Method and system for cell size measurement based on machine vision

Abstract

The application discloses a battery cell size measurement method and system based on machine vision, and relates to the technical field of battery cell size measurement. The method comprises the following steps: calling a 3D vision module to capture the 3D point cloud of a battery cell and constructing an initial point cloud set; obtaining the deformation data of the battery cell under typical working conditions, performing deformation analysis, and constructing a clustering deformation field; performing layout optimization of strain gauges, configuring the strain gauges according to the layout optimization results, performing battery cell monitoring, and generating a monitoring deformation data set; after registering the initial point cloud set, establishing the contour position coordinates, performing coordinate compensation based on the monitoring deformation data set, and generating a battery cell size measurement result according to the coordinate compensation result. The technical problems that the dynamic deformation characteristics of the battery cell under real charging and discharging cycle working conditions are difficult to capture in the prior art, which further leads to insufficient data fusion accuracy and real-time performance and poor size measurement accuracy are solved, and the technical effects of improving the size measurement accuracy of the battery cell and the safety of the battery management are achieved.

Description

Technical Field

[0001] This application relates to the technical field of battery cell size measurement, specifically to a method and system for measuring battery cell size based on machine vision. Background Technology

[0002] As the basic unit of a battery, the precise measurement of the battery cell's geometric dimensions directly affects the assembly accuracy, thermal management efficiency, and long-term cycle stability of the battery module. Traditional cell size measurement results are easily affected by other factors, resulting in large accuracy deviations, which makes it difficult to meet the high precision requirements of current cell manufacturing. Even with the introduction of contact measurement equipment, although accuracy has been improved, the data fusion accuracy is insufficient, and contact measurement can cause potential damage to the cell surface, such as scratches and wear, affecting the cell's performance and lifespan. In addition, under dynamic operating conditions (such as charging and discharging, temperature changes, and mechanical vibration), the cell is prone to local deformation, leading to the accumulation of dimensional deviations. It is difficult to capture dynamic deformation characteristics, accurately characterize the three-dimensional deformation distribution and complex surface features of the cell, and even trigger safety hazards such as thermal runaway.

[0003] Therefore, current technologies have limitations in capturing the dynamic deformation characteristics of battery cells under real charge and discharge cycles, leading to insufficient data fusion accuracy and real-time performance, as well as poor dimensional measurement accuracy. Summary of the Invention

[0004] This application provides a machine vision-based method and system for measuring battery cell dimensions, which solves the technical problem in the prior art that it is difficult to capture the dynamic deformation characteristics of battery cells under real charge and discharge cycle conditions, resulting in insufficient data fusion accuracy and real-time performance, and poor size measurement accuracy. This achieves the technical effect of improving the accuracy of battery cell size measurement and the safety of battery management.

[0005] This application provides a machine vision-based method for measuring the size of battery cells. The method includes: calling a 3D vision module to capture 3D point clouds of the battery cell and constructing an initial point cloud set; establishing typical working conditions of the battery cell and acquiring deformation data of the battery cell under typical working conditions; performing deformation analysis on the deformation data of the battery cell and constructing a clustered deformation field, wherein the clustered deformation field includes a high-strain region, a transition region, and a low-strain region; optimizing the placement of strain gauges in the clustered deformation field, establishing the placement optimization result, configuring strain gauges according to the placement optimization result, performing battery cell monitoring, and generating a monitoring deformation dataset. The objective function of the placement optimization includes a monitoring accuracy function, a cost constraint function, and a safety constraint function; after registering the initial point cloud set, performing contour recognition, establishing contour position coordinates, performing coordinate compensation of the contour position coordinates based on the monitoring deformation dataset, and generating battery cell size measurement results based on the coordinate compensation results.

[0006] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: establishing a target function for layout optimization, and performing layout optimization based on the target function, wherein the target function is as follows:

[0007] ;

[0008] in, Characterize the objective function, Characterizing the strain gauge set, Characterization strain gauge set The number of strain gauges inside Characterization based on strain gauge set Full-field deformation reconstruction error under the scheme Characterizing the maximum permissible reconstruction error, The maximum number of strain gauges Characterizing the safety constraint function, , , These are the weighting coefficients for the monitoring precision item, cost constraint item, and safety constraint item, respectively.

[0009] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: the monitoring accuracy function is as follows:

[0010] ;

[0011] in, This represents the total number of discrete grid points on the surface of the battery cell. Representing any grid point, Characterizing the first The actual deformation of each grid point Characterization through strain gauge set Deformation derived from position strain data inversion.

[0012] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: configuring the initial distribution of strain gauges using historical experience data to establish an initial distribution set, where each subset of the initial distribution set represents a distribution scheme for strain gauges; performing strain gauge fitness analysis under the initial distribution set according to the objective function to generate fitness analysis results; comparing the fitness of subset regions within the initial distribution set according to the regional division of the clustered deformation field, and generating a regional optimization scheme based on the regional fitness comparison results and fitness analysis results; and iteratively updating the initial distribution set according to the regional optimization scheme to complete the optimal placement of strain gauges.

[0013] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: establishing a benchmark target value for optimization based on the fitness analysis results; performing trust superposition based on the regional fitness comparison results and the benchmark target value, and determining the optimization target based on the trust superposition results; performing update analysis on the initial distribution set using the optimization target, and establishing a regional optimization scheme.

[0014] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: activating the laser sensor, performing laser data acquisition of the cell, and establishing an additional dataset; extracting the key point position coordinates based on the additional dataset, performing data registration of the initial point cloud based on the key point position coordinates, and performing contour recognition based on the data registration result to establish contour position coordinates.

[0015] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: the 3D vision module and the laser sensor are installed at the production line station, and the strain gauge is integrated into the cell carrier.

[0016] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: configuring a preset cycle range, and when the preset cycle range is triggered by a time node, performing cycle adaptive calibration, wherein the cycle adaptive calibration includes performing a loading test using a standard cell, generating a loading test result, calculating a drift compensation coefficient based on the loading test result, and performing measurement calibration management based on the drift compensation coefficient.

[0017] In a possible implementation, the machine vision-based cell size measurement method further performs the following processing: identifying size anomalies based on the cell size measurement results and establishing a size anomaly identification early warning; and managing anomaly reporting based on the size anomaly identification early warning.

[0018] This application also provides a machine vision-based battery cell size measurement system, including: an initial point cloud construction module, used to call a 3D vision module to capture the 3D point cloud of the battery cell and construct an initial point cloud; a clustered deformation field construction module, used to establish typical working conditions of the battery cell, acquire battery cell deformation data under typical working conditions, perform deformation analysis on the battery cell deformation data, and construct a clustered deformation field, wherein the clustered deformation field includes a high strain region, a transition region, and a low strain region; a monitoring deformation dataset generation module, used to optimize the placement of strain gauges in the clustered deformation field, establish placement optimization results, configure strain gauges with the placement optimization results, perform battery cell monitoring, and generate a monitoring deformation dataset, wherein the objective function of placement optimization includes a monitoring accuracy function, a cost constraint function, and a safety constraint function; and a battery cell size measurement result generation module, used to register the initial point cloud, perform contour recognition, establish contour position coordinates, perform coordinate compensation of the contour position coordinates based on the monitoring deformation dataset, and generate battery cell size measurement results based on the coordinate compensation results.

[0019] This application proposes a machine vision-based cell size measurement method and system. The method involves using a 3D vision module to capture the 3D point cloud of the cell and constructing an initial point cloud set. It then acquires cell deformation data under typical operating conditions, performs deformation analysis, and constructs a clustered deformation field. Next, it optimizes the placement of strain gauges, configures them based on the optimization results, performs cell monitoring, and generates a monitoring deformation dataset. After registering the initial point cloud set, it establishes contour position coordinates, performs coordinate compensation based on the monitoring deformation dataset, and generates cell size measurement results based on the coordinate compensation results. This solves the technical problem in existing technologies where it is difficult to capture the dynamic deformation characteristics of the cell under real charge-discharge cycle conditions, leading to insufficient data fusion accuracy and real-time performance, and poor size measurement accuracy. This achieves the technical effect of improving the accuracy of cell size measurement and the safety of battery management. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0021] Figure 1 This is a schematic flowchart of a machine vision-based cell size measurement method provided in an embodiment of this application.

[0022] Figure 2 A schematic diagram of a machine vision-based cell size measurement system provided in an embodiment of this application.

[0023] Figure labeling: Initial point cloud construction module 10, clustered deformation field construction module 20, deformation monitoring dataset generation module 30, cell size measurement result generation module 40. Detailed Implementation

[0024] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" can be the same or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0027] This application provides a machine vision-based method for measuring battery cell dimensions, such as... Figure 1 As shown, the method includes:

[0028] Step S100: Call the 3D vision module to capture the 3D point cloud of the battery cell and construct the initial point cloud set.

[0029] Preferably, a 3D vision module is a device capable of acquiring three-dimensional spatial information of an object. It typically integrates multiple sensors, including structured light 3D cameras and LiDAR. Different types of 3D vision modules have different working principles, but their purpose is to acquire the three-dimensional coordinate information of the object's surface. Specifically, a structured light 3D camera projects a specific structured light pattern (such as stripes, coded patterns, etc.) onto the object's surface, and then uses the camera to capture the pattern modulated by the object's surface. Based on the deformation of the pattern, the three-dimensional coordinates of each point on the object's surface are calculated using the principle of triangulation. LiDAR calculates the distance between each point on the object's surface and the sensor by emitting a laser beam and measuring the time it takes for the laser to travel from emission to reflection back to the receiver (Time of Flight, TOF). Combined with the emission angle information of the laser beam, the three-dimensional coordinates of each point on the object's surface can be obtained.

[0030] Preferably, the 3D vision module is aligned with the battery cell to be measured, and the module is activated to acquire data. During the acquisition process, the 3D vision module emits light (such as laser or structured light) onto the surface of the battery cell and receives the light reflected back from the surface. By analyzing and processing the reflected light, the module can calculate the coordinates of each point on the surface of the battery cell in three-dimensional space. These coordinate points constitute the 3D point cloud data of the battery cell, which is a collection of a large number of discrete three-dimensional coordinate points. Each point represents a specific position on the surface of the battery cell, reflecting the shape and contour information of the surface of the battery cell. The 3D point cloud of the battery cell is then acquired. After collecting the data, the data is organized to form an initial point cloud set. This may include filtering the collected raw point cloud data to remove obviously erroneous or unacceptable points; storing the filtered point cloud data according to a certain format, such as PLY, XYZ, and OBJ; grouping and sorting the data according to certain rules, or establishing an index structure to improve data access efficiency; and finally, calling a 3D vision module to capture the 3D point cloud of the battery cell and construct the initial point cloud set, providing a data foundation for subsequent battery cell size measurement and analysis.

[0031] Step S200: Establish typical operating conditions for the battery cell and obtain battery cell deformation data under typical operating conditions. Perform deformation analysis on the battery cell deformation data and construct a clustered deformation field, which includes a high strain region, a transition region, and a low strain region.

[0032] Preferably, typical operating conditions for the battery cell are established, such as different charge / discharge states, ambient temperatures, usage frequency, and load size. Specifically, different charge / discharge rates and depths are considered. For example, the battery cell's state differs during fast charging and slow charging, and deep charge / discharge versus shallow charge / discharge also causes the battery cell to exhibit different characteristics. Taking the power battery of an electric vehicle as an example, the battery discharges with a large current during rapid acceleration, while it may be discharging with a small current or in a charging state under urban congestion conditions, which are all different charge / discharge conditions. Ambient temperature has a significant impact on the battery cell. High temperatures accelerate the chemical reaction rate but may lead to performance degradation, while low temperatures may reduce battery capacity and increase internal resistance. For example, the battery cells of electronic devices used in desert areas face high-temperature and dry environments, while the battery cells of devices used in polar regions must withstand low-temperature environments. The operating conditions are determined according to the application scenario of the battery cell. For example, the operating conditions of a mobile phone battery differ when frequently using high-energy-consuming applications (such as games) versus when only making simple calls and sending text messages. The operating modes of energy storage batteries also differ in grid peak shaving and home energy storage scenarios. By comprehensively analyzing these factors, several combinations of operating conditions are determined to form the typical operating conditions of the battery cell.

[0033] Preferably, under different typical operating conditions, complex physical and chemical changes occur inside the battery cell, leading to changes in the external shape and size of the battery cell, i.e., deformation. Obtaining battery cell deformation data under typical operating conditions is crucial. For example, during high-current charging and discharging, the rapid insertion and extraction of lithium ions inside the battery cell may cause the electrode material to expand or contract; temperature changes can cause thermal expansion and contraction of materials. Specifically, strain gauges and displacement sensors can be installed on or inside the battery cell. Strain gauges can convert the strain on the battery cell surface into electrical signals, and the strain of the battery cell can be obtained by measuring changes in these electrical signals. Displacement sensors can directly measure the displacement changes at a specific part of the battery cell, thereby obtaining deformation data. Using digital image correlation (DIC) technology, images of the battery cell surface under different operating conditions are analyzed to calculate the displacement and strain at each point on the surface. Using a 3D vision module, 3D point cloud data of the battery cell can be captured again under different operating conditions. By comparing this with the initial point cloud set and analyzing the changes in the point cloud data, the deformation information of the battery cell can be obtained, including overall expansion, contraction, and local deformation.

[0034] Preferably, deformation analysis is performed on the cell deformation data, and a clustered deformation field is constructed to divide the data into high-strain regions, transition regions, and low-strain regions, in order to gain a deeper understanding of the deformation characteristics of the cell under different working conditions. Specifically, the acquired cell deformation data is preprocessed, such as removing noise and filling missing values, to improve data quality. Then, deformation-related features, such as the magnitude, rate of change, and direction of strain, are extracted from these data to more intuitively reflect the deformation of the cell. Then, cluster analysis is performed, that is, different regions of the cell are divided into different categories based on the similarity of the data. Each location on the surface or inside of the cell is used as a sample point, and the extracted deformation features are used as variables. A clustering algorithm is used to cluster all sample points. Through continuous iteration and optimization, the entire area of ​​the battery cell is eventually divided into different cluster regions, each corresponding to a specific deformation feature. For example, the K-Means algorithm assigns deformation data points to different clusters based on the set number of clusters (assuming 3 clusters, corresponding to high strain region, transition region, and low strain region). This makes the data points within the same cluster have high similarity, while the data points between different clusters have large differences. The high-strain region refers to the area where the battery cell undergoes significant deformation under specific operating conditions. The performance and stability of the battery cell may be significantly affected, and it is more prone to fatigue, cracking, and other problems. The low-strain region is the opposite of the high-strain region. This region has relatively small deformation under various typical operating conditions because the material has good mechanical properties or the stress inside the battery cell is small. Therefore, it can maintain a relatively stable shape when the operating conditions change, and the impact on the overall performance of the battery cell is relatively small. The transition region is located between the high-strain region and the low-strain region. Its strain level is between the two and plays a connecting and transitional role. The strain change in the transition region is more complex. It is affected by the high-strain region and interacts with the low-strain region. Its strain characteristics may fluctuate within a certain range with changes in operating conditions.

[0035] Step S300: In the clustered deformation field, the strain gauges are optimized for placement, the optimization results are established, and the strain gauges are configured according to the optimization results. Cell monitoring is performed, and a monitoring deformation dataset is generated. The objective function of the optimization includes the monitoring accuracy function, the cost constraint function, and the safety constraint function.

[0036] Preferably, a strain gauge is a sensor used to measure the strain on the surface of an object. By rationally arranging strain gauges on the surface of the battery cell, the deformation of the battery cell under different operating conditions can be monitored in real time. The clustered deformation field has divided the battery cell into high-strain region, transition region, and low-strain region. Different regions have different deformation characteristics. The optimal placement of the strain gauges is determined based on the deformation characteristics. That is, different strain gauge placement schemes are continuously tried in different regions of the clustered deformation field. By evaluating the advantages and disadvantages of each scheme, the optimal placement method is finally found. Specifically, each strain gauge placement scheme is evaluated according to the preset objective function (i.e., monitoring accuracy function, cost constraint function, and safety constraint function). By calculating the value of different schemes under the objective function, it is determined whether the scheme meets the requirements. Finally, the scheme that makes the objective function reach the optimal value is selected as the final placement optimization result, so as to clarify the specific position and number of strain gauges in the clustered deformation field.

[0037] Preferably, according to the optimized layout results, the strain gauges are accurately installed at designated positions on the surface of the battery cell to ensure good contact between the strain gauges and the surface of the battery cell, so as to ensure accurate measurement of the strain of the battery cell. After the strain gauges are configured, real-time monitoring of the battery cell is started, the strain on the surface of the battery cell is converted into an electrical signal and processed and analyzed to obtain the deformation information of the battery cell under different operating conditions. The electrical signals output by the strain gauges are continuously collected and converted into corresponding strain values ​​to generate a monitoring deformation dataset, which may include corresponding operating condition information (such as charging and discharging status, ambient temperature, etc.). The monitoring accuracy function measures the accuracy of strain gauge placement in monitoring cell deformation. It is typically related to the location and number of strain gauges, as well as the characteristics of the clustered deformation field. For example, appropriately increasing the number of strain gauges in high-strain areas can improve the monitoring accuracy of deformation in that area. By optimizing the monitoring accuracy function, strain gauges can more accurately capture cell deformation information. The cost constraint function controls the overall cost of the monitoring system. While meeting monitoring accuracy requirements, it minimizes the number of strain gauges used or selects cost-effective strain gauge products. For example, in low-strain areas, the number of strain gauges can be appropriately reduced to lower costs. The safety constraint function ensures that the strain gauge placement does not negatively impact cell safety. For example, it avoids excessively dense placement of strain gauges in critical parts of the cell or areas prone to failure, to prevent affecting normal cell operation or increasing safety risks such as short circuits. By integrating these three objective functions, the optimal placement of strain gauges is achieved, enabling efficient, accurate, and safe cell monitoring.

[0038] Furthermore, step S300 also includes establishing an objective function for deployment optimization, and performing deployment optimization based on the objective function, wherein the objective function is as follows:

[0039] ;

[0040] in, Characterize the objective function, Characterizing the strain gauge set, Characterization strain gauge set The number of strain gauges inside Characterization based on strain gauge set Full-field deformation reconstruction error under the scheme Characterizing the maximum permissible reconstruction error, The maximum number of strain gauges Characterizing the safety constraint function, , , These are the weighting coefficients for the monitoring precision item, cost constraint item, and safety constraint item, respectively.

[0041] Preferred, The monitoring accuracy is measured by the ratio of the full-field deformation reconstruction error to the maximum allowable reconstruction error under the strain gauge set S scheme. The smaller the value, the higher the monitoring accuracy. To measure cost constraints, it reflects the ratio of the number of strain gauges in strain gauge set S to the upper limit of the number, and is used to control costs; The safety constraint function is used to evaluate the impact of strain gauge placement on cell safety. This method is used to calculate the full-field deformation reconstruction error based on the strain gauge set S scheme by calculating the true deformation of discrete grid points on the cell surface. and the deformation inverted from the strain data at position S of the strain gauge set The average of the sum of squares of the differences between them is used to measure the accuracy. The smaller the value, the closer the inverted deformation is to the true value, and the higher the monitoring accuracy.

[0042] Furthermore, step S300 also includes the monitoring accuracy function as follows:

[0043] ;

[0044] in, This represents the total number of discrete grid points on the surface of the battery cell. Representing any grid point, Characterizing the first The actual deformation of each grid point Characterization through strain gauge set Deformation derived from position strain data inversion.

[0045] Furthermore, step S300 also includes step S310, configuring the initial distribution of strain gauges using historical experience data to establish an initial distribution set, where each subset in the initial distribution set represents a distribution scheme for strain gauges; step S320, performing strain gauge fitness analysis under the initial distribution set according to the objective function to generate fitness analysis results; step S330, comparing the fitness of subset regions within the initial distribution set according to the regional division of the clustered deformation field, and generating a regional optimization scheme based on the regional fitness comparison results and fitness analysis results; step S340, iteratively updating the initial distribution set according to the regional optimization scheme to complete the optimization of strain gauge placement.

[0046] Preferably, historical experience data includes information such as the deformation characteristics of the battery cell under different operating conditions and the monitoring effects corresponding to different strain gauge distribution schemes. Based on historical experience data, multiple different strain gauge distribution schemes are designed, each scheme corresponding to a way of arranging strain gauges on the surface of the battery cell. These schemes are then combined to form an initial distribution set. Each subset in the initial distribution set represents a set of strain gauge distribution schemes, including information such as the specific location and number of strain gauges. A fitness analysis is performed on each subset (i.e., each strain gauge distribution scheme) in the initial distribution set using an objective function. That is, the fitness of the scheme is evaluated based on the calculation results of the objective function. The higher the fitness, the more the scheme meets our requirements and the better its overall performance in terms of monitoring accuracy, cost, and safety. By analyzing the fitness of all schemes, fitness analysis results are generated, which typically include the fitness score of each scheme and its corresponding ranking.

[0047] Preferably, for each subset of the initial distribution set, the distribution of strain gauges in each region of the clustered deformation field is analyzed to evaluate the fitness of each scheme in different regions. For example, in high-strain regions, can the distribution of strain gauges accurately capture the large deformation in that region? In low-strain regions, does it avoid overly dense deployment to reduce costs? Through regional fitness comparison, the fitness evaluation of each scheme in different regions is obtained (i.e., regional fitness comparison results). Then, by comprehensively considering the regional fitness comparison results and the previous fitness analysis results, the schemes that perform better in each region and overall are identified, and regional optimization schemes are generated. This approach improves the monitoring effect of strain gauges in different regions while considering cost and safety factors. Finally, the optimized regional scheme is applied to the initial distribution set, and subsets within it are updated. For example, the position or number of strain gauges in certain schemes is adjusted according to the optimized scheme. The fitness analysis, regional fitness comparison, and iterative update process described above are repeated until certain stopping conditions are met, such as the fitness score no longer showing significant improvement or reaching a preset number of iterations. Ultimately, the optimal placement of strain gauges is achieved, and the scheme with the highest fitness in the initial distribution set becomes the final strain gauge placement scheme, thereby enabling efficient and accurate monitoring of the battery cells.

[0048] Furthermore, step S330 also includes step S331, establishing a baseline target value for optimization based on the fitness analysis results; step S332, performing trust superposition based on the regional fitness comparison results and the baseline target value, and determining the optimization target based on the trust superposition results; and step S333, performing an update analysis of the initial distribution set using the optimization target to establish a regional optimization scheme.

[0049] Preferably, based on multiple fitness values ​​in the fitness analysis results, a reasonable threshold is set as the baseline target value to measure whether the new scheme is better. Then, the regional fitness comparison results and the baseline target value are superimposed with trust. For example, by using a certain weighting method, the regional fitness and the baseline target value are weighted and summed. Schemes that perform well in high-strain regions are given higher weights because accurate monitoring of high-strain regions is crucial to the overall monitoring effect. Through trust superposition, a new value that considers regional performance and overall fitness is obtained. Then, based on the trust superposition result, a clear optimization target is determined, that is, a more comprehensive and accurate evaluation standard that considers both the fitness of the scheme in each region and the previously established baseline target value.

[0050] Preferably, guided by the optimization target, each subset of the initial distribution set is re-evaluated and analyzed, comparing the gap between each scheme and the optimization target to identify areas for improvement. For example, if a scheme has low adaptability in a certain area or its overall adaptability differs significantly from the optimization target, the scheme is adjusted. Finally, based on the results of the updated analysis, the schemes in the initial distribution set are optimized, including adjusting parameters such as the position and number of strain gauges according to the characteristics and needs of different areas, to obtain regional optimized schemes that are closer to the optimization target in each area and overall. This improves the overall performance of the strain gauge deployment scheme and ensures that the final strain gauge distribution scheme can achieve good monitoring results in different areas and overall, while also taking into account factors such as cost and safety.

[0051] Step S400: After registering the initial point cloud, perform contour recognition, establish contour position coordinates, perform coordinate compensation of the contour position coordinates based on the monitored deformation dataset, and generate cell size measurement results based on the coordinate compensation results.

[0052] Preferably, when acquiring the initial point cloud set, factors such as measurement errors of the 3D vision module and differences in the placement of the battery cells may cause positional deviations in the point cloud data. Therefore, registration is performed on the initial point cloud set, which involves aligning point cloud data acquired from different viewpoints or at different times to ensure accurate relative positional relationships within the same coordinate system. For example, the Iterative Closest Point (ICP) algorithm can be used for registration. By iteratively searching for the optimal transformation matrix (including translation and rotation) between two point clouds, the distance between corresponding points in the two point clouds is minimized, thereby achieving accurate registration of the point clouds. Then, contour recognition is performed, which involves extracting the outer edge information of the battery cells from the registered initial point cloud set and identifying the contours. Clearly defining the shape and extent of the battery cell, that is, determining the boundary of the battery cell, is crucial. For example, edge detection algorithms, such as the Canny edge detection algorithm, calculate the gradient values ​​of each point in the point cloud data, find the points with larger gradient values ​​as edge points, and connect these edge points to form the outline of the battery cell. Then, accurate position coordinates are established for each point on the identified outline. Usually, a Cartesian coordinate system is used, with a fixed point as the origin, to determine the directions of the three coordinate axes X, Y, and Z. Based on the registered point cloud data, the position information of each point on the outline is converted into coordinate values ​​in this coordinate system, thus obtaining the set of position coordinates of the outline, which accurately describes the position and shape of the battery cell outline in three-dimensional space.

[0053] Preferably, the battery cell will deform under various operating conditions (such as charging and discharging, temperature changes, etc.). The deformation monitoring dataset records the deformation information of the battery cell under different operating conditions. The contour position coordinates are then corrected based on the deformation monitoring dataset, taking into account the actual deformation of the battery cell, to obtain more accurate battery cell size measurement results. Specifically, the contour position coordinates are adjusted accordingly based on the deformation (such as displacement, strain, etc.) of each point recorded in the deformation monitoring dataset. For example, if a point undergoes a displacement ∆X along the X-axis under a certain operating condition, then the X coordinate value of that point is added to ∆X to achieve coordinate compensation. Finally, various size parameters of the battery cell are calculated based on the compensated coordinate data. For example, by finding the coordinates of the corresponding endpoints on the contour and calculating the distance between them, the basic dimensions such as the length, width, and height of the battery cell can be obtained, thereby generating accurate battery cell size measurement results that reflect the actual size of the battery cell.

[0054] Furthermore, step S400 also includes step S410, activating the laser sensor, performing laser data acquisition of the battery cell, and establishing an additional dataset; step S420, after extracting the key point position coordinates according to the additional dataset, performing data registration of the initial point cloud based on the key point position coordinates, and performing contour recognition according to the data registration result to establish contour position coordinates.

[0055] Preferably, the laser sensor emits a laser beam and then receives the laser signal reflected from the surface of the battery cell. Based on the time-of-flight (TOF) or triangulation principle, the distance from the sensor to various points on the surface of the battery cell is calculated. By scanning the surface of the battery cell, a large amount of distance information can be obtained. The laser sensor is then activated to scan the battery cell according to a predetermined path or pattern, collecting distance data and corresponding spatial position information of various points on the surface of the battery cell, thus obtaining an additional dataset containing detailed geometric information of the surface of the battery cell. The data in the additional dataset is analyzed, and key points are selected from multiple points reflecting the shape and structural characteristics of the battery cell, such as the vertices, edge turning points, and points with large curvature changes. After the key points are determined, the position coordinates of these key points are extracted from the additional dataset, including the X, Y, and Z coordinate values ​​in three-dimensional space, for subsequent data registration and contour recognition operations.

[0056] Preferably, data registration is performed on the initial point cloud. This involves using the coordinates of key points as a reference and employing a specific registration algorithm (such as the Iterative Closest Point (ICP) algorithm) to calculate the transformation relationship (including translation and rotation) between the initial point cloud and the supplementary dataset. Then, the initial point cloud is transformed accordingly to ensure that it is spatially consistent with the supplementary dataset. Finally, image processing or point cloud processing algorithms are used to identify the outer contour of the battery cell from the registered data. For example, by detecting edge points in the point cloud data, these edge points are connected to form a closed curve, i.e., the contour of the battery cell. The position coordinates of each point on the contour are then extracted to form a set of contour position coordinates.

[0057] Preferably, step S400 further includes installing the 3D vision module and the laser sensor at the production line station, and integrating the strain gauge into the battery cell carrier.

[0058] Preferably, on the battery cell production line, 3D vision modules and laser sensors are installed at specific work positions (production line stations). The 3D vision modules can acquire three-dimensional spatial information of the battery cell surface and construct an initial point cloud. The laser sensors can collect laser data of the battery cell by emitting and receiving laser signals. Data collection can be performed on the battery cell in a timely manner when the battery cell production or testing process passes through this position. Strain gauges are components used to monitor the deformation of the battery cell and are integrated and installed on the battery cell carrier. The battery cell carrier is a tool used to carry the battery cell as it moves on the production line. Integrating the strain gauges onto it allows the strain gauges to monitor the deformation data of the battery cell under various working conditions in real time as the battery cell moves with the carrier and goes through different production stages, without affecting the normal transmission of the battery cell on the production line.

[0059] Furthermore, step S400 also includes configuring a preset cycle range, and when the preset cycle range is triggered by a time node, performing cycle adaptive calibration. The cycle adaptive calibration includes performing a loading test using a standard cell, generating a loading test result, calculating a drift compensation coefficient based on the loading test result, and performing measurement calibration management based on the drift compensation coefficient.

[0060] Preferably, a time period range is set according to production process requirements, equipment characteristics, and testing accuracy requirements. When the time progresses to a time node that matches the preset period range, the system automatically starts the periodic adaptive calibration process. That is, according to the pre-set rules, calibration operations are started at a specific time point to ensure that the equipment or measurement system always maintains the best working state and adapts to various changes and possible error accumulation in the production process. Among them, the periodic adaptive calibration includes using standard cells to perform loading tests, generating loading test results, calculating drift compensation coefficients based on loading test results, and performing measurement calibration management based on drift compensation coefficients.

[0061] Preferably, the standard battery cell is a battery cell with precisely known characteristics and parameters, serving as a benchmark to measure the performance of other battery cells and the accuracy of testing equipment. The standard battery cell is placed on a battery cell carrier, simulating the loading conditions in actual production, undergoing the same processes and operations as in actual production, such as moving along the production line and passing through various testing equipment. Various data related to the loading of the standard battery cell are recorded, such as the operating status of the battery cell carrier and the feedback data from various sensors in contact with the standard battery cell. Loading test results are generated, which may include the positional accuracy of the standard battery cell on the carrier, the pressure distribution of the carrier on the standard battery cell, and the differences between the measured data of each sensor and the standard values. The loading test results are then analyzed in depth to identify the actual... The measurement process involves analyzing the deviation between actual measurement data and standard values. Based on these deviations, a drift compensation coefficient is calculated to quantify the degree of drift generated by the equipment or measurement system during loading. Finally, the calculated drift compensation coefficient is applied to the actual measurement data. In subsequent measurements and tests of ordinary battery cells, the measurement results are corrected according to the drift compensation coefficient to make the measurement data closer to the true value. In addition, the process also includes managing the entire measurement calibration process, such as recording the time of each calibration, the specific value of the drift compensation coefficient, and the changes in measurement data before and after calibration, so as to track and evaluate the performance of the equipment and the stability of the measurement system, thereby ensuring the accuracy of battery cell size measurement and production quality.

[0062] Furthermore, step S400 also includes step S430, identifying size anomalies based on the cell size measurement results and establishing a size anomaly identification early warning; step S440, managing anomaly reporting based on the size anomaly identification early warning.

[0063] Preferably, based on the cell size design and quality requirements, a standard range for cell size is established, including allowable tolerances for key dimensions such as length, width, and height. For example, if the standard length of a certain cell is 50mm, the allowable tolerance range is ±0.2mm, meaning the length of a qualified product should be between 49.8mm and 50.2mm. Then, a size anomaly identification and early warning system is implemented, which compares the collected cell size measurement results with the pre-set size standard. If the measurement result exceeds the standard range, the system determines that the cell has a size anomaly and establishes a size anomaly early warning mechanism. Upon receiving a size anomaly identification and early warning, the system records detailed information about the abnormal cell, including but not limited to the cell's unique identifier (such as serial number), production batch, production time, specific abnormal size value, and the degree of deviation from the standard value. Finally, anomaly reporting management is implemented to ensure that size anomalies are identified and handled in a timely manner.

[0064] In the above text, refer to Figure 1A machine vision-based method for measuring battery cell size according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 A machine vision-based battery cell size measurement system according to an embodiment of the present invention is described.

[0065] The machine vision-based cell size measurement system according to embodiments of the present invention addresses the technical problems in the prior art, namely, the difficulty in capturing the dynamic deformation characteristics of cells under real charge-discharge cycle conditions, leading to insufficient data fusion accuracy and real-time performance, and poor size measurement accuracy. This system achieves the technical effect of improving the accuracy of cell size measurement and the safety of battery management. Figure 2 As shown, the machine vision-based cell size measurement system includes: an initial point cloud construction module 10, a clustering deformation field construction module 20, a deformation monitoring dataset generation module 30, and a cell size measurement result generation module 40.

[0066] The initial point cloud construction module 10 is used to call the 3D vision module to capture the 3D point cloud of the battery cell and construct the initial point cloud. The clustered deformation field construction module 20 is used to establish typical working conditions of the battery cell, acquire the deformation data of the battery cell under typical working conditions, perform deformation analysis on the deformation data of the battery cell, and construct a clustered deformation field, which includes a high strain region, a transition region, and a low strain region. The monitoring deformation dataset generation module 30 is used to optimize the placement of strain gauges in the clustered deformation field, establish the placement optimization result, configure the strain gauges with the placement optimization result, perform battery cell monitoring, and generate a monitoring deformation dataset. The objective function of the placement optimization includes a monitoring accuracy function, a cost constraint function, and a safety constraint function. The battery cell size measurement result generation module 40 is used to register the initial point cloud, perform contour recognition, establish contour position coordinates, perform coordinate compensation of the contour position coordinates based on the monitoring deformation dataset, and generate battery cell size measurement results based on the coordinate compensation result.

[0067] The specific configuration of the deformation monitoring dataset generation module 30 will be described in detail below. The deformation monitoring dataset generation module 30 further includes: establishing an objective function for deployment optimization, and performing deployment optimization based on the objective function, wherein the objective function is as follows:

[0068] ;

[0069] in, Characterize the objective function, Characterizing the strain gauge set, Characterization strain gauge set The number of strain gauges inside Characterization based on strain gauge set Full-field deformation reconstruction error under the scheme Characterizing the maximum permissible reconstruction error, The maximum number of strain gauges Characterizing the safety constraint function, , , These are the weighting coefficients for the monitoring precision item, cost constraint item, and safety constraint item, respectively.

[0070] The specific configuration of the deformation monitoring dataset generation module 30 will be described in detail below. The deformation monitoring dataset generation module 30 further includes the following: the monitoring accuracy function is as follows:

[0071] ;

[0072] in, This represents the total number of discrete grid points on the surface of the battery cell. Representing any grid point, Characterizing the first The actual deformation of each grid point Characterization through strain gauge set Deformation derived from position strain data inversion.

[0073] The specific configuration of the deformation monitoring dataset generation module 30 will be described in detail below. The deformation monitoring dataset generation module 30 further includes: configuring the initial distribution of strain gauges using historical experience data to establish an initial distribution set, where each subset in the initial distribution set represents a set of strain gauge distribution schemes; performing strain gauge fitness analysis under the initial distribution set according to the objective function to generate fitness analysis results; comparing the fitness of subset regions within the initial distribution set according to the regional division of the clustered deformation field, and generating a regional optimization scheme based on the regional fitness comparison results and fitness analysis results; and iteratively updating the initial distribution set according to the regional optimization scheme to complete the optimal placement of strain gauges.

[0074] The specific configuration of the deformation monitoring dataset generation module 30 will be described in detail below. The deformation monitoring dataset generation module 30 further includes: establishing a baseline target value for optimization based on the fitness analysis results; performing trust superposition based on the regional fitness comparison results and the baseline target value; determining the optimization target based on the trust superposition result; and performing an update analysis of the initial distribution set using the optimization target to establish a regional optimization scheme.

[0075] The specific configuration of the cell size measurement result generation module 40 will be described in detail below. The cell size measurement result generation module 40 further includes: activating the laser sensor, performing laser data acquisition of the cell, and establishing an additional dataset; extracting the key point position coordinates based on the additional dataset, performing data registration of the initial point cloud based on the key point position coordinates, and performing contour recognition based on the data registration result to establish contour position coordinates.

[0076] The specific configuration of the cell size measurement result generation module 40 will be described in detail below. The cell size measurement result generation module 40 further includes: the 3D vision module and the laser sensor are installed at the production line station, and the strain gauge is integrated into the cell carrier.

[0077] The specific configuration of the cell size measurement result generation module 40 will be described in detail below. The cell size measurement result generation module 40 further includes: configuring a preset period range; when the preset period range is triggered by a time node, performing period adaptive calibration; the period adaptive calibration includes performing a loading test using a standard cell, generating loading test results, calculating a drift compensation coefficient based on the loading test results, and performing measurement calibration management based on the drift compensation coefficient.

[0078] The specific configuration of the cell size measurement result generation module 40 will be described in detail below. The cell size measurement result generation module 40 further includes: identifying size anomalies based on the cell size measurement results and establishing a size anomaly identification early warning; and managing anomaly reporting based on the size anomaly identification early warning.

[0079] The machine vision-based cell size measurement system provided in this embodiment of the invention can execute the machine vision-based cell size measurement method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0080] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0081] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for measuring battery cell size based on machine vision, characterized in that, The method includes: The 3D vision module is invoked to capture the 3D point cloud of the battery cell and construct an initial point cloud set; Establish typical operating conditions for battery cells and acquire cell deformation data under typical operating conditions. Perform deformation analysis on the cell deformation data and construct a clustered deformation field, which includes a high strain region, a transition region, and a low strain region. In the clustered deformation field, strain gauge placement optimization is performed, placement optimization results are established, and strain gauges are configured according to the placement optimization results to perform cell monitoring and generate a monitoring deformation dataset. The objective function of placement optimization includes monitoring accuracy function, cost constraint function, and safety constraint function. After registering the initial point cloud, contour recognition is performed to establish contour position coordinates. Based on the monitored deformation dataset, coordinate compensation of the contour position coordinates is performed, and cell size measurement results are generated based on the coordinate compensation results.

2. The cell size measurement method based on machine vision as described in claim 1, characterized in that, The optimization of strain gauge placement in the clustered deformation field includes: Establish an objective function for layout optimization, and perform layout optimization based on the objective function. The objective function is as follows: ； in, Characterize the objective function, Characterizing the strain gauge set, Characterization strain gauge set The number of strain gauges inside Characterization based on strain gauge set Full-field deformation reconstruction error under the scheme Characterizing the maximum permissible reconstruction error, The maximum number of strain gauges Characterizing the safety constraint function, , , These are the weighting coefficients for the monitoring precision item, cost constraint item, and safety constraint item, respectively.

3. The cell size measurement method based on machine vision as described in claim 2, characterized in that, The monitoring accuracy function is as follows: ； in, This represents the total number of discrete grid points on the surface of the battery cell. Representing any grid point, Characterizing the first The actual deformation of each grid point Characterization through strain gauge set Deformation from position strain data inversion.

4. The cell size measurement method based on machine vision as described in claim 2, characterized in that, After establishing the objective function for layout optimization, the following is included: The initial distribution of strain gauges is configured using historical experience data to establish an initial distribution set, where each subset of the initial distribution set represents a set of strain gauge distribution schemes. Based on the objective function, strain gauge fitness analysis is performed on the initial distribution set to generate fitness analysis results; Based on the regional division of the clustered deformation field, the fitness of subset regions within the initial distribution set is compared, and a regional optimization scheme is generated based on the regional fitness comparison results and fitness analysis results. The initial distribution set is iteratively updated according to the regional optimization scheme to complete the optimal placement of strain gauges.

5. The cell size measurement method based on machine vision as described in claim 4, characterized in that, The process of generating a regional optimization scheme based on the regional fitness comparison results and fitness analysis results includes: Establish a baseline target value for optimization based on the fitness analysis results; Trust is superimposed based on the regional fitness comparison results and the baseline target value, and the optimization target is determined based on the trust superposition results. The initial distribution set is updated and analyzed based on the optimization objective to establish a regional optimization scheme.

6. The cell size measurement method based on machine vision as described in claim 1, characterized in that, After registering the initial point cloud, contour recognition is performed to establish contour position coordinates, including: Activate the laser sensor to acquire laser data from the battery cell and build an additional dataset; After extracting the key point coordinates from the additional dataset, data registration of the initial point cloud is performed based on the key point coordinates. Contour recognition is then performed based on the data registration results to establish contour coordinates.

7. The cell size measurement method based on machine vision as described in claim 6, characterized in that, The 3D vision module and the laser sensor are installed at the production line station, and the strain gauge is integrated into the battery cell carrier.

8. The cell size measurement method based on machine vision as described in claim 1, characterized in that, After generating the cell size measurement results based on the coordinate compensation results, the process includes: Configure a preset cycle range. When the preset cycle range is triggered by a time node, perform cycle adaptive calibration. The cycle adaptive calibration includes performing a loading test using a standard cell, generating a loading test result, calculating a drift compensation coefficient based on the loading test result, and performing measurement calibration management based on the drift compensation coefficient.

9. The cell size measurement method based on machine vision as described in claim 1, characterized in that, The process of generating cell size measurement results based on coordinate compensation results also includes: Based on the cell size measurement results, size anomalies are identified, and a size anomaly identification early warning system is established. Anomaly reporting management is implemented based on the aforementioned size anomaly identification and early warning system.

10. A machine vision-based battery cell size measurement system, characterized in that, The system is used to implement the machine vision-based cell size measurement method according to any one of claims 1 to 9, the system comprising: The initial point cloud construction module is used to call the 3D vision module to capture the 3D point cloud of the battery cell and construct the initial point cloud. The clustered deformation field construction module is used to establish typical working conditions of the battery cell, acquire the deformation data of the battery cell under typical working conditions, perform deformation analysis on the deformation data of the battery cell, and construct a clustered deformation field, which includes a high strain region, a transition region, and a low strain region. The deformation monitoring dataset generation module is used to optimize the placement of strain gauges in the clustered deformation field, establish the optimization results, configure strain gauges according to the optimization results, perform cell monitoring, and generate a deformation monitoring dataset. The objective function of the optimization includes a monitoring accuracy function, a cost constraint function, and a safety constraint function. The cell size measurement result generation module is used to register the initial point cloud, perform contour recognition, establish contour position coordinates, perform coordinate compensation of the contour position coordinates based on the monitored deformation dataset, and generate cell size measurement results based on the coordinate compensation results.

Images