A battery cell testing system and method
The non-contact solar cell testing system uses a light-emitting module and a data acquisition module to acquire image data, and combines this with a control module to generate corresponding instructions. This solves the problem of damage to solar cells caused by contact testing and achieves high-precision solar cell quality testing.
Patent Information
- Application Number
- CN202410532138.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-29
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-04-29
AI Technical Summary
In existing solar cell testing technologies, contact testing may damage the solar cells, and the differences between solar cells may result in insufficient image clarity, affecting the testing quality.
A non-contact solar cell testing system is adopted, which provides a light source through a light-emitting module, acquires image data through a data acquisition module, and uses a control module to determine the detection results based on the image data, and generates drive commands, illumination adjustment commands and conveyor cycle adjustment commands to achieve accurate detection of solar cells.
This reduces secondary damage to the solar cells during testing, improves the accuracy and reliability of testing, and ensures the accuracy and efficiency of solar cell quality testing.
Smart Images

Figure CN118399889B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of battery testing technology, and in particular to a battery cell testing system and method. Background Art
[0002] As solar cells play an increasingly important role in clean energy and climate change, the requirements for their performance are also getting higher and higher. The defects of solar cells caused in the production process, such as broken grid, leakage, black core, missing corners, cracks, cold solder joints, etc., need to be detected to ensure the quality of the solar cells.
[0003] Patent CN212568513U discloses an EL test device, i.e. an electroluminescent device, which is provided with a hydraulic cylinder to drive the glass plate to descend, so that the pressing device presses the battery cell onto the electrical module on the probe to prevent the battery cell from being inaccurately detected due to false contact. The servo motor drives the screw rod to rotate, so that the camera rotates with the screw rod through the moving block, and can observe the battery cell in multiple positions. However, EL detection is a contact detection, which may cause unpredictable damage to the battery cell, and when performing EL detection, due to the differences in the battery cells themselves, the clarity of the detection image sometimes does not meet the requirements, affecting the detection quality.
[0004] Therefore, a battery cell testing system and method are provided to better detect the quality of the battery cells. Summary of the invention
[0005] One of the embodiments of the present specification provides a cell testing system, including a light-emitting module, a cell conveying device, a data acquisition module, a detection device, a driving component and an analysis and control module; wherein the light-emitting module is configured to provide lighting for the cell in the detection area based on the light-emitting component; the conveying device includes a fixing component, which is configured to convey the cell to the monitoring area based on a preset detection rate, and convey the cell with different detection results to different areas to be processed based on a conveying instruction, and the fixing component is used to fix the cell; the data acquisition module is configured to obtain image data of the cell based on the detection device in response to the light-emitting module being in a working state ; The driving component includes a translation component and a rotation component, and is configured to control the position of the detection device based on a driving instruction; the control module includes a processor, and the processor is configured to: determine a first detection result of the battery cell based on the image data; based on the first detection result, perform at least one of the following: generate the driving instruction, and send the driving instruction to the driving component to control the driving component to adjust the position of the battery cell; generate a control instruction, and send the control instruction to the conveying device to adjust the conveying rhythm of the conveying device; generate a lighting adjustment instruction, and send the lighting adjustment instruction to the light-emitting module to adjust the light-emitting parameters of the light-emitting module.
[0006] One of the embodiments of the present specification provides a battery cell testing method, characterized in that the method is executed based on a processor of a control module, including: determining a first detection result of the battery cell based on the image data; executing at least one of the following based on the first detection result: generating the drive instruction, and sending the drive instruction to the drive component to control the drive component to adjust the orientation of the battery cell; generating a control instruction, and sending the control instruction to the conveying device to adjust the conveying rhythm of the conveying device; generating a lighting adjustment instruction, and sending the lighting adjustment instruction to the light-emitting module to adjust the light-emitting parameters of the light-emitting module. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:
[0008] Figure 1 is an exemplary module diagram of a battery cell testing system according to some embodiments of this specification;
[0009] Figure 2 An exemplary flow chart of a cell testing method according to some embodiments of this specification;
[0010] Figure 3 is a schematic diagram of performing a re-inspection according to some embodiments of this specification;
[0011] Figure 4 is a schematic diagram of determining a first detection result according to some embodiments of this specification;
[0012] Figure 5 is a schematic diagram of determining a driving instruction according to some embodiments of this specification;
[0013] Figure 6 It is a schematic diagram of determining control instructions according to some embodiments of this specification. DETAILED DESCRIPTION
[0014] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.
[0015] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.
[0016] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.
[0017] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.
[0018] The quality of newly produced solar cells needs to be inspected. CN212568513U uses an EL contact detection device to detect the quality of the cells. Contact detection may cause damage to the cells, and the camera in the device moves axially following the rotation of the lead screw. It can only control the detection angle but not the detection distance. Due to the differences in the cells themselves, the shooting distance required to obtain clear image data is different, so a fixed shooting distance may result in the inability to obtain a clear image of the cell.
[0019] Therefore, in some embodiments of the present specification, the battery cell testing system determines a first detection result through a control module based on image data obtained by a detection device, generates a drive instruction that can control the orientation of the detection device according to the first detection result, generates a control instruction for adjusting the conveying rhythm, and generates a lighting adjustment instruction for adjusting the light-emitting parameters, thereby reducing secondary damage to the battery cell during detection and more accurately detecting the quality of the battery cell.
[0020] Figure 1 is an exemplary schematic diagram of a battery cell testing system according to some embodiments of this specification. Figure 1 As shown, the cell testing system 100 includes a light emitting module 110 , a cell conveying device 120 , a data acquisition module 130 , a detection device 140 , a driving assembly 150 and a control module 160 .
[0021] The light-emitting module 110 is configured to provide lighting for the battery cells in the detection area based on the light-emitting component. The light-emitting component refers to an electronic device that can emit light. For example, the light-emitting component can be a light source, an optical element, or other light-emitting device. The detection area refers to a specific area where the battery cells are detected. The light-emitting module 110 is connected in communication with the data acquisition module 130, and the light-emitting module transmits the working status information of the light-emitting module 110 to the data acquisition module 130. The light-emitting module 110 is also connected in communication with the control module 160, and the control module 160 transmits the lighting adjustment instruction to the light-emitting module 110.
[0022] The cell conveying device 120 includes a fixing assembly 121. The cell conveying device 120 is configured to convey cells to a detection area based on a preset detection rate, and to convey cells with different detection results to different areas to be processed based on a conveying instruction. The cell conveying device 120 is in communication connection with the control module 160, and the control module 160 transmits the control instruction to the cell conveying device 120.
[0023] The fixing component 121 is used to fix the battery cell. For example, the fixing component 121 can be a clamp, a bracket, a back plate, etc.
[0024] The preset detection rate refers to the expected detection rate of the battery cells in the detection area initially set by the battery cell test system. The preset detection rate can guide the speed of battery cell conveyance to ensure that the battery cell test system can process the battery cells at the expected rate. In some embodiments, the preset detection frequency can be determined in advance based on production requirements, equipment performance, and other factors.
[0025] In some embodiments, the preset detection rate is related to the distribution of the first detection results. For example, if the proportion of "may have defects" and "doubtful" in the first detection results of the battery cells in the recent period is greater than a preset threshold, the preset detection rate can be reduced. The preset threshold can be determined based on historical experience. For more information about the first detection results, please refer to the control module 160 and its related description.
[0026] The transport instruction refers to an instruction to transport battery cells with different test results to different areas to be processed.
[0027] The processing area refers to the area used to perform different processing on the battery cells. For example, the processing area may include the normal production area, the re-inspection area, and the recycling area. Among them, the normal production area refers to the area where the battery cells that meet the specifications and can continue to be produced or used in the next step are processed. The re-inspection area refers to the area where the battery cells that may need to be verified again are processed. The recycling area refers to the area where the battery cells that are judged to be unqualified or need further processing are processed.
[0028] The data acquisition module 130 is configured to obtain the image data of the battery cell based on the detection device 140 in response to the light emitting module 110 being in the working state. The image data refers to the data obtained after the detection device 140 photographs the battery cell. The data acquisition module 130 is connected in communication with the light emitting module 110, and the data acquisition module 130 receives the working state information of the light emitting module 110. The data acquisition module 130 is also connected in communication with the detection device 140, and the data acquisition module 130 collects the image data of the battery cell photographed by the detection device 140. The data acquisition module 130 is also connected in communication with the control module 160, and the data acquisition module 130 transmits the image data of the battery cell to the control module 160.
[0029] The detection device 140 refers to a device used to obtain information about the characteristics or quality of the battery cell. For example, the detection device can take a full-scale photo of the battery cell under illumination to obtain image data. For example, the detection device 140 can be an infrared camera, an optical camera, etc. The detection device 140 is connected to the data acquisition module 130 for communication, and the detection device 140 transmits the image data of the battery cell taken to the data acquisition module 130. The detection device 140 is mechanically connected to the drive assembly 150.
[0030] The drive assembly 150 includes a translation assembly 151 and a rotation assembly 152. The drive assembly 150 is configured to control the orientation of the detection device based on the drive instruction. The drive assembly 150 is in communication connection with the control module 160, and the drive assembly 150 receives the drive instruction transmitted by the control module 160. The drive assembly 150 is mechanically connected to the detection device 140.
[0031] The translation assembly 151 refers to a component used to translate the detection device. For example, the translation assembly 151 may include a slider, a guide rail, or a conveyor belt.
[0032] The rotating assembly 152 refers to a component used to rotate the detection device. For example, the rotating assembly 152 may include a rotating shaft, a motor, or a gyroscope.
[0033] The synergistic effect of the driving component 150 and the detection device 140 can comprehensively and flexibly detect the battery cell, ensuring that the detection device can obtain the required data on the entire battery cell.
[0034] The control module 160 includes a processor, which is configured to: determine a first detection result of the battery cell based on the image data; based on the first detection result, perform at least one of the following: generate a drive instruction, and send the drive instruction to the drive component to control the drive component to adjust the orientation of the battery cell; generate a control instruction, and send the control instruction to the conveying device to adjust the conveying rhythm of the conveying device; generate a lighting adjustment instruction, and send the lighting adjustment instruction to the light-emitting module to adjust the light-emitting parameters of the light-emitting module.
[0035] The control module 160 is connected in communication with the light-emitting module 110, and the control module 160 transmits the lighting adjustment instruction to the light-emitting module 110. The control module 160 is also connected in communication with the battery cell conveying device 120, and the control module 160 transmits the control instruction to the battery cell conveying device 120. The control module 160 is also connected in communication with the data acquisition module 130, and the control module 160 receives the image data of the battery cell transmitted by the data acquisition module 130. The control module 160 is also connected in communication with the driving component 150, and the control module 160 transmits the driving instruction to the driving component 150.
[0036] The battery cell testing system provides light through a light-emitting module, which can provide light without contacting the battery cell, avoiding the need to contact the battery cell during EL testing, thereby generating electroluminescence and reducing secondary damage to the battery cell during testing; through the control module, based on the image data obtained by the detection device, a first detection result is determined, and a drive instruction that can control the orientation of the detection device is generated according to the first detection result, as well as a control instruction for adjusting the conveying rhythm and a light adjustment instruction for adjusting the light-emitting parameters. According to the first detection result, for questionable battery cells, the drive component can be controlled to translate or rotate according to the drive instruction, and other relevant parameters can be adjusted accordingly, so as to perform a secondary inspection on the battery cell and obtain clearer image data, thereby being able to more accurately detect the quality of the battery cell.
[0037] Figure 2 is an exemplary flow chart of a battery cell testing method according to some embodiments of this specification. Figure 2 As shown, the process 200 may include the following steps. In some embodiments, the process 200 may be executed by a processor included in the control module.
[0038] Step 210: Determine a first detection result of the battery cell based on the image data.
[0039] The first test result refers to the preliminary test result obtained by testing the battery cell.
[0040] In some embodiments, the first test result can classify the battery cells into different types. For example, the tested battery cells can be classified into: normal type, significant defect type, uncertain type, or questionable type.
[0041] Among them, the normal class indicates that the image data corresponding to the battery cell is clearly visible and no abnormalities are found; the significant defect class indicates that the image data corresponding to the battery cell is clearly visible and there are obvious abnormalities, defects or flaws; the uncertain class indicates that the image data corresponding to the battery cell is clearly visible, but it is impossible to accurately determine whether the current battery cell has defects through analysis; the questionable class indicates that the image data corresponding to the battery cell is abnormal (for example, the image is blurred, missing, etc.), and the battery cell cannot be analyzed based on the current image data.
[0042] The processor can determine the first detection result of the battery cell in a variety of ways. In some embodiments, the processor can preprocess the image data of the battery cell, perform image recognition on the image obtained after the preprocessing, and determine the first detection result by analyzing and processing the result of the image recognition.
[0043] The preprocessing may include adjusting the size or resolution of the image, and may also include reducing the noise of the image, and adjusting parameters such as contrast and brightness.
[0044] Image recognition can be implemented based on methods such as Histogram of Oriented Gradient (HOG). By performing image recognition on the image data obtained after preprocessing, feature information related to the cell defect can be obtained. In some embodiments, the feature information of the cell defect may include the texture, color, shape, etc. of the image data.
[0045] In some embodiments, the processor may analyze and process the characteristic information of the battery cell defect to determine whether the battery cell has a defect, thereby determining the first detection result. Among them, a classifier or a convolutional neural network model may be used for judgment, and the classifier or convolutional neural network model may be obtained through training.
[0046] In some embodiments, the processor may also determine the first detection result of the battery cell based on the image data of the battery cell through the image detection model. For more details, please refer to Figure 4 and related instructions.
[0047] In some embodiments, the processor may determine whether the battery cell testing system needs to be adjusted based on the first detection result. When it is determined that adjustment is needed, the processor may execute at least one of step 220, step 230 and step 240.
[0048] In some embodiments, the processor may collect statistics on the first detection results of the battery cells, and collect statistics on the proportion of battery cells with uncertain or questionable first detection results, and when the proportion is greater than a preset adjustment threshold, it is determined that adjustment is required. The preset adjustment threshold may be determined based on prior experience or historical data.
[0049] Step 220, generating a driving instruction, and sending the driving instruction to the driving component to control the driving component to adjust the orientation of the battery cell.
[0050] The driving instruction is an instruction for adjusting the orientation of the driving component. In some embodiments, the driving instruction may include at least three-dimensional coordinate information and inclination information. The three-dimensional coordinate information may represent the spatial position of the battery cell, which may be represented by mathematical methods such as spherical coordinates and three-dimensional coordinates; the inclination information may represent the degree of inclination of the battery cell with respect to a certain axis or plane, which may be represented by mathematical methods such as Euler angles and quaternions.
[0051] In some embodiments, when the defects on the battery cell are mainly concentrated in a certain area, the processor can adjust the orientation of the driving component based on the location of the defect to more centrally detect the area. Exemplarily, the processor can generate a driving instruction to adjust the orientation of the detection device so that the imaging center of the detection device is exactly perpendicular to the defect concentration area on the battery cell, and appropriately shorten the distance between the detection device and the battery cell, so as to achieve more centralized detection of a part of the battery cell.
[0052] In some embodiments, the processor may generate a driving instruction based on a first preset table. The processor may construct a first preset table based on a historical first detection result and a historical driving instruction. The first preset table includes a correspondence between different first detection results and different historical driving instructions. The processor may determine a current driving instruction by querying the first preset table based on the current first detection result.
[0053] In some embodiments, the processor may also generate candidate driving instructions based on at least one target area and defect distribution characteristics of the battery cell, and determine the driving instructions from the candidate driving instructions through the orientation effect model. For more details, please refer to this specification Figure 5 Description of the species.
[0054] Step 230: Generate a control instruction and send the control instruction to the conveying device to adjust the conveying rhythm of the conveying device.
[0055] The control instruction refers to an instruction for adjusting the working state of the conveying device. In some embodiments, the control instruction may include adjusting the conveying rhythm of the conveying device.
[0056] In some embodiments, the conveying rhythm can be represented by the conveying speed of the conveying device.
[0057] In some embodiments, the conveying rhythm can be adjusted according to the actual detection situation. For example, when the detection device needs to adjust the position or perform a certain operation, the processor can control the conveying device to slow down the conveying speed of the battery sheet to ensure that the battery sheet stays in the detection area for a sufficient time.
[0058] In some embodiments, the processor may generate control instructions in a variety of ways. For example, the conveying speed of the conveying device may be negatively correlated to the proportion of battery cells of the type of the first detection result being a questionable type or a misdetected type. When the proportion of battery cells of the type of the first detection result being an uncertain type or a questionable type is too high, it may be because the conveying speed of the conveying device is too fast, resulting in insufficient time to collect image data. In this case, the conveying speed of the conveying device may be appropriately reduced.
[0059] In some embodiments, the processor may also determine a control instruction based on the second detection result. For more details, please refer to this specification. Figure 6 See the relevant instructions in .
[0060] Step 240: Generate a light adjustment instruction, and send the light adjustment instruction to the light emitting module to adjust the light emitting parameters of the light emitting module.
[0061] The illumination adjustment instruction refers to a control parameter used to adjust the illumination parameters of the illumination module. In some embodiments, the illumination parameters may at least include illumination intensity, and the illumination adjustment instruction may at least include adjusting the illumination intensity.
[0062] In some embodiments, the processor may analyze the image data of the cell to obtain the brightness of the image data. When the brightness of the image data is greater than the maximum brightness threshold, the light adjustment instruction may be to reduce the light intensity; when the brightness of the image data is less than the minimum brightness threshold, the light adjustment instruction may be to increase the light intensity. The adjustment amplitude of the light intensity may be positively correlated to the difference between the brightness and the maximum / minimum brightness threshold. For example, when the brightness of the image data is higher than the maximum brightness threshold, the greater the brightness of the image data, the greater the corresponding adjustment amplitude.
[0063] In some embodiments of the present specification, based on the analysis of the first detection result, when the first detection result is abnormal, the abnormality is analyzed in time, and the modules in the battery cell testing system are adjusted, which can effectively improve the accuracy and reliability of the battery cell test results, reduce the low detection credibility due to equipment reasons, and reduce the situation where multiple tests are required to determine whether there are defects.
[0064] Figure 3 It is an exemplary schematic diagram of re-examination according to some embodiments of the present specification.
[0065] In some embodiments, in response to the number of battery cells to be re-inspected 310 in the re-inspection area reaching a preset threshold 320, the processor can generate a re-inspection instruction 330, and the re-inspection instruction 330 is used to control the detection device 140 to re-inspect the battery cells to be re-inspected. The processor can also generate an adjustment instruction 340, and the adjustment instruction 340 is used to adjust the conveying rhythm of the battery cell conveying device 120.
[0066] Re-inspection refers to the system performing additional inspection or verification on some cells after the cells have been initially inspected. The re-inspection includes adjusting the inspection surface of the cells to be re-inspected.
[0067] A battery cell that needs additional testing or verification is a battery cell to be retested. When the first test result is uncertain or questionable, the battery cell can be determined as a battery cell to be retested.
[0068] The number of battery cells to be re-inspected refers to the number of battery cells to be re-inspected in the re-inspection area.
[0069] For more information about the re-examination area, see Figure 1 and its related description.
[0070] The preset threshold value 320 refers to the pre-set boundary value of the number of cells to be re-inspected. If the number of cells to be re-inspected 310 in the re-inspection area does not reach the preset threshold value 320, the cells to be re-inspected in the re-inspection area do not need to be re-inspected, that is, the control module does not need to control or adjust the parameters of the drive component, the light-emitting module, and the cell conveying device. In this way, the re-inspection area can accumulate cells to the preset threshold value 320 before the cells to be re-inspected in the re-inspection area are re-inspected, which can optimize the allocation of resources.
[0071] When the number of cells to be re-inspected 310 in the re-inspection area reaches a preset threshold 320, the re-inspection instruction 330 generated by the processor will be converted into a drive instruction, and the drive component adjusts the orientation of the detection device based on the drive instruction. The adjustment instruction 340 generated by the processor will be converted into a control instruction, and the cell conveying device adjusts the conveying rhythm of the cell based on the control instruction. The processor will also generate a lighting adjustment instruction, and the light-emitting module adjusts the light-emitting parameters of the light-emitting module based on the lighting adjustment instruction.
[0072] In some embodiments, the conveying rhythm may be adjusted so that during re-inspection, the battery cell conveying device stops conveying new battery cells so that the system can focus on processing battery cells that need to be re-inspected.
[0073] In some embodiments, the preset threshold 320 may be determined based on a historical threshold.
[0074] For example, historical thresholds are counted and the preset threshold is obtained by calculating the average of the historical thresholds.
[0075] For another example, suppose that a certain historical threshold has remained unchanged for a period of time in the past few months, and under this threshold, the quality control of the cell test system has performed well, and there are no preset number of defective or unqualified cells. The preset number can be set based on experience. Therefore, based on past experience and the consideration of maintaining system stability, the preset threshold continues to use the historical threshold. This can maintain the consistency of the system under similar conditions and reduce interference with the production process.
[0076] In some embodiments, the preset threshold 320 can be determined based on the mean and standard deviation of the historical thresholds, weighted according to a preset weight; the preset weight is determined based on the consistency between the first test result and the second test result of the battery cell to be reviewed in the historical production data.
[0077] For more information about the first test result, please refer to this manual. Figure 4 For more information about the second test, please refer to this manual. Figure 6 Related description in .
[0078] As an example only, the preset threshold 320 can be determined by calculation using formula (1):
[0079] Q=k1*n+k2*m (1)
[0080] Wherein, Q is the preset threshold, k1 is the preset weight corresponding to the mean of the historical threshold, n is the mean of the historical threshold, k2 is the preset weight corresponding to the standard deviation of the historical threshold, m is the standard deviation of the historical threshold, k1 and k2 are related to the consistency between the first detection result and the second detection result of multiple battery cells in the historical production data, the greater the consistency between the first detection result and the second detection result, the larger k1 and the smaller k2, wherein k1 is 1 at most and k2 is 0 at least. Wherein, multiple battery cells refer to battery cells of the same production batch or the same time period in the historical production batch, and the aforementioned multiple battery cells use at least one historical threshold; the consistency refers to the similarity or consistency between the first detection result and the second detection result.
[0081] In some embodiments, the consistency between the first test result and the second test result can be represented by a Kappa coefficient.
[0082] The Kappa coefficient is a statistical indicator used to compare the consistency of data among multiple groups.
[0083] In some embodiments, the Kappa coefficient can be obtained based on the observed accuracy and the expected accuracy, wherein the observed accuracy refers to the actual consistency between the first test result and the second test result, and the expected accuracy refers to the expected consistency between the first test result and the second test result.
[0084] In some embodiments, the Kappa coefficient value ranges from 0 to 1, where 0 indicates complete inconsistency and 1 indicates complete consistency. The larger the Kappa coefficient value, the higher the consistency between the first test result and the second test result of the battery cell.
[0085] As an example only, the Kappa coefficient can be obtained by formula (2):
[0086] Kappa = (Po - Pe) / (1 - Pe) (2)
[0087] Among them, Kappa is the Kappa coefficient, Po is the observed accuracy, and Pe is the expected accuracy.
[0088] In some embodiments, the observed accuracy rate may be determined based on the ratio of the number of battery cells for which the first detection result and the second detection result are consistent to the total number of battery cells.
[0089] As an example only, the observed accuracy can be calculated using formula (3):
[0090] Po= (A+C+D) / (A+B+C+D) (3)
[0091] The expected accuracy can be calculated using formula (4):
[0092]
[0093] Among them, A means that the first test result is that there may be defects, and the second test result is the number of battery cells with obvious defects; B means that the first test result is that there may be defects, but the second test result is the number of normal battery cells; C means that the first test result is questionable, and the second test result is that there are obvious defects; D means that the first test result is questionable, and the second test result is the number of normal battery cells.
[0094] Among them, when the first test result is questionable, it includes two situations: possible defects and possible normal conditions. At this time, as long as the second test result is defective or normal, the first test result and the second test result can be considered to be consistent.
[0095] For more information about the first test result, please refer to Figure 1 For details about the second test result, please refer to Figure 6 and its related description.
[0096] When the number of cells to be re-inspected reaches a preset threshold, the processor performs re-inspection processing, which optimizes resources while also paying closer attention to areas where problems may exist, thereby improving the accuracy and comprehensiveness of detection and ensuring that the system can more reliably determine the quality and performance of the cells.
[0097] Figure 4 It is an exemplary schematic diagram of determining the first detection result according to some embodiments of this specification.
[0098] In some embodiments, the processor can determine regional features 430 based on the image data 410 through the image detection model 420, and the regional features include defect probabilities 431 and corresponding first confidence levels 432 of multiple sub-regions in the battery cell; based on the defect probabilities 431 and the first confidence levels 432, determine at least one target region 440 in the battery cell; based on the at least one target region 440, determine a first detection result 450 of the battery cell.
[0099] A sub-region refers to a partial region of the cell image data.
[0100] In some embodiments, the sub-regions may be obtained by dividing the battery cell based on a preset rule. In some embodiments, the preset rule may be dividing the battery cell into n sub-regions (sub-regions 1, 2, ..., n) according to an equidistant grid.
[0101] The regional features are used to characterize the features of each sub-region, including the defect probability of the sub-region and the confidence of the defect probability.
[0102] The defect probability refers to the probability that a sub-region may have a defect. For example, the defect probability of each sub-region may be a value between 0 and 100%. The higher the defect probability, the greater the possibility that a sub-region may have a defect.
[0103] The first confidence refers to the credibility of the defect probability. For example, the first confidence of each sub-region can be a value between 0 and 100%. The higher the first confidence, the greater the reliability of the defect probability.
[0104] In some embodiments, the processor may determine the regional features 430 of the sub-regions in the cell by using the image detection model 420 based on the image data 410 of the cell.
[0105] In some embodiments, the image detection model is a machine learning model. For example, the image detection model may be a convolutional neural network (CNN), or other models that can achieve the same or corresponding functions.
[0106] In some embodiments, the input of the image detection model includes image data of the battery cell, and the output is a region feature.
[0107] In some embodiments, the image detection model can be obtained by training the initial detection model through a plurality of first training samples with first labels. For example, the first training sample is input into the initial detection model, a loss function is constructed based on the output of the initial detection model and the training label, and the parameters of the initial detection model are iteratively updated based on the loss function through gradient descent or other methods until the iteration is terminated when the training end condition is met, thereby obtaining a trained image detection model.
[0108] Among them, the training end conditions may include the convergence of the loss function, the number of iterations reaching a threshold, etc.
[0109] In some embodiments, the first training samples used to train the image detection model include positive samples and negative samples, wherein the positive samples refer to samples whose defect condition judgment results are consistent with the actual defect condition, and the negative samples refer to samples whose defect condition judgment results are inconsistent with the actual defect condition.
[0110] In some embodiments, the first confidence label corresponding to the positive sample is 1, and the first confidence label corresponding to the negative sample is 0; if there are actually defects in the sample sub-region corresponding to the training sample, the label of the defect probability corresponding to the training sample is 1, if there are actually no defects, the label of the defect probability corresponding to the training sample is 0.
[0111] In some embodiments, the ratio of the number of positive samples to the number of negative samples is determined based on the distribution of target regions in the positive samples.
[0112] The target region refers to a sub-region with a higher defect probability. For example, a sub-region with a defect probability greater than a defect threshold is a target region.
[0113] The defect threshold is a threshold used to determine whether a sub-region is a target region, and may include a maximum allowable value of a defect probability, such as 0.6, etc. When the defect probability of a sub-region is greater than the defect threshold, the sub-region may be considered a target region. In some embodiments, the defect threshold may be determined based on historical experience or actual production requirements.
[0114] As an example, the more dispersed the target area in the positive sample is, the greater the proportion of negative samples can be. At this time, the data distribution of the training set is more complex. Adding more negative samples can enable the image detection model to better learn the confidence. Among them, the distribution of the target area can be calculated based on the coordinates of the target area.
[0115] In some embodiments, the processor can analyze the battery cell detection area through historical data, and determine the number of positive samples based on the analysis results. For example, the number of positive samples can be determined based on the defect frequency in the battery cell detection area. The higher the defect frequency, the greater the number of positive samples, to ensure that the positive samples can effectively cover various defect types, so that the model can fully learn the characteristics of different defects.
[0116] In some embodiments, the processor may process some positive samples to obtain negative samples and count the number of the negative samples. The types of processing may include but are not limited to adding Gaussian blur, adding noise, etc. The above processing can help the model learn more different scenarios and situations, thereby improving the generalization ability of the model, making the model better adaptable to different situations, and enhancing the robustness of the model.
[0117] In some embodiments, the processor calculates the ratio of the number of positive and negative samples based on the aforementioned number of positive samples and number of negative samples; evaluates the performance of the image detection model through cross-validation, analyzes the performance of the image detection model under different ratios of the number of positive and negative samples based on the evaluation results, and gradually adjusts the ratio of the number of positive and negative samples, and judges the model performance based on the indicators, and determines the ratio of the number of positive and negative samples that makes the model performance best as the optimal ratio of the number of positive and negative samples. The indicators may include but are not limited to accuracy, recall rate, F1 score, etc.
[0118] In some embodiments, the processor may adjust the ratio of the number of positive and negative samples based on a preset adjustment step, for example, each adjustment range does not exceed ±10%, etc. The preset adjustment step may be determined based on historical experience or actual needs.
[0119] By setting the optimal ratio of the number of positive and negative sample sub-regions and training the image detection model, the model can be helped to better learn the characteristics and background information of the target object, improve the accuracy and generalization ability of detection, thereby improving the accuracy of regional feature recognition, and further improving the accuracy of the first detection result.
[0120] In some embodiments, the processor may determine at least one target region 440 according to the defect probability 431 of each sub-region and the first confidence 432. For example, the processor may determine whether the defect probability and the first confidence of each sub-region meet a preset judgment condition, and if so, the sub-region is the target region.
[0121] In some embodiments, the preset judgment condition may include a preset judgment threshold. For example, the preset judgment condition may include that the defect probability is greater than the probability threshold, and the first confidence level is greater than the confidence threshold.
[0122] In some embodiments, the preset judgment threshold may be set by system default or based on historical experience.
[0123] In some embodiments, the processor may determine the first detection result 450 of the cell according to at least one target region 440 in the cell. For example, the processor may determine the first detection result of the cell according to the ratio of the number of target regions to the number of sub-regions.
[0124] As an example only, if the proportion of the number of target areas in the battery cell to the number of sub-areas is greater than 95%, the first inspection result of the battery cell is that there are obvious defects; if the proportion of the number of target areas in the battery cell to the number of sub-areas is 50% to 95%, the first inspection result of the battery cell is that there may be defects; if the proportion of the number of target areas in the battery cell to the number of sub-areas is 1% to 50%, the first inspection result of the battery cell is questionable; if the proportion of the number of target areas in the battery cell to the number of sub-areas is 0%, the first inspection result of the battery cell is normal.
[0125] The setting of the threshold values of each interval is only exemplary and can be set based on historical data or experience.
[0126] By judging the defects of battery cells through the image detection model, the battery cells can be quickly classified and a preliminary first detection result can be obtained to provide a basis for the subsequent second detection, thereby improving the efficiency and accuracy of battery cell quality detection.
[0127] Figure 5 This is a schematic diagram of determining a driving instruction schematic diagram according to some embodiments of this specification.
[0128] In some embodiments, the processor can determine the defect distribution characteristics 520 of the battery cell based on at least one target area 510 in the battery cell; generate a candidate driving instruction 550 based on at least one target area 510 and the defect distribution characteristics 520; determine the estimated effect 580 of the candidate driving instruction 550 through an orientation effect model 570; and in response to the estimated effect of the candidate driving instruction meeting a preset condition, determine the candidate driving instruction as a target driving instruction 590.
[0129] The defect distribution feature can reflect the distribution of various defects at different locations on the battery cell. In some embodiments, the defect distribution feature can include the defect density and the maximum likelihood value of the defect number.
[0130] Among them, the maximum likelihood value of the number of defects can reflect the number of defects, and the defect density can reflect the concentration of defect distribution. For example, the more defects there are, the greater the maximum likelihood value of the number of defects, and the fewer defects there are, the smaller the maximum likelihood value of the number of defects; the more uniform the defect distribution on the battery cell, the smaller the defect density, and the more concentrated the distribution, the greater the defect density.
[0131] In some embodiments, the defect concentration may be inversely related to the distance between defects, and the processor may determine an average distance between defects and determine the defect concentration based on the average distance.
[0132] In some embodiments, the processor can select any two target areas on the cell and calculate the distance between them. After the calculation is completed, the target area selection is repeated until all the selectable methods are traversed, and the distance corresponding to each selection method is calculated. The average value of the distances corresponding to all selection methods is calculated and the reciprocal is taken, and the obtained value is determined as the defect density.
[0133] In some embodiments, the processor may determine the maximum likelihood value of the defect count by clustering method.
[0134] In some embodiments, the target area can be represented by a target area feature vector. The elements in the target area feature vector may include at least one of the position coordinates and area of the target area. For example, the target area feature vector may be (A, B), where A represents the position of the target area and B represents the area of the target area.
[0135] In some embodiments, the processor may cluster the target region feature vectors corresponding to the battery cells through a clustering algorithm to determine multiple target region cluster sets. Each cluster set includes one or more target region feature vectors, and each cluster set has a corresponding cluster center.
[0136] In some embodiments, after clustering is completed, the processor may evaluate the effectiveness of the clustering and determine a clustering score, wherein the evaluation algorithm may include but is not limited to algorithms such as silhouette coefficient.
[0137] In some embodiments, the processor may perform multiple clustering by adjusting the type or parameters of the clustering algorithm, and evaluate the effects of the multiple clustering to obtain multiple clustering scores. The clustering algorithm may include K-Means clustering, density-based clustering method (DBSCAN), or other example algorithms; the parameters of the clustering algorithm may include a preset number of clusters, number of iterations, etc.
[0138] In some embodiments, after multiple clusterings are completed, the processor can sort based on the clustering scores, select the clustering result corresponding to the highest clustering score (best clustering effect), and determine the number of cluster centers of the clustering result as the maximum likelihood value of the number of defects.
[0139] The candidate driving instruction refers to a selectable driving instruction used to control the detection device to make adjustments.
[0140] In some embodiments, the parameters included in the candidate drive instruction are the same as the parameters included in the drive instruction. For more details, see Figure 2 Related description of the drive instructions.
[0141] In some embodiments, the processor may compare the defect distribution characteristics of the battery cell with the historical defect distribution characteristics of detected battery cells in historical data, select detected battery cells with a similarity level higher than a similarity threshold as sample battery cells, and obtain historical driving instructions for the sample battery cells as reference driving instructions.
[0142] In some embodiments, the processor may construct a feature vector based on the maximum likelihood value of the defect type, defect location, defect density and defect number of the cell, and construct a reference feature vector based on the corresponding features of the detected cell; determine the similarity between the cell and the detected cell based on the vector distance between the feature vector and the reference feature vector, and determine the detected cell with a similarity higher than a similarity threshold as a reference cell. The similarity threshold may be set based on prior experience or manual input.
[0143] In some embodiments, the processor may adjust the parameters included in the reference driving instruction, and randomly generate a large number of different candidate driving instructions by modifying the value range of the parameters.
[0144] In some embodiments, the processor may also generate candidate driving instructions based on algorithms such as a generative adversarial network model and a variational autoencoder.
[0145] In some embodiments, the processor may determine the estimated effect of the candidate driving instruction 550 through the orientation effect model 570. The estimated effect refers to the prediction result of whether the detection effect is improved after adjusting the orientation of the detection device based on the candidate driving instruction.
[0146] The position effect model refers to a prediction model used to evaluate the adjustment effect of the candidate driving instructions on the battery cell. In some embodiments, the position effect model can be a machine learning model, such as a deep neural network model (Deep-Learning Neural Network, DNN).
[0147] In some embodiments, the input of the orientation effect model may include at least one of the location information 530 of the target area in the battery cell, regional features 540, defect distribution features 520 and candidate drive instructions 550, and the output may include the estimated effect 580 of the candidate drive instructions 550.
[0148] The processor can mesh the surface of the cell based on the mesh parameters to obtain several mesh sub-regions and label them. The labels of the mesh sub-regions represent the location information of the target region. The mesh parameters can include the size and shape of the mesh. The mesh parameters can be determined based on prior experience. For more information about regional features, please refer to Figure 4and its related descriptions; for more information about defect distribution characteristics and candidate drive instructions, please refer to the related descriptions above.
[0149] In some embodiments, the input of the orientation effect model further includes first detection results 560 of a plurality of battery cells that have completed detection.
[0150] In some embodiments of the present specification, when there are obvious defects or flaws in a battery cell, or when there are areas in the battery cell that function normally and do not need to be inspected, the first inspection results of multiple battery cells that have completed inspection can be directly input into the orientation effect model to reduce the time spent on the model predicting known areas, reduce computing power consumption, and make the model more efficient.
[0151] In some embodiments, the processor may train the orientation effect model based on a large number of second training samples with second labels. In some embodiments, the second training samples may include sample position information of target areas in each sample battery cell, sample area characteristics, sample defect distribution characteristics, and sample drive instructions. In some embodiments, the second training samples also need to include sample first test results of multiple battery cells that have completed testing. The second training samples may be acquired based on historical data.
[0152] In some embodiments, the second label may be an actual effect corresponding to the sample driving instruction in the second training sample.
[0153] As an example only, the processor can obtain sample inspection results of multiple sample battery cells after adjustment based on sample drive instructions in historical production data, as well as actual verification results determined by manual review, and determine the inspection accuracy based on whether the sample estimated effect is consistent with the actual verification result.
[0154] When the detection accuracy is greater than or equal to the detection threshold, it means that the adjustment effect of the historical drive instruction is good and the defects on the battery cell can be fully and carefully detected. At this time, the first label is 1. When the detection accuracy fails to reach the detection threshold, it means that the adjustment effect of the historical drive instruction is poor, and the first label is 0. The detection accuracy can be the number of correct sample detections / the total number of sample detections*100%; the detection threshold can be preset based on prior experience, for example, 95%; for more information about the actual verification results, please refer to Figure 6 section and its related instructions.
[0155] In some embodiments, the processor can perform training based on the second training sample and the second label by various methods. For example, training can be performed based on the gradient descent method. As an example only, multiple second training samples with second labels can be input into the initial orientation effect model, a loss function can be constructed by the second label and the result of the initial orientation effect model, and the parameters of the initial orientation effect model can be iteratively updated based on the loss function. When the loss function of the initial orientation effect model meets the preset conditions, the model training is completed and a trained orientation effect model is obtained. Among them, the preset conditions can be that the loss function converges, the number of iterations reaches a threshold, etc.
[0156] The preset condition refers to a screening condition for determining a target drive instruction from candidate drive instructions. In some embodiments, the preset condition may be: selecting a candidate drive instruction with the highest detection accuracy and determining it as the target drive instruction. The highest detection accuracy may mean that the candidate drive instruction has the best estimated effect.
[0157] In some embodiments of the present specification, the adjustment effects of different candidate drive instructions are predicted in advance through an orientation effect model, and the candidate drive instruction with the best estimated effect is selected and determined as the target drive instruction, which can effectively improve the efficiency and accuracy of battery cell defect detection.
[0158] Figure 6 It is a schematic diagram of determining control instructions according to some embodiments of this specification.
[0159] In some embodiments, the driving instruction may include three-dimensional coordinate information 610 and inclination information 620 for controlling the detection device. In some embodiments, the processor may also adjust the orientation of the detection device 140 according to the three-dimensional coordinate information 610 and the inclination information 620, and obtain additional data 630 based on the detection device 140; determine a second detection result 640 through the control module based on the additional data 630; determine a control instruction 650 based on the second detection result 640; and adjust the configuration of the fixing component 121.
[0160] In some embodiments, the processor can adjust the orientation of the detection device in a variety of ways. For example, the processor can count the number of questionable areas in the target area. When the number exceeds a preset threshold, it indicates that the collected image data may be fuzzy. At this time, the processor can appropriately shorten the distance between the detection device and the battery cell and adjust the inclination of the detector so that the detection device can clearly capture the battery cell, thereby improving the quality of image data collection.
[0161] Additional data refers to extra detection data obtained after adjusting the orientation of the detection device.
[0162] In some embodiments, the additional data may be image data containing local details or different angles. In some embodiments, the specific composition form of the additional data may be the same as or different from the image data, depending on the actual situation.
[0163] The second test result refers to the test result obtained by further analyzing the doubtful area. For example, based on the additional data, the target area of the uncertain class or doubtful class in the first test result is analyzed to determine whether it belongs to the normal class or the significant defect class. The final result is the second test result.
[0164] In some embodiments, the processor may determine the second detection result in the same manner as determining the first detection result. For example, the processor may input the additional data and previously acquired image data into an image detection model, determine the regional features based on the image detection model, further determine the target region, and determine the second detection result based on the target region.
[0165] For more details about the image detection model, please refer to this manual Figure 4 and related instructions.
[0166] In some embodiments, when the second detection result shows that more than a preset number of battery cells have similar characteristic damage, it may be due to defects in the conveying method of the battery cell conveying device, resulting in damage to the battery cell during the process of conveying the battery cell and performing operations such as clamping and flipping. At this time, the processor can send control instructions such as reducing the clamping force of the fixing component, replacing the flexible protective material included in the fixing component, and reducing the conveying speed of the conveying device to avoid similar situations.
[0167] The configuration of the fixing component refers to the operating parameters of the fixing component. In some embodiments, the configuration of the fixing component may include the conveying speed of the conveying device, the clamping force of the fixing component, etc.
[0168] In some embodiments, the processor can adjust the configuration of the fixed component based on the control instruction. In some embodiments, the processor can also establish a second preset table based on historical control instructions, historical conveying speeds, and historical clamping forces. The second preset table includes the correspondence between historical control instructions and different historical conveying speeds and historical clamping forces. The processor can determine the current conveying speed and clamping force according to the current control instruction by consulting the second preset table.
[0169] In some embodiments of the present specification, based on the three-dimensional coordinate information and inclination information in the driving instructions, high-precision positioning and directional adjustment of the detection device can be achieved, thereby ensuring that the detection device can accurately align with the key parts or potential defect areas of the battery cell for detection, thereby improving the pertinence and accuracy of the re-inspection; based on the analysis of the second detection results, abnormal situations in which defects are caused by the battery cell during transportation and fixation can be discovered in time, and the system can be adjusted in time through control instructions to reduce the number of defective battery cells caused thereby.
[0170] In some embodiments, in response to the second detection result indicating that a defect exists, the processor may evaluate the cause of the defect in the battery cell using a preset algorithm based on the image data and the additional data.
[0171] The preset algorithm is an algorithm for determining the cause of the cell defect. In some embodiments, the processor can evaluate different grid sub-areas of the cell through the preset algorithm to determine whether the defective grid sub-area is generated during the inspection process or the production process. The processor can also perform a weighted summation on all grid sub-areas to comprehensively determine the cause of the cell defect.
[0172] In some embodiments, the cause of the cell defect can be determined based on the defect type value. When the defect type value is greater than the defect threshold, the defect cause can be determined to be the production process. When the defect type value is not greater than the defect threshold, the defect cause can be determined to be the detection process. The defect threshold can be determined based on historical experience or actual needs. In some embodiments, the defect threshold can be 50.
[0173] In some embodiments, the defect type value may be determined by weighting based on the number of defects in at least one sub-region of the cell. As an example only, the processor may determine the defect type value by formula (5):
[0174] Y=k1×d1+k2×d2+ … +k n ×d n (5)
[0175] Among them, Y represents the defect type value, k n represents the weight coefficient corresponding to the grid sub-region n, d n Indicates the number of defects corresponding to the grid sub-region n. When the defect type value Y>50, the cause of the defect is determined to be the production process, and when the defect type value Y≤50, the cause of the defect is determined to be the detection process. Weight coefficient k n The value of can be preset based on prior experience or manual input.
[0176] In some embodiments, in response to the production process being the cause, the battery cell is transported to a recycling area by a transport device. For more information about the recycling area, see Figure 1 and its related contents.
[0177] In some embodiments of the present specification, by determining the cause of defects in each grid of the battery cell separately and weighting them based on the number of defects, the cause of defects in the entire battery cell can be comprehensively determined, thereby reducing the misjudgment rate of the cause of defects and quickly determining the module or equipment that currently needs to be adjusted; by transporting defective battery cells in the production process to a recycling area, resources can be maximized and production costs can be reduced.
[0178] In some embodiments, in the preset algorithm, the weight coefficient of each sub-region is related to the distance between each sub-region and the fixed component.
[0179] In some embodiments, the weight coefficient of each sub-region is positively correlated with the distance between the sub-region and the fixed component. The closer the distance between the sub-region and the fixed component, the greater the weight coefficient corresponding to the sub-region.
[0180] For more information on fixing components, please refer to this manual. Figure 1 Related description in .
[0181] When the fixing component clamps the battery cell, different parts of the battery cell receive different pressures. Sub-areas closer to the fixing component tend to be under greater pressure and have a greater risk of damage than sub-areas farther away. Defects are more likely to be caused by the fixing component rather than the production process.
[0182] In some embodiments of the present specification, by making the weight coefficient of each sub-region related to the distance between each sub-region and the fixing component, the influence of the fixing component can be fully considered when determining the cause of the defect of the battery cell.
[0183] In some embodiments, the processor may also determine the reliability of the aforementioned preset method; and adjust the three-dimensional coordinate information in the driving instruction based on the reliability of the preset algorithm.
[0184] The reliability can reflect the reliability of the preset algorithm. In some embodiments, the reliability can be represented by the detection accuracy. For example, the reliability of the preset algorithm can be the average value of the reliability predicted by the preset algorithm for different batches of battery cells.
[0185] In some embodiments, the reliability of the preset algorithm can be determined based on the proportion of the correct results of identifying the causes of battery cell defects in the same batch to the total number of battery cells.
[0186] As an example only, the processor can determine the reliability of different batches of battery cells by formula (6):
[0187]
[0188] Among them, Z represents the reliability of the judgment results of a batch of preset algorithms, T p Indicates the number of correctly identified normal cells (true examples), T n represents the number of defective cells correctly identified (true negatives), F p Indicates the number of defective cells mistakenly identified as normal cells (false positives); F n Indicates the number of correct cells that are mistakenly identified as defective cells (false negatives).
[0189] Among them, the positive example indicates that the preset algorithm determines that it is caused by the detection process, and the negative example indicates that the preset algorithm determines that it is caused by the production process.
[0190] The truth or falsity of the judgment result can be determined based on the consistency between the judgment result of the preset algorithm and the actual verification result. The actual verification result is the cause of the battery cell defect obtained based on professional manual inspection. For a battery cell, if the judgment of the preset algorithm is consistent with the actual verification result, the sample is true; if the judgment of the preset algorithm is inconsistent with the actual verification result, the sample is false.
[0191] In some embodiments, the reliability of the preset algorithm can be determined based on the average reliability of the judgment results of at least one batch of battery cells by the preset algorithm.
[0192] In some embodiments, the processor can reversely adjust the three-dimensional coordinate information based on the reliability of the preset algorithm. For example, when the reliability of the preset algorithm is lower than the minimum reliability threshold, it means that the current image data and additional data may not meet the needs. At this time, the three-dimensional coordinate information of the detection device can be adjusted to increase the number of positions and / or angles for obtaining additional data. By obtaining additional data from multiple positions and / or angles, it is helpful to make a more accurate judgment on the situation of the battery cell.
[0193] In some embodiments, the minimum reliability threshold may be set based on historical experience or actual production requirements.
[0194] In some embodiments of the present specification, the three-dimensional coordinate information in the driving instruction is adjusted based on the reliability of the preset algorithm. When the preset algorithm cannot make an accurate prediction, the detection device can be adjusted in time to improve the prediction accuracy of the battery cell defects, thereby ensuring that the reliability of the preset algorithm is high, thereby reducing the occurrence of missed detection and false detection.
[0195] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.
[0196] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.
[0197] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.
[0198] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.
[0199] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values is as accurate as possible within the feasible range.
[0200] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.
[0201] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.
Claims
1. A battery cell testing system, characterized in that: It includes a light-emitting module, a cell conveying device, a data acquisition module, a detection device, a driving component and an analysis and control module; wherein, The light emitting module is configured to provide illumination for the cell sheet in the detection area based on the light emitting component; The conveying device includes a fixing component configured to convey the battery sheet to the inspection area based on a preset inspection rate, and to convey the battery sheet with different inspection results to different to-be-processed areas based on a conveying instruction, wherein the fixing component is used to fix the battery sheet; The data acquisition module is configured to, in response to the light emitting module being in a working state, acquire image data of the cell based on the detection device; The driving assembly includes a translation assembly and a rotation assembly, and is configured to control the orientation of the detection device based on a driving instruction; The control module includes a processor configured to: Determining a first detection result of the battery cell based on the image data includes: Based on the image data, determining regional features through an image detection model, the regional features including defect probabilities of multiple sub-regions in the battery cell and corresponding first confidences, the multiple sub-regions are obtained by dividing the image data based on preset rules, and the image detection model is a machine learning model; Determining at least one target area in the cell based on the defect probability and the first confidence level; Determining the first detection result of the battery cell based on the at least one target area; Based on the first detection result, perform at least one of the following: Generate the driving instruction, and send the driving instruction to the driving component to control the driving component to adjust the orientation of the battery sheet, wherein the processor is further configured to: Determining defect distribution characteristics of the cell based on the at least one target area in the cell; generating a candidate driving instruction based on the at least one target area and the defect distribution characteristics; Determine the estimated effect of the candidate driving instruction by means of an orientation effect model, wherein the input of the orientation effect model includes the position information of the target area in the battery cell, the area feature, the defect distribution feature and at least one of the candidate driving instructions, and the orientation effect model is a machine learning model; In response to the estimated effect of the candidate driving instruction meeting a preset condition, determining the candidate driving instruction as a target driving instruction; Generate a control instruction, and send the control instruction to the conveying device to adjust the conveying rhythm of the conveying device; Generate a lighting adjustment instruction, and send the lighting adjustment instruction to the lighting module to adjust the lighting parameters of the lighting module.
2. The battery cell testing system according to claim 1, characterized in that: The area to be processed includes at least one of a production area, a re-inspection area and a recycling area; The processor is further configured to: In response to the number of battery cells to be re-inspected in the re-inspection area reaching a preset threshold, re-inspecting the battery cells to be re-inspected and adjusting the conveying rhythm; The re-inspection process includes adjusting the inspection surface of the battery cell to be re-inspected.
3. The cell testing system according to claim 1, characterized in that: The driving instruction includes three-dimensional coordinate information and inclination information for controlling the detection device; The processor is further configured to: Based on the target area, determining the three-dimensional coordinate information and the inclination information; adjusting the orientation of the detection device according to the three-dimensional coordinate information and the inclination information, and acquiring additional data based on the detection device; Based on the additional data, determining a second detection result by a control module; determining the control instruction based on the second detection result; as well as, Adjust the configuration of the fixing assembly.
4. A battery cell testing method, characterized in that: The method is executed based on a processor of a control module, and includes: Determining a first detection result of the battery cell based on the image data includes: Based on the image data, determining regional features through an image detection model, the regional features including defect probabilities of multiple sub-regions in the battery cell and corresponding first confidences, the multiple sub-regions are obtained by dividing the image data based on preset rules, and the image detection model is a machine learning model; Determining at least one target area in the cell based on the defect probability and the first confidence level; Determining the first detection result of the battery cell based on the at least one target area; Based on the first detection result, perform at least one of the following: Generate a driving instruction, and send the driving instruction to a driving component to control the driving component to adjust the orientation of the battery sheet, wherein the processor is further configured to: Determining defect distribution characteristics of the cell based on the at least one target area in the cell; generating a candidate driving instruction based on the at least one target area and the defect distribution characteristics; Determine the estimated effect of the candidate driving instruction by means of an orientation effect model, wherein the input of the orientation effect model includes the position information of the target area in the battery cell, the area feature, the defect distribution feature and at least one of the candidate driving instructions, and the orientation effect model is a machine learning model; In response to the estimated effect of the candidate driving instruction meeting a preset condition, determining the candidate driving instruction as a target driving instruction; Generate a control instruction, and send the control instruction to the conveying device to adjust the conveying rhythm of the conveying device; Generate a lighting adjustment instruction, and send the lighting adjustment instruction to the lighting module to adjust the lighting parameters of the lighting module.
5. The method according to claim 4, characterized in that The area to be processed includes at least one of a production area, a re-inspection area, and a recycling area, and the method further includes: In response to the number of battery cells to be re-inspected in the re-inspection area reaching a preset threshold, re-inspecting the battery cells to be re-inspected and adjusting the conveying rhythm; The re-inspection process includes adjusting the inspection surface of the battery cell to be re-inspected.
6. The method according to claim 4, characterized in that The driving instruction includes three-dimensional coordinate information and inclination information for controlling the detection device, and the generating driving instruction also includes: Based on the target area, determining the three-dimensional coordinate information and the inclination information; adjusting the orientation of the detection device according to the three-dimensional coordinate information and the inclination information, and acquiring additional data based on the detection device; Based on the additional data, determining a second detection result by a control module; determining the control instruction based on the second detection result; and Adjust the configuration of fixed components.
