and the response signal.

[0008] However, in some cases, impedance spectroscopy cannot be performed and / or is not very reliable. For example, it is difficult to perform impedance measurements while the motor vehicle is moving, because the DC current applied to the battery cells should remain constant in order to be able to perform impedance measurements reliably.

[0009] It is also known that measurable characteristics of a battery cell (e.g., OCV curve and / or impedance spectrum) depend on the previous history of the battery cell, i.e., have hysteresis. That is, if the voltage or impedance measurement yields a specific value, then different SOC and SOH may have to be assigned to the measurement. The assignment of SOC and SOH depends on the measurement method of the battery cell. Here, for example, the temporal development of the current direction and / or the current intensity are of great significance, as described in "Characterization of charge and discharge behavior of lithium ion batteries with olivine based cathode active material" (ROSCHER, Michael A; VETTER, Jens; SAUER, Dirk Uwe; "Journal of Power source", 2009, Vol. 191, No. 2, pp. 582-590). Summary of the invention

[0010] In the context of the prior art, the object of the present disclosure is to provide an improved method for monitoring a battery cell of a battery of a motor vehicle, which method is suitable for enriching the prior art. A specific design of the present disclosure can solve the task of enabling more reliable and more comprehensive monitoring of battery cells.

[0011] This object is achieved by the features of the independent claims. The contents of the parallel claims and, for example, the dependent claims are optional aspects and / or developments of the disclosure.

[0012] The object is thus achieved by a method for monitoring a battery cell of a battery of a motor vehicle. The method comprises: detecting a time series of a cell state variable characterizing the battery cell; determining a state vector by inputting the time series into a first neural network; determining an impedance spectrum of the battery cell by means of impedance spectroscopy; and determining state information for monitoring the battery cell by inputting the impedance spectrum and the state vector into a second neural network.

[0013] A motor vehicle has a battery. The battery includes a plurality of battery cells to be monitored. The state of each battery cell can be characterized by a cell state parameter. Here, the cell state parameter can include, in particular, information characterizing the current state of the battery cell and / or the current effective power of the battery cell. The cell state parameter can be measured by measurements at the battery cell. By repeatedly measuring the cell state parameter, a time series of the cell state parameter, i.e., the temporal sequence of the measured values ​​of the cell state parameter, can be determined. The time series of the cell state parameter can give a description of the time dependency of the cell state parameter. For this purpose, the time series of each cell state parameter can include a timestamp, which indicates the time when the cell state parameter was measured. The history or the past of the battery cell, or the state of the battery cell, can be characterized by the time series. In order to determine the time series, in particular during driving or during operation of the motor vehicle, the cell state parameter can be measured.

[0014] The time series of the cell state variables is input into the first neural network. The first neural network is configured to determine a state vector from the time series of the cell state variables. A state vector is a one-dimensional or multi-dimensional vector which includes one or more values ​​that indicate the state of the battery cell and the development of the battery cell over time. The state vector does not have to have information that is readable and / or understandable to a user, since the state vector is configured to be input into a second neural network for further processing.

[0015] The impedance spectrum of a battery cell is determined by impedance spectroscopy. For example, impedance spectroscopy is performed as disclosed in DE 10 2019 125 014 A1. Here, a current signal with currents of different frequencies is applied to the battery cell as an excitation signal. In addition, the cell voltage signal of the battery cell is detected as a frequency-dependent response signal to the excitation signal, and the impedance spectrum of the battery cell is determined based on the excitation signal and the response signal. The impedance spectrum depends on the state of the battery cell, i.e. the state of charge and the aging state, as well as the history or past of the battery cell.

[0016] The state information is used to monitor the battery cells and may include information that can be read by a user and / or become information that can be read by a user through further processing. For example, the state information may include a state of charge and / or an aging state. To determine the state information, the impedance spectrum and the state vector are input into a second neural network. The second neural network is configured to determine the state information based on the impedance spectrum and the state vector.

[0017] It has been recognized that the problem of time dependency or hysteresis of the measurement can be reliably avoided by using two neural networks. In order to overcome the problem of time dependency of the impedance measurement, it is proposed that the previous history of the monomer chemistry is recorded in a first step by means of existing measurement inputs in the vehicle and processed by the first neural network without carrying out impedance spectroscopy. In order to improve the estimation of the state of the monomer by means of the impedance spectrum and the time series, it is proposed to combine the output of the first neural network, i.e. the state vector, with the data of the impedance spectrum in one vector and to input this joint vector into the second neural network in order to calculate the state information.

[0018] Thus, an improved determination of the state of charge and / or state of aging can be achieved and, for example, the range reserve can be reduced. Thus, by reducing the range reserve, a significant reduction in battery costs can be achieved (particularly in LFP batteries). Furthermore, the remaining range can be increased by a precise cell state estimation that takes into account both the cell prior history by means of the first neural network and the impedance spectrum by means of the second neural network. The method makes it possible to obtain the temporal profile of the cell chemistry as important information for the state estimation during dynamic driving with a non-constant direct current.

[0019] The single cell state parameter may include a single cell current, a single cell voltage, a single cell temperature and / or a parameter derived therefrom. The single cell current is the current of the battery cell to be monitored. The current is the current that can be used for the electric drive device, and may have a different sign when the battery cell is charged and / or recovered than when the motor vehicle is driven. The current is distributed by the single cells connected in parallel in the memory. The single cell voltage is the voltage of the battery cell to be monitored. The single cell temperature is the temperature of the battery cell to be monitored. The current state of the battery cell can be fully characterized by the single cell current, the single cell voltage and the single cell temperature. Therefore, according to the time series of the single cell state parameter, it is possible to fully characterize the development of the state of the battery cell in time. The derived parameter may include, for example, the gradient of the current curve, the voltage curve and / or the temperature curve. The derived parameter may include the integral of the current parameter, the voltage parameter and / or the temperature parameter in time. Alternatively or additionally, the derived parameter may in particular include the difference between two current values, two voltage values ​​and / or two temperature values.

[0020] When the development of the monomer current and / or the monomer voltage in time meets the predetermined conditions, the impedance spectrum can be measured. Here, the predetermined conditions can be defined so that the impedance measurement can be reliably implemented. For example, the predetermined conditions can involve fluctuations that characterize the dynamics of the monomer current and / or the monomer voltage, the variance and / or the standard deviation of the monomer current and / or the monomer voltage. When the dynamics of the monomer current and / or the monomer voltage is less than a predetermined threshold value, the impedance measurement can be performed. Otherwise, when the dynamics of the monomer current and / or the monomer voltage is greater than the threshold value, the impedance measurement stops and is performed at a later moment, at which the development of the monomer current and / or the monomer voltage in time meets the predetermined conditions.

[0021] When the cell current is constant, that is, when the dynamics in the chemistry of the battery cell are low, the impedance spectrum can be measured. For example, a constant cell current indicates that the chemical dynamics are low, and this means that the cell current, and thus the cell voltage, is sufficiently constant for impedance measurement by impedance spectroscopy. Here, the temporal development of the cell voltage and / or the cell current is largely fixed. For example, it is sufficient if the cell voltage and / or the cell current only change by a few percent and / or a few thousandths within a few seconds. If the cell voltage and / or the cell current are constant in this sense, the temporal development of the cell current meets the predetermined conditions. Otherwise, when the cell voltage and / or the cell current are not constant in this sense, the impedance measurement stops and is performed at a later time, at which the cell current is largely constant and meets the predetermined conditions. For the same reason, the impedance spectrum can be measured when the cell current is below a predetermined cell current threshold. Here, the cell current threshold can depend on the chemistry of the battery cell.

[0022] The impedance spectrum can be determined when the cell voltage is constant and / or the temporal development of the cell voltage satisfies a predetermined fluctuation condition. For example, a constant cell voltage indicates a low chemical dynamics and means that the cell voltage is sufficiently constant for the impedance measurement by impedance spectroscopy. In this case, the temporal development of the cell voltage is largely fixed. For example, it is sufficient if the cell voltage changes only by a few percent and / or a few thousandths within a few seconds. In particular, it is sufficient if the cell voltage changes only by a few millivolts within a few seconds. If the cell voltage is constant in this sense, the temporal development of the cell voltage satisfies the predetermined fluctuation condition. Otherwise, when the cell voltage is not constant in this sense, the impedance measurement is stopped and carried out at a later time, at which the cell voltage is largely constant and satisfies the predetermined fluctuation condition. In this case, the fluctuation condition can depend on the chemistry of the battery cell.

[0023] The first neural network may include a recurrent neural network (RNN), such as a long short-term memory network (LSTM). RNN is capable of detecting information in time. It is thus possible to effectively characterize time series. Therefore, the first neural network includes temporal feedback that combines measurements at later times with measurements at earlier times. Therefore, the previous history of the battery cell can be reliably taken into account in order to determine the state vector.

[0024] The second neural network can be a fully connected neural network (DNN) and / or a convolutional neural network (CNN). Therefore, the second neural network can be efficiently trained and can simultaneously process complex and highly dimensional inputs, such as inputs generated by the combination of the state vector and the impedance spectrum.

[0025] The second neural network can have a first layer for inputting the impedance spectrum and a second layer arranged downstream of the first layer for inputting the state vector. Thus, the impedance spectrum can be efficiently processed first and the state vector can then be input into the second neural network, for example, to reduce the dimension of the variable to be input.

[0026] In other words, with reference to a specific design scheme described as not limiting the present disclosure, the content described above can be summarized as follows: The present disclosure relates to a KI (artificial intelligence) method for performing cell state identification in a vehicle by means of electrochemical impedance spectroscopy. The prior art for realizing battery cell state identification includes monitoring the OCV curve of the battery cell. In addition, energy integrators are used. These energy integrators integrate the current over time in order to determine the SOC, for example, by adding up the consumed energy. In order to determine the SOH, current pulse measurement methods are usually applied in laboratories and / or on test benches. The prior art needs to be improved in the following aspects: 1) The OCV curve is not applicable to all chemistries or battery cell types, and the cell state cannot be identified. In particular, in LFP cells, the OCV curve method is even less applicable due to the lack of slope (Steigung) and hysteresis. 2) Energy integrators that integrate energy over time are very conducive to state determination and are mainly used on test benches. Here, they meet high requirements for voltage and current sampling rates. They have the disadvantage that they are too expensive for use in motor vehicles. 3) Known cell state identification methods rarely take into account chemical and electrical prior history. In order to circumvent these problems, the following method is proposed, which constructs a sequence with corresponding KI structures: 1) In order to circumvent the problem of impedance measurement during driving, it is proposed that in a first step, the LSTM KI structure is used to record the prior history of the cell chemistry during driving based on the existing measurement inputs in the vehicle (such as current, voltage, temperature), without having to perform impedance measurements. 2) In order to significantly improve the cell state estimation with the help of impedance measurement in terms of prior history, it is proposed that impedance measurements are generated when the DC current is constant, such as when the motor vehicle is parked, stopped at a traffic light and / or a constant current is drawn from the battery. The output of the LSTM Kl structure is then combined with the impedance measurement data into a vector. This joint vector is input into a second KI structure (such as a fully connected layer / dense layer with a sigmoid output activation function) in order to calculate the target cell state parameter, such as SOC or SOH.

[0027] Furthermore, a computer program is provided, which comprises instructions which, when said program is executed by a computer, cause the computer to at least partially implement or execute the method described above.

[0028] The program code of the computer program may be in any desired code form, in particular in a code form suitable for a motor vehicle control unit.

[0029] The above description regarding the method also applies analogously to the computer program and vice versa.

[0030] Furthermore, a data processing device, for example a control device, for an automated motor vehicle is provided, wherein the data processing device is configured to at least partially implement or execute the method described above. The method is therefore a computer-implemented method.

[0031] The data processing device may be or represent a part of a driver assistance system. The data processing device may be, for example, an electronic control unit (ECU). The electronic control device may be an intelligent unit controlled by a processor, which may communicate with other modules, for example, via a central gateway (CGW), and may form a vehicle on-board network, for example, via a field bus (such as CAN bus, LIN bus, MOST bus and FlexRay) or via automotive Ethernet, for example, together with a telematics control device. It is conceivable that the control device controls functions that are critical to the driving performance of the motor vehicle, such as engine control, transmission, brake system and / or tire pressure control system. The data processing device may also be located in the cloud or run by a server outside the vehicle, for example, and may be connected and / or connectable to the motor vehicle by means of data transmission.

[0032] The above description regarding the method and the computer program also applies analogously to the data processing device and vice versa.

[0033] Furthermore, a motor vehicle is provided which comprises a data processing device as described above.

[0034] The motor vehicle may be a passenger car, in particular a car. The motor vehicle may be designed to be optionally automated so that the longitudinal guidance and / or transverse guidance is at least partially and / or at least temporarily taken over during the automated driving of the motor vehicle. The automated driving may be performed so that the forward movement of the motor vehicle is (largely) autonomous. The automated driving may be controlled at least partially and / or temporarily by a data processing device. The motor vehicle may be a motor vehicle of autonomy level 0 to 5.

[0035] The above description with respect to the method, the data processing device and the computer program also applies analogously to the motor vehicle and vice versa.

[0036] Furthermore, a computer-readable medium, in particular a computer-readable storage medium, is provided. The computer-readable medium comprises instructions, which, when the program is executed by a computer, cause the computer to at least partially implement the method described above.

[0037] That is, a computer-readable medium comprising the computer program defined above may be provided. The computer-readable medium may be any digital data storage device, such as a USB stick, a hard drive, an SD card or an SSD card. The computer program need not necessarily be stored on such a computer-readable storage medium to be provided to the motor vehicle, but may also be obtained via the Internet or in other external ways.

[0038] The above description regarding the method, the data processing device, the computer program and the automated vehicle also applies analogously to the computer-readable medium and vice versa. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Refer to the following Figures 1 to 3 An implementation method is described below.

[0040] Figure 1 A motor vehicle according to one aspect of the present disclosure is schematically illustrated;

[0041] Figure 2 Schematically illustrates an architecture for performing data processing by a data processing device according to one aspect of the present disclosure; and

[0042] Figure 3 A flow chart of a method according to one aspect of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0043] Figure 1 A motor vehicle 200 according to one aspect of the disclosure is schematically shown. The motor vehicle 200 is an electrically drivable motor vehicle 200. For this purpose, the motor vehicle 200 has an electric drive 220 and a battery 210. The electric drive 220 can be supplied with electrical energy 225 via the battery 210, which the electric drive 220 can convert into mechanical energy. In this case, the state of charge of the battery 210 decreases. The electric drive 220 is provided for recuperation, i.e. for performing regenerative braking, in which mechanical energy is converted into electrical energy 225 which can be supplied to the battery 210. In this case and when the battery 210 is charged via an external charging station (not shown), the state of charge of the battery 210 may increase.

[0044] The motor vehicle 200 has a data processing device 250. The data processing device 250 is configured to monitor the battery cells 215 of the battery 210 of the motor vehicle 200. For this purpose, the data processing device 250 is configured to monitor the battery cells 215 of the battery 210 of the motor vehicle 200. Figure 2 The described architecture is used to implement the reference Figure 3 Method 100 is described.

[0045] The battery 210 has a plurality of battery cells 215, such as Figure 1Each battery cell 215 can be characterized by a cell state parameter 265. The cell state parameter 265 includes a cell current I, a cell temperature T, and a cell voltage U. The cell current I is typically a direct current (DC current), which is used for the electric drive device 220 and becomes negative during charging and / or recovery.

[0046] The cell current I, the cell voltage U and the cell temperature T can be measured by a measuring device (not shown) connected to the control device of the battery 210 and can be transmitted to the data processing device 250. Other variables such as gradients, differences and / or integrals can be determined or derived from the cell current I, the cell voltage U and / or the cell temperature T. By repeatedly measuring the cell state parameters 265, the time series 260 of the cell state parameters 265 can be determined and transmitted to the data processing device 250, or in another embodiment, determined by the data processing device 250 based on a set of received cell state parameters 265. The measurement of the cell current I, the cell voltage U and the cell temperature T is performed, for example, during driving, and the time series 260 is provided with a time stamp corresponding to the time of the measurement. The time series 260 can include cell state parameters 265, which are measured, for example, every second and characterize the development in time within thousands of seconds.

[0047] The impedance Z or impedance spectrum 275 of the battery cell 215 can be measured by a measuring device (not shown). For this purpose, an alternating current is applied to the battery cell 215 and a response signal is measured. The impedance spectrum 275 is transmitted to the data processing device 250 or, in another embodiment, is determined by the data processing device 250 based on the received measured values ​​of the impedance Z.

[0048] Figure 2 The diagram schematically shows an architecture for performing data processing by a data processing device 250 according to one aspect of the present disclosure.

[0049] The data processing device 250 includes a memory (not shown) for storing data, a processor (not shown) for processing data, and an interface for receiving input of measurement values ​​and outputting information. Figure 1 As input, the described cell state variables 265 (i.e., cell current I, cell temperature T, and cell voltage U) or the time series 260 of the cell state variables 265 as well as the impedance Z and the impedance spectrum 275 are used, as indicated by the incoming arrows. The data processing device 250 is configured to determine and output state information 290, as indicated by the outgoing arrows.

[0050] Furthermore, data processing device 250 is configured to process inputs using first neural network 280 and second neural network 285 and to determine state information 290 .

[0051] The first neural network 280 receives as input the cell state variables 265 (i.e., the cell current I, the cell temperature T, and the cell voltage U) or the time series 260 of the cell state variables 265, wherein the time series 260 has time stamps of the cell state variables 265. For example, these time stamps can be explicitly included as variables in the time series 260 and / or can be implicitly obtained by the periodicity of the measurement of the cell state variables 265.

[0052] The first neural network 280 comprises a recurrent neural network (RNN), for example a long short-term memory network (LSTM). In this case, the neural network 280 comprises a time feedback 281 so that the history of the individual state variables 265 represented by the time series 260 can be appropriately taken into account.

[0053] The first neural network 280 generates a state vector 270. The state vector 270 is a vector with information about the previous history of the measured battery cell 215. The state vector 270 can include one or more values, in other words, can be an n-tuple with a natural number n. A corresponding state vector 270 can be calculated for each of the battery cells 215. The second neural network 285 includes a first layer 286 for inputting an impedance spectrum 275 and a second layer 287 arranged downstream of the first layer 286 for inputting the state vector 270. The second neural network 285 thus receives the information or state vector 270 from the first neural network 280 and thus receives the information about the previous history of the battery cell 215 in the vector plus the data of the impedance spectroscopy, and creates a concatenated vector therefrom. The concatenated vector, i.e. the state vector 270 and the impedance spectrum 275, is then used as input to the second layer 287. If, for example, the impedance spectrum 275 is measured at 75 frequencies, 2×75 measurement values ​​are obtained, 75 measurement values ​​each for the real part of the impedance Z and the imaginary part of the impedance Z. Thus, 75 data points are obtained that are input into the second layer 287. Therefore, the first layer 286 for inputting the impedance spectrum 275 includes as many nodes as the frequencies used when measuring the impedance spectrum 275.

[0054] The first layer 286 is connected to the second layer 287. The second neural network 285 may have other layers (not shown). The second neural network 285 includes a convolutional neural network (CNN) and / or a fully connected neural network (DNN) with fully connected layers (FC).

[0055] The output of the second neural network 285 is the state information 290 to be predicted, such as the state of charge SOC and / or the state of aging SOH. The second neural network 285 can incorporate the valuable information of the impedance spectroscopy into the determination of the state information 275 based on the output of the first neural network 280 (the output includes, for example, measured variables during driving before the impedance spectroscopy) plus the impedance spectrum 275 after driving (for example, when the DC current is constant). The two neural networks 280, 285 are connected in two stages and form an artificial intelligence that can achieve the best single cell state estimation. The first neural network 280 can be trained by detecting test bench data. Here, the state of charge of the battery cell 215 is measured during the operation of the motor vehicle 200, and the time series 260 of the single cell state variable 265 is recorded and labeled according to the state of charge. By recording the time series 260, the dependence on past measurements or history and / or hysteresis is taken into account. The second neural network 285 can be trained by simulating a predetermined scenario and / or driving profile for driving the motor vehicle 200. For example, over a period of hours and / or days, a time series 260 of individual state variables 265 is measured and processed by a first neural network 280. In each case, i.e., for example, when the motor vehicle 200 is stationary, an impedance spectrum 275 is measured and compared with test bench data, wherein the two networks 280, 285 are trained by back propagation according to the comparison.

[0056] Figure 3 Schematically shows a flow chart of a method 100 according to one aspect of the present disclosure. The method 100 is a method 100 for monitoring a battery cell 215 of a battery 210 of a motor vehicle 200. Figure 1 Such a motor vehicle 200 is described, wherein reference is made to the accompanying drawings. Figure 2 The architecture on which the monitoring of the battery cells 215 is based is described. Figure 1 and Figure 2 Description Figure 3 Method 100.

[0057] according to Figure 3 The method 100 includes: detecting 110 a time series 260 of a cell state variable 265 characterizing a battery cell 215 .

[0058] The determination 120 of the state vector 270 takes place by inputting the time series 260 into a first neural network 280 .

[0059] The impedance spectrum 275 of the battery cell 215 is determined 130 by means of impedance spectroscopy. The impedance spectrum 275 is determined 130 when the temporal development of the cell voltage U and / or the cell current I meets a predetermined condition. The impedance spectrum 275 is determined 130 when the cell voltage U and / or the cell current I are constant. The impedance measurement can be performed when the dynamics of the cell voltage U and / or the cell current I are less than a threshold value. Otherwise, when the dynamics of the cell voltage U and / or the cell current I are greater than a threshold value, the impedance measurement is stopped and performed at a later time, at which the temporal development of the cell voltage U and / or the cell current I meets a predetermined condition. The impedance spectrum 275 is determined 130 when the cell current I is below a predetermined cell current threshold value and / or when the cell voltage meets a predetermined fluctuation condition.

[0060] Determination 140 of state information 290 for monitoring battery cell 215 is performed by inputting impedance spectrum 275 and state vector 270 into second neural network 285 .

[0061] A person skilled in the art recognizes that the steps of method 100 may be performed simultaneously and / or continuously. Detection 110 of time series 260 can be performed substantially continuously and may also be performed, for example, during determination 120 of state vector 270 and / or determination of state information 290 .

[0062] Reference numerals list

[0063] 100 Methods

[0064] 110 detection

[0065] 120 Determine the state vector

[0066] 130 determinations

[0067] 140 Confirm status information

[0068] 200 Motor Vehicles

[0069] 210 battery

[0070] 215 battery cells

[0071] 220 electric drive

[0072] 225 Electricity

[0073] 250 data processing equipment

[0074] 260 Time Series

[0075] 265 monomer state parameters

[0076] 270 state vector

[0077] 275 Impedance Spectroscopy

[0078] 280 First Neural Network

[0079] 281 Feedback

[0080] 285 Second Neural Network

[0081] 286 First Floor

[0082] 287 Second Floor

[0083] 290 Status Information

[0084] RNN Recurrent Neural Network

[0085] LSTM Long Short-Term Memory Network

[0086] CNN Convolutional Neural Network

[0087] DNN fully connected neural network

[0088] FC fully connected layer

[0089] I Single-cell current

[0090] T monomer temperature

[0091] U single cell voltage

[0092] Z single impedance

Claims

1. A method (100) for monitoring a battery cell (215) of a battery (210) of a motor vehicle (200), wherein: The method (100) comprises: - detecting (110) a time series (260) of a cell state parameter (265) characterizing a battery cell (215); - determining (120) a state vector (270) by inputting the time series (260) into a first neural network (280); - determining (130) an impedance spectrum (275) of the battery cell (215) by means of impedance spectroscopy; and - determining (140) state information (290) for monitoring the battery cell (215) by inputting the impedance spectrum (275) and the state vector (270) into a second neural network (285).

2. The method according to claim 1, wherein: The cell state variables (265) include cell current (I), cell voltage (U), cell temperature (T) and / or variables derived therefrom.

3. The method according to claim 1 or 2, wherein: The impedance spectrum (275) is determined (130) when the temporal development of the cell current (I) and / or the cell voltage (U) meets predetermined conditions.

4. The method according to claim 3, wherein: The determination (130) of the impedance spectrum (275) is performed when the cell current (I) is constant and / or below a predetermined cell current threshold.

5. The method according to claim 3 or 4, wherein: The impedance spectrum (275) is determined (130) when the cell voltage (U) is constant and / or the temporal development of the cell voltage (U) satisfies a predetermined fluctuation condition.

6. The method according to any one of the preceding claims, wherein: The first neural network (280) includes a recurrent neural network (RNN) and / or a long short-term memory network (LSTM).

7. The method according to any one of the preceding claims, wherein: The second neural network (285) includes a convolutional neural network (CNN) and / or a fully connected neural network (DNN).

8. The method according to any one of the preceding claims, wherein: The second neural network (285) has a first layer (286) for inputting an impedance spectrum (275) and a second layer (287) arranged downstream of the first layer (286) for inputting the state vector (270).

9. Computer program and / or computer-readable medium comprising instructions which, when the program or instructions are executed by a computer, cause the computer to execute the method (100) and / or the steps of the method (100) according to one of claims 1 to 8.

10. A data processing device (250) for a motor vehicle (200), wherein: The data processing device (250) is configured to carry out the method (100) according to one of claims 1 to 8.

11. A motor vehicle (200) comprising a data processing device (250) according to claim 10.

Images