Method and system for determining a characteristic value of a storage unit for electrical energy

Abstract

A method for determining a characteristic value of a storage unit for electrical energy comprises the following steps: detecting distance sensor data by means of at least one distance sensor, wherein the distance sensor data is based on at least one distance measurement between the distance sensor and the storage unit for electrical energy, and determining the characteristic value of the storage unit for electrical energy on the basis of the distance sensor data and using a neural network.

Description

[0001] Method and system for determining a characteristic value of a storage unit for electrical energy

[0002] The present disclosure relates to methods for determining a characteristic value of an electrical energy storage unit and a system for determining a characteristic value of an electrical energy storage unit.

[0003] background

[0004] Batteries, such as lithium-ion batteries, are used in a variety of energy storage applications, including storing electrical energy in electric vehicles or buildings. Battery manufacturing is complex and requires a multi-stage process that is prone to errors. Due to the manufacturing process, the proportion of batteries that are either non-functional or only partially functional is relatively high.

[0005] Known defects in lithium-ion batteries include gas formation, lithium plating, and the formation of a solid electrolyte interface layer (SEI). Gas formation can occur due to decomposition of the electrolyte and electrodes. In lithium plating, lithium ions can be deposited on the anode surface and reduced to metallic lithium. An SEI layer typically forms during battery cell manufacturing between the electrolyte and the negative electrode as a solid, electrically insulating layer. The formation of the SEI layer is thought to be caused by electrolyte decomposition at the electrode-electrolyte interface. These defects can severely limit battery performance.

[0006] Battery cell manufacturers test the battery cells at the end of production. However, they also test the battery cells before they are used in batteries. Battery cell and battery manufacturing can be carried out by one company or by different companies. This is intended to reduce the number of defective or underperforming battery cells, battery modules, or batteries delivered to a customer. Known methods for testing battery cells are often relatively imprecise, meaning that usable battery cells can be classified as unusable, and vice versa. Often, end-of-production tests are limited to electrical measurements, with the battery or battery cell capacity being the primary focus. However, capacity is not the only indicator of a battery's or battery cell's quality.

[0007] Disclosure of the invention

[0008] The purpose of the present disclosure is to determine a characteristic value for an electrical energy storage unit with improved accuracy.

[0009] The problem is solved by the features of the independent claims. Preferred embodiments are specified in the dependent claims and the description.

[0010] A method for determining a characteristic value of an electrical energy storage unit is disclosed, the method comprising the steps of: acquiring distance sensor data using at least one distance sensor, wherein the distance sensor data is based on at least one distance measurement between the distance sensor and the electrical energy storage unit, and determining the characteristic value of the electrical energy storage unit based on the distance sensor data and using a neural network.

[0011] Also disclosed is a system for determining a characteristic value of an electrical energy storage unit. The system comprises at least one distance sensor configured to acquire distance sensor data, wherein the distance sensor data is based on at least one distance measurement between the distance sensor and the electrical energy storage unit. The system comprises an evaluation unit configured to determine the characteristic value of the electrical energy storage unit based on the distance sensor data and using a neural network.

[0012] Based on distance measurements, the expansion or volume of the storage unit can be determined. This expansion or volume correlates with the lithium transport process within the unit. Low lithium transport indicates reduced performance or quality. During charging, the storage unit typically expands, while during discharging, it shrinks. Typical expansion and shrinkage values ​​range from approximately 5% to 10% of the unit's thickness. If these values ​​are exceeded, or if expansion is irreversible, a defect or reduced performance of the storage unit may be detected.

[0013] Determining the characteristic value of the electrical energy storage unit can still be based on storage unit data. This data can be fed into the neural network, particularly in conjunction with the distance sensor data. Using the neural network, the characteristic value of the storage unit can be determined based on both the distance sensor data and the storage unit data. The storage unit data can be, or become, linked to the distance sensor data, especially in terms of timing.

[0014] The storage unit data may include information about the state of charge of the storage unit, information about the history of the charging and / or discharging processes of the storage unit, information about the type and / or model of the storage unit, information about the shape and / or geometry of the storage unit, and / or information about the planned type of use of the storage unit.

[0015] The characteristic value can contain or enable information about the quality of the electrical energy storage unit. This quality information can be qualitative or quantitative. For example, the quality information can be "good" or "poor." Alternatively, the quality information can be a numerical value used to assess the quality of the storage unit, e.g., a percentage, where 100% represents a storage unit without defects and 0% represents an unusable storage unit.

[0016] The characteristic value can contain or enable information about the estimated service life of the electrical energy storage unit. The service life can be a duration for a specific application. The characteristic value can contain or enable information about the estimated number of charge and / or discharge cycles of the electrical energy storage unit. For example, the information can indicate how many charge and / or discharge cycles are expected before the storage unit loses its functionality. Based on the number of estimated charge and / or discharge cycles, different types of use can be planned for the storage unit.

[0017] The characteristic value can contain or enable information about the type of use of the electrical energy storage unit. Storage units without defects can be used for demanding applications, and storage units with few defects can be used for less demanding applications.

[0018] The characteristic value can contain or enable information about the state of the electrical energy storage unit. This information about the state can be qualitative or quantitative.

[0019] The characteristic value can contain or enable information about a defect in the electrical energy storage unit.

[0020] Based on the characteristic value, the electrical energy storage unit can be classified. The characteristic value can contain or enable information about the classification of the electrical energy storage unit. For example, the storage unit is classified as "normal" or "abnormal".

[0021] The characteristic value can contain or enable quality prediction of the storage unit.

[0022] The electrical energy storage unit can be a battery or a part of a battery. Preferably, the electrical energy storage unit is a lithium-ion battery or a part of a lithium-ion battery. The electrical energy storage unit can be a cell. A battery can contain a plurality of cells.

[0023] The storage unit for electrical energy can have a cylindrical shape or a pouch shape.

[0024] The distance sensor can comprise or be an optical distance sensor. Preferably, the distance sensor comprises or is a laser triangulation sensor, a lidar sensor (light detection and ranging or light imaging, detection and ranging), an optical coherence tomograph, and / or a chromatic confocal sensor. Particularly preferably, the distance sensor comprises a chromatic confocal sensor. The chromatic confocal sensor, in particular, enables a compact design for the measurement system. Likewise, the chromatic confocal sensor enables high-accuracy and non-contact measurement.

[0025] The distance sensor can be configured to determine the distance between itself and the storage unit without physical contact. This means the distance sensor cannot make contact with the storage unit while determining the distance between the sensor and the storage unit.

[0026] To determine the distance between the distance sensor and the storage unit, the distance sensor can be configured to project a measuring beam, in particular an optical measuring beam, onto a surface of the storage unit. Based on the reflection (of a portion) of the measuring beam from the surface of the storage unit, the distance between the two storage units can be determined.

[0027] Distance sensor data can be acquired using at least two distance sensors. Preferably, the distance sensor data is acquired using at least three distance sensors. The distance sensors can be identical or different types. The distance sensors can be configured to determine the distance between the respective sensor and the surface of the storage unit. Distance sensor data from all sensors can be used to determine the characteristic value of the storage unit. For this purpose, distance sensor data from all sensors can be fed into the neural network.

[0028] Distance sensor data can be collected over a period of at least 1 minute. Preferably, distance sensor data is collected over a period of at least 1 hour. The distance sensor data can be fed continuously into the neural network or collected over a defined period and the collected data fed into the neural network.

[0029] The distance sensor data acquired over the specified period can be acquired by one or more distance sensors. Acquisition can be performed at a measurement frequency of at least 0.1 Hz, preferably at least 1 Hz. The measurement frequency can be between 0.1 Hz and 2 Hz, preferably between 0.1 Hz and 1 Hz.

[0030] The distance sensor data can be based on measurements taken at least at two different locations or positions of the electrical energy storage unit. Preferably, the distance sensor data is based on measurements taken at least at three different locations or positions of the electrical energy storage unit.

[0031] The locations or positions can lie on (exactly) one surface of the storage unit. Alternatively, the locations or positions can lie on at least two different surfaces of the storage units.

[0032] If exactly one distance sensor is used, the distance sensor can be configured to perform distance measurements at at least two different locations or positions. For this purpose, the distance sensor can, for example, include a scanner device to direct the measuring beam to different locations or positions of the storage unit. The distance sensor can have a high measurement frequency (e.g., at least or approximately 1 kHz) so that the measurements at different locations or positions take place quasi-simultaneously. Exactly one distance sensor can be provided for each location or position of the storage unit. Distance measurements between a first distance sensor and a first location or position can be performed by the first distance sensor. Distance measurements between a second distance sensor and a second location or position can be performed by the second distance sensor.Similarly, more than one distance sensor for baseline measurements can be provided for each location or position of the storage unit.

[0033] The distance sensor data can include data from all distance sensors and all locations or positions of the storage unit. The storage unit's characteristic value can be determined from this distance sensor data.

[0034] The neural network can be a trained neural network. Preferably, the neural network is a semi-supervised neural network (or a semi-supervised trained neural network). The neural network can also be a transformer-type neural network.

[0035] The semi-supervised approach can use an autoencoder (unsupervised) to identify abnormal cells. These abnormal cells can then be further investigated (e.g., through destructive post-mortem analysis or aging ("dg / ' / zg") to determine potential lifetime problems). The results of further tests (defect type, lifetime, etc.) can then be used as labels for training a supervised neural network (possibly fully connected, convolutional, sequential, or a combination of all three). The neural network can be a fully connected neural network, a convolutional neural network, or a sequential neural network. Likewise, the neural network can be a combination of at least two of these networks, and in particular, a combination of all three.After training, this network can be used to classify cells based on new test data.

[0036] Distance measurement(s) can be performed during a charging and / or discharging process of the electrical energy storage unit. Distance measurements can be performed during multiple charging and / or discharging processes of the electrical energy storage unit.

[0037] During charging, the storage unit typically expands, resulting in a shorter distance being measured between the distance sensor and the storage unit. Conversely, during discharging, the storage unit typically shrinks, leading to a larger distance being measured between the distance sensor and the storage unit.

[0038] If a minimum distance threshold is not met during a charging process, it can be determined that the storage unit is defective or has limited performance. Similarly, if a minimum distance threshold is exceeded during a discharging process, it can be determined that the storage unit is defective or has limited performance. The defect or limited performance can be inferred from the measured characteristic value.

[0039] Before a charging and / or discharging process, particularly the first one, the distance between the distance sensor(s) and the storage unit can be determined. After one or more charging and / or discharging processes, the distance between the distance sensor(s) and the storage unit can be determined. These distances can be compared. If a deviation exists, especially above a threshold value, it can be determined that the storage unit is defective or has limited performance. Preferably, identical charge states of the storage unit are considered for this purpose. That is, for example, the distance before the first charging of the storage unit is determined and compared with a distance determined after the storage unit has been charged and discharged once or several times. This allows for the detection of irreversible expansion of the storage unit.

[0040] The distance sensor(s) can be stationary. This allows changes in the distance between the sensor and the storage unit to be attributed to expansion or contraction of the storage unit. At least one distance measurement can be performed at the end of the electrical energy storage unit's production process ("end-of-line testing"). After production and before delivery, at least one distance measurement can be taken to determine the characteristic value. This allows defective storage units to be rejected before delivery, or for units with limited performance to be assigned to less demanding applications.At least one distance measurement can be performed on one or more electrical energy storage units before these units are further processed into larger storage unit sets or battery packs. There, the distance measurement can be carried out as an incoming test.

[0041] Alternatively or additionally, at least one distance measurement can be carried out upon receipt of goods or before manufacturing a battery with multiple storage units. When a battery manufacturer acquires storage units, these can be checked for defects or classified according to their performance. This allows defective storage units to be excluded from battery production or assigned to specific uses within the subsequently manufactured battery.

[0042] The distance sensor data can be fed into the neural network as raw data. Alternatively, feature extraction can be applied to the distance sensor data. After feature extraction, the extracted data can be fed into the neural network. Feature extraction can also be performed within the neural network itself.

[0043] The method may further include: acquiring electrical sensor data using at least one electrical sensor, wherein the electrical sensor data is based on at least one measurement of at least one electrical property of the electrical energy storage unit, and determining the characteristic value of the electrical energy storage unit based on the distance sensor data and the electrical sensor data, and using the neural network.

[0044] If both the distance sensor data and the electrical sensor data are fed into the neural network, the characteristic value of the storage unit can be determined more precisely. The characteristic value of the electrical energy storage unit can be determined based on the distance sensor data, the electrical sensor data, and the storage unit data, and using the neural network.

[0045] The electrical sensor can be configured to perform a current and / or voltage measurement at the storage unit.

[0046] The electrical sensor can be (physically) connected to the storage unit during the measurement of the electrical property.

[0047] The electrical property can be a voltage and / or a current.

[0048] The system for determining the characteristic value of the electrical energy storage unit can include at least one electrical sensor. The electrical sensor can be configured to acquire electrical sensor data, wherein the electrical sensor data is based on at least one measurement of at least one electrical property of the electrical energy storage unit. The evaluation unit can be configured to determine the characteristic value of the electrical energy storage unit based on the distance sensor data and the electrical sensor data, and using the neural network.

[0049] At least one measurement of at least one electrical property can be performed during a charging and / or discharging process of the electrical energy storage unit.

[0050] At least one measurement of at least one electrical property can be carried out at the end of the production process of the electrical energy storage unit.

[0051] At least one measurement of the at least one electrical property and at least one distance measurement can be performed simultaneously.

[0052] Electrical sensor data can be acquired over a period of at least 1 minute. Preferably, electrical sensor data is acquired over a period of at least 1 hour. The electrical sensor data can be fed continuously to the neural network or collected over a defined period, and the collected data can then be fed to the neural network. The electrical sensor data can be fed to the neural network simultaneously with the distance sensor data. Acquisition can be performed at a measurement frequency of at least 1 Hz, preferably at least 5 Hz.

[0053] At least the distance sensor data, the electrical sensor data, and / or the storage unit data can be an input to the neural network. In particular, at least the distance sensor data and the electrical sensor data are an input to the neural network.

[0054] Distance sensor data and / or electrical sensor data can be acquired during a charging and / or discharging process of the storage unit or during multiple charging and discharging cycles. The distance sensor data and / or electrical sensor data can be linked to, or be associated with, a property of the storage unit. That is, each distance measurement and / or measurement of the electrical property can be assigned to a property of the storage unit. This property of the storage unit can be its state of charge.

[0055] In the Ab stand measurement, the distance between the distance sensor and the storage unit for electrical energy can be determined directly or indirectly.

[0056] Direct distance measurement allows for the determination of the distance between the distance sensor and a surface of the electrical energy storage unit. In this method, no (opaque) element can be present between the distance sensor and the surface of the electrical energy storage unit. During distance measurement, an (optical) measuring beam from the distance sensor can either directly scan the surface of the electrical energy storage unit or be reflected from it. Direct distance measurement is preferred when the electrical energy storage unit is cylindrical.

[0057] In an indirect distance determination, a distance can be determined between the distance sensor and an element connected to or linked with a surface of the electrical energy storage unit. Preferably, the element contacts a surface of the electrical energy storage unit. The element can be a plate that contacts or rests on a surface of the electrical energy storage unit. The element can be a holder or mounting by which the electrical energy storage unit is held or secured, particularly during a charging and / or discharging process of the electrical energy storage unit. The element can prevent an (optical) measuring beam of the distance sensor from directly accessing a surface of the electrical energy storage unit.The (optical) measuring beam may be obstructed by the element from directly scanning a surface of the electrical energy storage unit or from being directly reflected by the surface. The element may be connected or linked to the electrical energy storage unit in such a way that movement of the electrical energy storage unit is transmitted to the element. A surface of the element may be directly scanned by an (optical) measuring beam from the distance sensor or reflected by it. Based on a change in the distance between the distance sensor and the element, a change in the distance between the distance sensor and the electrical energy storage unit can be inferred.For example, a change in the distance between the distance sensor and the element (directly or immediately) can represent a change in the distance between the distance sensor and the electrical energy storage unit. Indirect distance determination is preferred when the electrical energy storage unit is pouch-shaped (a "pouch cell").

[0058] Brief description of the drawings

[0059] The invention is described in detail below with reference to figures.

[0060] Fig. 1 shows a system 100 for determining a characteristic value of a storage unit 10 for electrical energy;

[0061] Fig. 2 shows a system 100 for determining a characteristic value of a storage unit 10 for electrical energy; and

[0062] Fig. 3 shows a system 200 for determining a characteristic value of a storage unit 10 for electrical energy. Detailed description of the drawings

[0063] Fig. 1 shows a system 100 for determining a characteristic value of a storage unit 10 for electrical energy. Each method disclosed herein can be carried out by the system.

[0064] System 100 includes a distance sensor 20. The distance sensor 20 is configured to acquire distance sensor data. The distance sensor data is based on at least one distance measurement between the distance sensor 20 and the electrical energy storage unit 10 (hereinafter also referred to as storage unit 10).

[0065] The distance sensor data can be processed by a distance evaluation unit 30. For example, the distance evaluation unit 30 is configured to determine the distance between the distance sensor 20 and the storage unit 10 from the distance sensor data. The distance evaluation unit 30 can be part of the distance sensor 20. The distance sensor data can be raw data from the distance sensor 20 or pre-processed data, for example, by the distance evaluation unit 30.

[0066] To acquire distance sensor data, the distance sensor 20 can be configured to project an optical measuring beam 22 (directly or indirectly) onto the storage unit 10. For example, the optical measuring beam 22 can be projected directly onto a surface of the storage unit 10. Likewise, the optical measuring beam 22 can be projected onto a surface of an element (not shown in the figures) that is linked or connected to, or in contact with, the storage unit 10. In both cases, the distance sensor 20 can acquire distance sensor data based on at least one position measurement between the distance sensor 20 and the storage unit 10. Distance sensor data can be acquired over a longer period, e.g., over a period of at least 1 minute or at least 1 hour. This allows for the detection of any deformation (e.g., expansion) of the storage unit.During this period, a large number of measurements of the baseline can be carried out and included in the distance sensor data.

[0067] Figure 1 shows exactly one distance sensor 20, but at least two or at least three distance sensors can be provided. Each of the distance sensors can be configured to project a measurement beam onto the storage unit 10. In particular, each of the measurement beams from the distance sensors can be projected onto a different location or position of the storage unit 10, so that distance sensor data from different locations or positions of the storage unit are acquired. The total distance sensor data can be forwarded to the distance evaluation unit 30 for preprocessing. Alternatively, the total distance sensor data can be fed to the neural network as raw data.

[0068] Likewise, exactly one distance sensor 20 can be provided. The distance sensor 20 can be configured to perform a position measurement between the distance sensor 20 and exactly one location or position of the storage unit 10. Alternatively, the distance sensor 20 can be configured to perform a position measurement between the distance sensor 20 and at least two locations or positions of the storage unit 10. For this purpose, the measuring beam 22 can be deflected. The measurement frequency can be high, so that the measurements at different locations or positions of the storage unit 10 take place quasi-simultaneously or simultaneously.

[0069] At least one of the distance sensors 20, preferably each of the distance sensors 20, can be an optical distance sensor 20. Preferably, at least one of the distance sensors 20, preferably each of the distance sensors 20, is a chromatic confocal sensor or includes one. The chromatic confocal sensor has a compact design.

[0070] The storage unit 10 can be a lithium-ion battery or a cell of a lithium-ion battery. Various shapes of the storage unit 10 are possible. In particular, the storage unit is cylindrical or pouch-shaped.

[0071] The distance measurement can be a measurement between a surface of the storage unit 10 and the distance sensor 20. During the distance measurement, the distance sensor 20 can be stationary. This allows a change in the measured distance to be attributed to a deformation of the storage unit 10. The distance sensor data can be fed to a characteristic value evaluation unit 50. The characteristic value evaluation unit 50 can determine the characteristic value of the storage unit 10 using a neural network and based on the distance sensor data.

[0072] The neural network can be a semi-supervised neural network. The neural network can be a trained neural network.

[0073] By using the neural network, a variety of parameters can be processed to precisely determine the characteristic value of storage unit 10.

[0074] The characteristic value of storage unit 10 can provide information about its quality. For example, the characteristic value might indicate a low expected number of charge and / or discharge cycles before the storage unit 10 loses its functionality. Such a storage unit 10 can be selected for less demanding applications. Conversely, if the characteristic value indicates high quality, for example, because the expected maximum number of charge and / or discharge cycles is high, the storage unit 10 can be selected for demanding applications.

[0075] The neural network can classify storage units 10, for example as "normal" or "abnormal". The characteristic value can contain or enable information about the classification.

[0076] After the production of storage unit 10, an initial distance measurement can be performed. At least one further distance measurement can be performed after one or more charging and / or discharging cycles of storage unit 10. Distance sensor data, including information about the state of charge of storage unit 10, can be fed into the neural network. From this, the characteristic value of storage unit 10 can be determined.

[0077] If an expansion of the storage unit 10 is detected after one or more charging and discharging cycles, i.e., if the baseline measurement before charging and discharging shows a greater distance between the distance sensor 20 and the storage unit 10 than after charging and discharging, this may indicate irreversible expansion of the storage unit 10. Accordingly, the performance of the storage unit 10 may be limited.

[0078] Distance measurement can also provide information about the quality of the storage unit 10 during a charging and / or discharging process. As described earlier, a storage unit 10 can expand during a charging process and shrink during a discharging process. If the expansion exceeds or falls below a limit, a defect can be inferred. Similarly, if the shrinkage exceeds or falls below a limit, a defect can be inferred.

[0079] Using the neural network, a defect can be classified based on the distance sensor data. This means that the neural network can not only determine the quality of storage unit 10, for example using a characteristic value, but also identify or estimate the cause of the defect.

[0080] Distance sensor data can be acquired over an extended period, e.g., at least 1 minute, at least 1 hour, or during one or more charging and discharging cycles of storage unit 10. The distance sensor data can be continuously fed to the neural network. Alternatively, the distance sensor data can be buffered, and all buffered distance sensor data can be fed to the neural network at once.

[0081] A feature extraction unit 40 may be provided. The feature extraction unit 40 may be configured to extract features from the distance sensor data. The extracted features can be fed to the characteristic value evaluation unit 50 or the neural network. However, the feature extraction unit 40 is optional, and the distance sensor data can be fed to the characteristic value evaluation unit 50 or the neural network as raw data. Determining the characteristic value of the storage unit 10 can still be based on storage unit data. The storage unit data can be fed to the neural network, particularly together with the distance sensor data. Based on the storage unit data, the determination of the characteristic value can be refined, or limit values ​​for assessing the quality of the storage unit can be changed. For example, a change in shape (e.g.,Expansion of a first storage unit might indicate a defect, while the same change in shape of a second storage unit might be acceptable. The first and second storage units can be different models or types.

[0082] Fig. 2 shows the system 100 for determining the characteristic value of the storage unit as described with reference to Fig. 1. Features already described with reference to Fig. 1 are not repeated here, but are also disclosed with reference to the system shown and described in Fig. 2.

[0083] The system 100 can further include an electrical sensor 60. The electrical sensor 60 can be configured to acquire electrical sensor data based on at least one measurement of at least one electrical property of the storage unit. The electrical sensor data can be fed to the neural network or the characteristic value evaluation unit 50. Using the neural network and based on the distance sensor data and the electrical sensor data, the characteristic value of the storage unit 10 can be determined.

[0084] The electrical sensor 60 can be configured to perform a voltage and / or current measurement at the storage unit 10. The electrical sensor 60 can be, or include, a voltage and / or current sensor.

[0085] The electrical sensor data can be fed to the optional feature extraction unit 40. Feature extraction unit 40 can be configured to extract features from the electrical sensor data. The extracted features can be fed to the characteristic value evaluation unit 50 or the neural network. The electrical sensor data can also be fed to the characteristic value evaluation unit 50 or the neural network as raw data.

[0086] The measurement of at least one electrical property can be performed during a charging and / or discharging process of the storage unit 10. Alternatively or additionally, the measurement of the at least one electrical property can be performed when no charging and / or discharging process is taking place. For example, the measurement of the at least one electrical property is performed at a defined state of charge of the storage unit 10, preferably when the storage unit 10 is fully charged.

[0087] A large number of measurements of at least one electrical property can be carried out over a longer period of time, in particular over a period of at least 1 minute or 1 hour.

[0088] The measurement of at least one electrical property can be performed simultaneously with a distance measurement. Alternatively, the measurement of at least one electrical property can be performed without a distance measurement.

[0089] The precision and significance of the characteristic value can be improved by additionally using electrical sensor data to determine the characteristic value of the storage unit 10.

[0090] Using the neural network, a defect can be classified based on distance sensor data and electrical sensor data. The precision of defect detection can also be improved by feeding both distance sensor and electrical sensor data to the neural network for classification.

[0091] Fig. 3 shows a system 200 for training the neural network. System 200 comprises the components of system 100 described with reference to Figs. 1 and 2, the description of which is not repeated. However, the components are also disclosed for system 200.

[0092] If it is determined, for example using the neural network, that the examined storage unit 10 has at least one defect, i.e., the storage unit is assessed as abnormal, the storage unit 10 can be examined. For this purpose, an examination unit 80 can be provided, which is configured to perform the examination. The examination of the storage unit 10 can be destructive or non-destructive. During the examination, the storage unit 10 can be inspected for the identified defect and, if applicable, for further defects.

[0093] Another neural network 70 can be trained based on the test results. Alternatively, the neural network used to identify the defect can be trained directly.

[0094] If the identified defect is confirmed in the investigation, the defect identification can be considered correct. The subsequent neural network 70, or directly the neural network used to identify the defect, can then be trained based on further identified defects.

[0095] If feature extraction is used by the feature extraction unit 40, the neural network and / or the further neural network 80 can be trained based on the extracted features of the distance sensor data and / or the electrical sensor data.

[0096] The invention enables the identification of abnormal lithium-ion battery cells at the end of the production process based on volumetric expansion, optionally in addition to electrical measurements, using a neural network, for example, a semi-supervised neural network. Defective storage units can be sorted out, or less powerful storage units can be selected for less demanding applications.

Claims

Patent claims 1. Method for determining a characteristic value of a storage unit for electrical energy (10), the method comprising the steps: - Acquisition of distance sensor data by means of at least one distance sensor (20), wherein the distance sensor data are based on at least one distance measurement between the distance sensor (20) and the electrical energy storage unit (10), and - Determining the characteristic value of the electrical energy storage unit (10) based on the distance sensor data and using a neural network.

2. Method according to claim 1, wherein the characteristic value contains or enables information about a quality of the electrical energy storage unit (10), information about an estimated service life of the electrical energy storage unit (10), information about an estimated number of charge and / or discharge cycles of the electrical energy storage unit (10), information about a usage type of the electrical energy storage unit (10), information about a state of the electrical energy storage unit (10) and / or information about a defect of the electrical energy storage unit (10).

3. Method according to one of the preceding claims, wherein the storage unit for electrical energy (10) is a battery, in particular a lithium-ion battery, or is a part thereof.

4. Method according to one of the preceding claims, wherein the distance sensor (20) comprises an optical distance sensor, in particular a laser triangulation sensor, a lidar sensor, an optical coherence tomograph and / or a chromatic confocal sensor.

5. Method according to one of the preceding claims, wherein the distance sensor data are acquired by means of at least two distance sensors (20), preferably by means of at least three distance sensors (20).

6. Method according to one of the preceding claims, wherein the distance sensor data are recorded over a period of at least 1 min, preferably at least 1 h.

7. Method according to one of the preceding claims, wherein the distance sensor data are based on level measurements at at least two different locations or positions, preferably at least at three different locations or positions, of the electrical energy storage unit (10).

8. Method according to any of the preceding claims, wherein the neural network is a trained neural network, in particular wherein the neural network is a semi-supervised trained neural network.

9. Method according to one of the preceding claims, wherein at least one distance measurement is carried out during a charging and / or discharging process of the storage unit for electrical energy (10).

10. Method according to one of the preceding claims, wherein at least one distance measurement is carried out at the end of the production process of the electrical energy storage unit (10).

11. Method according to one of the preceding claims, wherein the distance sensor data are fed to the neural network as raw data, or wherein feature extraction is applied to the distance sensor data.

12. A method according to any one of the preceding claims, wherein the method further comprises: - Acquisition of electrical sensor data by means of at least one electrical sensor (60), wherein the electrical sensor data are based on at least one measurement of at least one electrical property of the electrical energy storage unit (10), and - Determining the characteristic value of the electrical energy storage unit (10) based on the distance sensor data and the electrical sensor data, and using the neural network.

13. Method according to claim 12, wherein the at least one measurement of the at least one electrical property is carried out during a charging and / or discharging process of the electrical energy storage unit (10), wherein the at least one measurement of the at least one electrical property is carried out at the end of the production process of the electrical energy storage unit (10) and / or wherein the at least one measurement of the at least one electrical property and the at least one distance measurement are carried out simultaneously.

14. Method according to one of the preceding claims, wherein, in the distance measurement, the distance between the distance sensor and the storage unit for electrical energy is determined directly or indirectly.

15. System (100) for determining a characteristic value of a storage unit for electrical energy (10), wherein the system (100) comprises: - at least one distance sensor (20) configured to acquire distance sensor data, wherein the distance sensor data is based on at least one distance measurement between the distance sensor (20) and the electrical energy storage unit (10), and - a characteristic value evaluation unit (50) which is set up to determine the characteristic value of the electrical energy storage unit (10) based on the distance sensor data and using a neural network.

Images