Intelligent detection method and system for working procedures of energy storage battery production line

Abstract

The invention provides an energy storage battery production line process intelligent detection method and system, and relates to the technical field of energy storage battery production detection. Collecting AC impedance and temperature data to generate a performance distribution diagram; performing charging identification on an abnormal area and determining a defect type; carrying out pulse discharge test on the abnormal region to calculate a capacity loss rate; and marking unqualified products according to the capacity loss rate and adjusting process parameters. The method can accurately identify the performance defect of the battery monomer, improve the detection precision and the production efficiency, and effectively reduce the defective product rate.

Description

Technical Field

[0001] The present invention relates to the production detection technology of energy storage batteries, and particularly to an intelligent detection method and system for the processes of an energy storage battery production line. Background Art

[0002] As a key component in the new energy field, the production quality of energy storage batteries directly affects the performance and lifespan of the entire energy storage system. With the rapid development of energy storage technology, the requirements for the intelligence and automation of battery production lines are continuously increasing. Traditional battery production lines mainly rely on manual inspection and simple electrical performance tests to judge the quality of batteries. This method is inefficient and prone to missed inspections.

[0003] In recent years, some advanced battery production lines have begun to adopt automated equipment and online detection systems to evaluate battery performance by measuring parameters such as the voltage and internal resistance of batteries. However, these methods still have some limitations. First, the single electrical performance parameter test cannot comprehensively reflect the internal state and potential defects of the battery. Second, the existing detection methods often can only give an overall performance evaluation and it is difficult to accurately locate the abnormal areas inside the battery. Finally, there is a lack of evaluation of the dynamic performance of the battery and it is impossible to accurately predict the performance of the battery during actual use.

[0004] In order to improve the production quality and consistency of energy storage batteries, it is urgent to develop a more comprehensive and accurate intelligent detection method. This method should be able to comprehensively analyze the multi-dimensional parameters of the battery, achieve accurate positioning and classification of internal abnormalities of the battery, and be able to automatically adjust the production process according to the detection results, thereby continuously optimizing the production process and improving the overall performance and yield rate of the battery. Summary of the Invention

[0005] Embodiments of the present invention provide a method and a system that can solve the problems in the prior art.

[0006] In the first aspect of the embodiments of the present invention, There is provided an intelligent detection method for the processes of an energy storage battery production line, including: Obtaining the electrical performance parameters of battery cells on an energy storage battery production line; Arranging a plurality of detection points along the length direction and the circumferential direction of the battery cell, collecting the AC impedance data and temperature data of the battery cell at each detection point, and generating a performance distribution map of the battery cell; Charging the battery cell with constant current and constant voltage to a fully charged state, recording the temperature change data of the battery cell during the charging process, identifying the abnormal area of the battery cell in the performance distribution map, and determining the type of performance defect corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data; Select the corresponding pulse discharge test parameters according to the type of performance defect, conduct pulse discharge tests on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculate the capacity loss rate of the abnormal area based on the dynamic response characteristics and the type of performance defect; When the capacity loss rate exceeds the preset loss rate threshold, mark the corresponding battery cell as a non-conforming product, control the manipulator to transfer the non-conforming product to the defective product collection area, and at the same time adjust the process parameters according to the distribution characteristics of the capacity loss rate, and send the adjusted process parameters to the production line control system.

[0007] In an alternative embodiment, Arrange multiple detection points along the length direction and the circumferential direction of the battery cell, collect the AC impedance data and temperature data of the battery cell at each detection point, and generate a performance distribution map of the battery cell including: Set multiple detection rings along the length direction of the battery cell, evenly set multiple detection points in the circumferential direction of each detection ring, construct a detection point array on the surface of the battery cell, apply an AC signal to the detection point array, collect the voltage signal and current signal of the detection point array, and calculate the AC impedance data of the detection point array; Collect the temperature data of the detection point array, establish the corresponding relationship between the AC impedance data and temperature data of the detection point array at each detection point, perform digital filtering on the AC impedance data and temperature data, and obtain the filtered AC impedance data and filtered temperature data of the detection point array; Based on the distribution coordinates of the detection point array, use the interpolation algorithm to reconstruct the filtered AC impedance data and filtered temperature data, and generate the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell; Normalize the AC impedance distribution surface and temperature distribution surface to obtain the normalized AC impedance distribution data and normalized temperature distribution data on the surface of the battery cell, and calculate the AC impedance weight coefficient and temperature weight coefficient of the battery cell; Weight and superimpose the normalized AC impedance distribution data and normalized temperature distribution data according to the AC impedance weight coefficient and temperature weight coefficient to generate the performance distribution map of the battery cell.

[0008] In an alternative embodiment, Based on the distribution coordinates of the detection point array, using the interpolation algorithm to reconstruct the filtered AC impedance data and filtered temperature data, and generating the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell includes: Calculate the Euclidean distance between adjacent detection points in the detection point array to generate a distance matrix, calculate the detection point density distribution value based on the distance matrix, compare the detection point density distribution value with the density threshold for region division, and generate detection region identification data; Perform wavelet transform filtering on the AC impedance data and temperature data, and partition and store the filtered AC impedance data and filtered temperature data according to the detection area identification data; Calculate the coefficient of variation for each partition, and allocate an interpolation algorithm based on the comparison result between the coefficient of variation and a preset variation threshold. When the coefficient of variation is less than the variation threshold, allocate the cubic spline interpolation algorithm; when the coefficient of variation is greater than or equal to the variation threshold, allocate the Kriging interpolation algorithm; Execute the allocated interpolation algorithm to perform reconstruction operations to obtain partition reconstruction data, and construct a physical constraint optimization function based on the internal resistance continuity and temperature conduction characteristics of the battery cell surface. Perform conjugate gradient optimization on the partition reconstruction data, calculate the spatial gradient value of the optimized data, determine the grid point fusion window parameters for adaptive fusion, and obtain the fused reconstruction data; Calculate the error between the data at the verification point and the fused reconstruction data. When the error value exceeds the preset error threshold, update the interpolation algorithm parameters and return to execute interpolation reconstruction until the error value does not exceed the preset error threshold, and generate the AC impedance distribution surface and temperature distribution surface of the battery cell surface.

[0009] In an alternative embodiment, Identify the abnormal area of the battery cell in the performance distribution map, and determine the performance defect types corresponding to the abnormal area based on the distribution characteristics of the abnormal area and the temperature change data, including: Calculate the Gaussian curvature value of each detection point in the performance distribution map. When the Gaussian curvature value exceeds the background field average value, mark the corresponding point as a performance abnormal feature point. Based on the position information of the performance abnormal feature point, expand the abnormal area using the performance value continuity as the criterion. When the performance value difference between adjacent detection points exceeds the expansion threshold, mark it as a regional boundary point, and connect the regional boundary points to obtain the performance abnormal area; Extract the area, roundness, and irregularity information of the performance abnormal area to generate a morphological feature vector. Calculate the shape tensor of the performance abnormal area based on the morphological feature vector to obtain the expansion characteristics of the performance abnormal area; Collect temperature time-series data based on the boundary range of the performance abnormal area, calculate the temperature rise rate and temperature fluctuation frequency of the temperature time-series data, and obtain the temperature change characteristics of the performance abnormal area; Combine the morphological feature vector, expansion characteristics, and temperature change characteristics of the performance abnormal area to construct a defect feature matrix of the performance abnormal area, and divide the performance abnormal area into defect areas to be classified according to the defect feature matrix; Match the defect areas to be classified with a pre-established battery cell defect feature library through differential feature encoding and a double-layer locally sensitive hashing index structure, and determine the battery cell defect type corresponding to the performance abnormal area based on the matching result.

[0010] In an alternative embodiment, The defective area to be classified is matched with the pre-established battery cell defect feature library through differential feature encoding and a two-layer locally sensitive hashing index structure, and the battery cell defect type corresponding to the performance abnormal area is determined based on the matching result, including: Obtain the morphological features, expansion features, and temperature features of the defective area to be classified and perform feature encoding to generate a feature fingerprint of the defective area to be classified; Construct a two-layer locally sensitive hashing index structure, use a random projection hashing function to perform the first-layer matching of the feature fingerprint with the pre-established battery cell defect feature library to determine candidate defect types, and use an angular sensitive hashing function to perform the second-layer matching of the candidate defect types to obtain a preliminary matching result; Calculate the discriminant information entropy of the morphological features, expansion features, and temperature features at different time scales, determine the feature importance index according to the discriminant information entropy, and dynamically adjust the feature weights of the preliminary matching result based on the feature importance index to obtain an optimized matching result; Construct a feature association graph of the optimized matching result, the feature association graph includes a feature node set, a feature association edge set, and an association weight matrix, and use a spectral clustering method to decompose the feature association graph to obtain feature subgraphs of different defect patterns; Perform repeated feature matching sampling on the defective area to be classified, calculate the feature matching probability distribution, and determine the battery cell defect type corresponding to the defective area to be classified when the highest matching probability is greater than a preset probability threshold.

[0011] In an alternative embodiment, Perform a pulsed discharge test on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculate the capacity loss rate of the abnormal area based on the dynamic response characteristics and the performance defect type, including: Perform a pulsed discharge test on the abnormal area to obtain a voltage response curve, perform wavelet transform on the voltage response curve to obtain a time-frequency domain feature map, extract the frequency features, energy features, and phase features in the time-frequency domain feature map, and construct the dynamic response characteristics of the abnormal area; Perform parameter identification based on electrochemical impedance spectroscopy data to obtain the polarization resistance and capacitance parameters of the abnormal area, and combine the polarization resistance and capacitance parameters with the dynamic response characteristics to form an abnormal area feature parameter set; Project the abnormal area feature parameter set into a high-dimensional feature space, use non-linear feature decomposition to obtain the coupling features between parameters, set a coupling strength threshold according to the performance defect type, extract the coupling features exceeding the coupling strength threshold, and construct a parameter coupling matrix; Determine the capacity attenuation characteristic curve as the reference benchmark according to the type of performance defect, fit the parameter coupling matrix with the capacity attenuation characteristic curve to obtain the actual capacity attenuation trend curve, calculate the deviation of the actual capacity attenuation trend curve relative to the reference benchmark, set different penalty weights for the positive deviation and the negative deviation respectively to construct an asymmetric loss function, and calculate the capacity loss rate of the abnormal area based on the asymmetric loss function.

[0012] In an alternative embodiment, Adjusting the process parameters according to the distribution characteristics of the capacity loss rate includes: Conduct statistical analysis on the capacity loss rate data of non-conforming products, use the kernel density estimation method to obtain the probability density function of the capacity loss rate, extract the peak distribution characteristics of the probability density function, and determine the abnormal process parameters according to the number of peak distribution characteristics; Extract the historical data of the abnormal process parameters from the process parameter records corresponding to the non-conforming products, calculate the change amount of the abnormal process parameters according to the historical data, calculate the sensitivity coefficient according to the corresponding relationship between the change amount and the change amount of the capacity loss rate, and calculate the adjustment amount of the process parameters based on the sensitivity coefficient; Use a PID controller to dynamically optimize the adjustment amount of the process parameters, where the real-time change amount of the capacity loss rate is used as the input signal of the PID controller, the output signal of the PID controller is used as the correction amount of the process parameters, and the process parameters are updated based on the correction amount to realize the adjustment of the process parameters according to the distribution characteristics of the capacity loss rate.

[0013] In the second aspect of the embodiments of the present invention, Provide an intelligent detection system for the processes of an energy storage battery production line, including: The first unit is used to obtain the electrical performance parameters of the battery cells on the energy storage battery production line; The second unit is used to arrange a plurality of detection points along the length direction and the circumferential direction of the battery cell, collect the AC impedance data and temperature data of the battery cell at each detection point, and generate a performance distribution map of the battery cell; The third unit is used to charge the battery cell with constant current and constant voltage to the full charge state, record the temperature change data of the battery cell during the charging process, identify the abnormal area of the battery cell in the performance distribution map, and determine the type of performance defect corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data; The fourth unit is used to select the corresponding pulse discharge test parameters according to the type of performance defect, perform a pulse discharge test on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculate the capacity loss rate of the abnormal area based on the dynamic response characteristics and the type of performance defect; The fifth unit is used to mark the corresponding battery cell as a non-conforming product when the capacity loss rate exceeds a preset loss rate threshold, control the manipulator to transfer the non-conforming product to the defective product collection area, and at the same time adjust the process parameters according to the distribution characteristics of the capacity loss rate, and send the adjusted process parameters to the production line control system.

[0014] In the third aspect of the embodiments of the present invention, a kind of electronic device is provided, including: a processor; a memory for storing instructions executable by the processor; wherein, the processor is configured to call the instructions stored in the memory to execute the method described above.

[0015] In the fourth aspect of the embodiments of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0016] In this embodiment, by obtaining the electrical performance parameters and multi-point AC impedance data of the battery cell, and combining the temperature change data, the abnormal area and the type of performance defect of the battery cell can be accurately identified, improving the accuracy and reliability of the detection. The pulse discharge test is used to obtain the dynamic response characteristics of the abnormal area, and the capacity loss rate is calculated based on this, which can quantitatively evaluate the degree of performance loss of the battery cell and provide an objective basis for determining non-conforming products. Automatically adjusting the process parameters according to the distribution characteristics of the capacity loss rate and feeding them back to the production line control system realizes the closed-loop control of the production process, helps to continuously optimize the production process, and improves the product yield. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic flow chart of the intelligent detection method for the energy storage battery production line process in the embodiments of the present invention; Figure 2 is a simulation diagram of the performance comparison of different interpolation algorithms in this embodiment; Figure 3 is a performance comparison diagram of the double-layer locally sensitive hashing index in the embodiments of the present invention; Figure 4 is a kernel density estimation analysis diagram of the capacity loss rate in the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] The following uses specific embodiments to elaborate in detail on the technical solutions of the present invention. These several specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0020] Figure 1 It is a schematic flowchart of the intelligent detection method for the process of an energy storage battery production line in an embodiment of the present invention. As Figure 1 shown, the method includes: Obtain the electrical performance parameters of battery cells on the energy storage battery production line; Arrange multiple detection points along the length direction and circumferential direction of the battery cell, collect the AC impedance data and temperature data of the battery cell at each detection point, and generate a performance distribution map of the battery cell; Charge the battery cell with constant current and constant voltage to a fully charged state, record the temperature change data of the battery cell during the charging process, identify the abnormal area of the battery cell in the performance distribution map, and determine the performance defect type corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data; Select the corresponding pulse discharge test parameters according to the performance defect type, perform a pulse discharge test on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculate the capacity loss rate of the abnormal area based on the dynamic response characteristics and the performance defect type; When the capacity loss rate exceeds the preset loss rate threshold, mark the corresponding battery cell as a non-conforming product, control the manipulator to transfer the non-conforming product to the defective product collection area, and at the same time adjust the process parameters according to the distribution characteristics of the capacity loss rate, and send the adjusted process parameters to the production line control system.

[0021] The energy storage battery production line refers to an automated or semi-automated production system for mass-producing energy storage battery products, usually including multiple process sections such as cell assembly, liquid injection and encapsulation, formation and grading, and testing and screening. The battery cell is the most basic functional unit in the energy storage battery, usually an independent cell that is not encapsulated into a group. During the detection process, each battery cell is an object of independent detection and evaluation. Its performance directly affects the stability and safety of the final grouped battery.

[0022] Electrical performance parameters include, but are not limited to, indicators such as voltage, current, resistance, internal resistance, conductivity, etc., which are used to quantify the basic electrical characteristics of a single battery cell in static or dynamic operating states. Obtaining these parameters is the basis for subsequent evaluation of the battery performance distribution and identification of defect areas.

[0023] Among them, constant current and constant voltage charging refers to the process of charging the battery with a constant current first during the charging process until the voltage reaches the set value, and then switching to a constant voltage to maintain until the current drops to the cut-off value. It is a commonly used standardized test method for battery activation and performance evaluation. The pulse discharge test parameters include control parameters such as pulse current amplitude, pulse time, and intermittent period, which are adaptively adjusted according to different defect types, aiming to induce and quantify the non-steady-state discharge response behavior of the abnormal area.

[0024] In an optional implementation manner, a plurality of detection points are arranged along the length direction and the circumferential direction of the single battery cell, and the AC impedance data and temperature data of the single battery cell at each detection point are collected. The performance distribution map of the single battery cell includes: A plurality of detection rings are arranged along the length direction of the single battery cell, and a plurality of detection points are evenly arranged in the circumferential direction of each detection ring to construct a detection point array on the surface of the single battery cell. An AC signal is applied to the detection point array, the voltage signal and current signal of the detection point array are collected, and the AC impedance data of the detection point array is calculated; The temperature data of the detection point array is collected, the corresponding relationship between the AC impedance data and the temperature data of the detection point array at each detection point is established, and digital filtering is performed on the AC impedance data and the temperature data to obtain the filtered AC impedance data and filtered temperature data of the detection point array; Based on the distribution coordinates of the detection point array, interpolation algorithms are used to reconstruct the filtered AC impedance data and filtered temperature data to generate the AC impedance distribution surface and temperature distribution surface on the surface of the single battery cell; The AC impedance distribution surface and the temperature distribution surface are normalized to obtain the normalized AC impedance distribution data and normalized temperature distribution data on the surface of the single battery cell, and the AC impedance weight coefficient and temperature weight coefficient of the single battery cell are calculated; The normalized AC impedance distribution data and normalized temperature distribution data are weighted and superimposed according to the AC impedance weight coefficient and temperature weight coefficient to generate the performance distribution map of the single battery cell.

[0025] In this embodiment, data is collected by constructing a detection point array on the surface of the battery cell and a performance distribution map is generated. Specifically, first, a detection ring is set every 50 mm in the length direction of the battery cell, and a detection point is set every 45-degree angle in the circumferential direction for each detection ring. If the length of the battery cell is 500 mm, 10 detection rings are set in the length direction, and 8 detection points are set for each detection ring. A total of 80 detection points form a detection point array. The detection points are made of metal probes, the probe material is selected as copper-gilded, the probe diameter is 1 mm, and the probe forms a 30-degree angle with the battery surface to ensure good electrical contact.

[0026] When an AC signal is applied to the detection point array, an impedance analyzer is used to generate the AC signal. The signal frequency range is from 0.1 Hz to 1000 Hz, the frequency scanning interval is 10 points / decade, and the amplitude of the AC signal is 1 / 20 of the rated capacity of the battery. The voltage signal and current signal of each detection point are respectively collected by a high-precision voltage acquisition module digital multimeter. The sampling frequency is set to 20 times the signal frequency, and the sampling duration is 5 times the signal period. When collecting data, the data of 8 detection points on the first detection ring are collected first, and so on until the data of all detection rings are collected. The AC impedance data of the detection points is calculated by the algorithm built in the impedance analyzer based on the collected voltage signal and current signal.

[0027] At the same time, a PT100 type temperature sensor is installed at each detection point to collect temperature data. The measurement accuracy of the temperature sensor is 0.1 °C, the range is from -50 °C to 150 °C, and the sampling frequency is 1 Hz. The temperature sensor collects data through a temperature acquisition module, and a four-wire connection method is used to eliminate the influence of lead resistance. When establishing the correspondence between the AC impedance data and the temperature data, the number, position coordinates, AC impedance value, and temperature value of the detection points are stored in the SQL database for subsequent data processing and analysis.

[0028] To eliminate the noise interference in the collected data, a band-pass filter is used to filter the AC impedance data and the temperature data. The order of the filter is 4th order, and the passband is from 0.01 Hz to 100 Hz. When filtering, the real part and the imaginary part of the AC impedance data are filtered separately first, and then the filtered AC impedance data is synthesized; the temperature data is directly filtered by the band-pass filter. The filtered data can effectively remove the power frequency interference and high-frequency noise and retain the effective measurement signals.

[0029] When performing data reconstruction based on the distribution coordinates of the detection point array, first establish a three-dimensional coordinate system on the surface of the battery cell. Take the center of one end of the battery cell as the origin, the axial direction as the Z-axis, the radial direction as the R-axis, and the circumferential angle as θ. The coordinates of each detection point can be expressed in the form of (R, θ, Z). Use the cubic spline interpolation algorithm to perform spatial reconstruction on the filtered AC impedance data and temperature data, and the interpolation nodes are the position coordinates of the detection points. In the axial direction, 50 interpolation points are used between adjacent detection rings; in the circumferential direction, 36 interpolation points are used between adjacent detection points to generate continuous AC impedance distribution surfaces and temperature distribution surfaces.

[0030] When normalizing the distribution surface, select the maximum values of the AC impedance data and temperature data respectively as the normalization benchmarks. Taking lithium-ion batteries as an example, at an ambient temperature of 25°C, the range of AC impedance data on the battery surface is 150 mΩ to 180 mΩ, and the range of temperature data is 26°C to 29°C. Taking 180 mΩ and 29°C as the normalization benchmarks for AC impedance and temperature respectively, through normalization, the AC impedance data and temperature data are mapped into the range of 0 to 1. Normalization eliminates the dimensional differences between different physical quantities and facilitates subsequent data fusion.

[0031] The calculation of the weight coefficient is based on the degree of data dispersion, and the data standard deviation is used as the basis for weight allocation. For the above example data, the calculated standard deviation of the normalized AC impedance data is 0.15, and the standard deviation of the normalized temperature data is 0.1. According to the ratio relationship of the standard deviations, the AC impedance weight coefficient is determined to be 0.6, and the temperature weight coefficient is 0.4. The setting of the weight coefficient reflects the influence degree of different parameters on the battery performance, and the parameter with a larger standard deviation has a higher weight.

[0032] The generation of the performance distribution map uses the method of weighted superposition. Multiply the normalized AC impedance distribution data by a weight coefficient of 0.6, and multiply the temperature distribution data by a weight coefficient of 0.4, and add the two to obtain the performance index of each point on the surface of the battery cell. The range of the performance index is 0 to 1, and the jet chromatogram is selected for visualization display, mapping the performance index of 0 to 1 into a gradient color spectrum from blue to red. Among them, the blue area indicates better performance (the performance index is close to 0), and the red area indicates worse performance (the performance index is close to 1).

[0033] In this embodiment, by constructing a detection point array on the surface of the battery cell and collecting multi-dimensional data, the accurate characterization and visualization analysis of the battery performance are achieved. By using high-precision data acquisition equipment and reasonable signal processing methods, the influence of measurement noise is effectively reduced, and the reliability of the data is improved. Through three-dimensional spatial interpolation reconstruction and normalization processing, a continuously distributed performance index surface is obtained, breaking through the limitations of traditional discrete point detection methods. The introduction of weight coefficients takes into account the influence degree of different parameters on the performance, making the performance evaluation more scientific and reasonable. The visualization method based on chromatographic mapping intuitively shows the distribution characteristics of the battery surface performance, facilitating quality control personnel to quickly identify abnormal areas. This solution can be used for quality inspection and screening in the battery production process. By analyzing the performance distribution characteristics, process anomalies can be discovered in a timely manner, guiding process optimization and improving the product yield rate, which has important practical application value.

[0034] In an alternative embodiment, based on the distribution coordinates of the detection point array, an interpolation algorithm is used to reconstruct the filtered AC impedance data and filtered temperature data, and generating the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell includes: Calculating the Euclidean distance between adjacent detection points in the detection point array to generate a distance matrix, calculating the detection point density distribution value based on the distance matrix, comparing the detection point density distribution value with a density threshold for region division, and generating detection region identification data; Performing wavelet transform filtering on the AC impedance data and temperature data, and storing the filtered AC impedance data and filtered temperature data in partitions according to the detection region identification data; Calculating the coefficient of variation of each partition, and allocating an interpolation algorithm according to the comparison result of the coefficient of variation and a preset coefficient of variation threshold. When the coefficient of variation is less than the coefficient of variation threshold, a cubic spline interpolation algorithm is allocated. When the coefficient of variation is greater than or equal to the coefficient of variation threshold, a Kriging interpolation algorithm is allocated; Performing reconstruction operations by executing the allocated interpolation algorithm to obtain partition reconstruction data, constructing a physical constraint optimization function according to the internal resistance continuity and temperature conduction characteristics on the surface of the battery cell, performing conjugate gradient optimization on the partition reconstruction data, calculating the spatial gradient value of the optimized data, determining the grid point fusion window parameter for adaptive fusion, and obtaining the fusion reconstruction data; Calculating the error between the data at the verification point and the fusion reconstruction data. When the error value exceeds the preset error threshold, updating the interpolation algorithm parameters and returning to execute interpolation reconstruction until the error value does not exceed the preset error threshold, and generating the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell.

[0035] Exemplarily, first calculate the Euclidean distance between adjacent detection points in the detection point array to generate a distance matrix. Specifically, any two adjacent detection points can be selected, the coordinate differences in the three-dimensional space are calculated, and then the square root of the sum of squares is obtained as the Euclidean distance. Repeat this process for all adjacent point pairs, and finally obtain an NxN distance matrix, where N is the total number of detection points.

[0036] Calculate the density distribution value of each detection point based on the distance matrix. The kernel density estimation method can be used. Taking each detection point as the center, calculate the distances from other points to this point, and apply the Gaussian kernel function to obtain the density value. For example, the density value of a certain detection point may be 0.85. Then compare the detection point density distribution value with a preset density threshold (such as 0.7) to divide the detection area. Points with density values higher than the threshold are classified into the high-density area, and points with density values lower than the threshold are classified into the low-density area to generate detection area identification data.

[0037] Perform wavelet transform filtering on the originally collected AC impedance data and temperature data. The db4 wavelet can be selected to decompose the data into 5 layers, and after removing the high-frequency noise, the filtered data is reconstructed. According to the detection area identification generated previously, store the filtered data into the corresponding high-density area and low-density area data sets respectively. Calculate the coefficient of variation of the data in each partition. The coefficient of variation is equal to the standard deviation divided by the mean, which reflects the degree of data dispersion. For example, the coefficient of variation of the AC impedance data in a certain high-density area is 0.15, and the coefficient of variation in a certain low-density area is 0.35.

[0038] Compare the calculated coefficient of variation with a preset variation threshold (such as 0.25) to allocate appropriate interpolation algorithms for different regions. For regions with a coefficient of variation less than the threshold, the cubic spline interpolation algorithm is used, and for regions greater than or equal to the threshold, the Kriging interpolation algorithm is used. Execute the allocated interpolation algorithm for reconstruction operations. For cubic spline interpolation, cubic spline functions are constructed in the x and y directions respectively, and then interpolation is performed in the z direction to obtain the reconstructed data. For Kriging interpolation, first calculate the experimental variogram, fit the theoretical variogram model, and then use the Kriging equation for interpolation calculation.

[0039] Construct a physical constraint optimization function according to the internal resistance continuity and temperature conduction characteristics on the surface of the battery cell. Smoothing constraint terms for the internal resistance gradient and temperature gradient, as well as a correlation constraint term for the internal resistance and temperature, can be set. Apply the conjugate gradient method to optimize the partitioned reconstructed data, and iterate the optimization until the objective function converges or reaches the maximum number of iterations.

[0040] Calculate the spatial gradient value of the optimized data and determine the grid point fusion window parameters. The central difference method can be used to calculate the gradient, and the fusion window size can be dynamically adjusted according to the gradient value. The larger the gradient value, the smaller the window. Use the determined window parameters to perform weighted average fusion on the partition boundary data to obtain the fused reconstructed data.

[0041] Select some verification points, compare their measured data with the fused reconstructed data, and calculate the root mean square error. If the error value exceeds the preset threshold (such as 5%), the interpolation algorithm parameters need to be updated and the interpolation reconstruction process needs to be executed again. The algorithm can be updated by adjusting the variogram model parameters of Kriging interpolation or modifying the node positions of cubic splines. Repeat this process until the error value does not exceed the preset threshold.

[0042] Based on the fused reconstructed data that meets the accuracy requirements, use the three-dimensional surface fitting method to generate the alternating current impedance distribution surface and temperature distribution surface on the surface of the battery cell. The bicubic spline surface fitting algorithm can be selected. Divide the grid in the x-y plane, construct bicubic polynomial surface patches for each grid unit, and finally splice them to obtain the complete distribution surface.

[0043] Figure 2 This is the simulation diagram for comparing the performance of different interpolation algorithms in this embodiment. As Figure 2 shown, this figure compares the reconstruction performance of cubic spline interpolation and Kriging interpolation in different data characteristic regions. The left side shows the reconstruction performance in the flat data region, and the right side shows the reconstruction performance in the data mutation region. The curves represent the change trend of the reconstruction error with the interpolation point density. The simulation results verify that cubic spline interpolation has lower computational complexity and good reconstruction accuracy in the flat region, while Kriging interpolation can better retain local features in the mutation region. This adaptive interpolation strategy significantly reduces the computational burden while ensuring the reconstruction accuracy compared with a single interpolation method.

[0044] In the prior art, a single interpolation algorithm is usually adopted to reconstruct the data on the battery surface, without considering the distribution characteristics of detection points and data variability, resulting in insufficient reconstruction accuracy. In this application, regional division is carried out by calculating the detection point density and data variation coefficient, different interpolation algorithms are adaptively allocated according to data characteristics, and a physical constraint optimization function is introduced to optimize the reconstruction result, realizing high-precision reconstruction of the performance distribution on the battery surface. Compared with the prior art, the solution proposed in this application can adapt to the data characteristics of different detection regions, using Kriging interpolation in sparse point regions and regions with large data fluctuations to ensure reconstruction accuracy, and using cubic spline interpolation in dense point regions and regions with stable data to improve calculation efficiency. Through physical constraint optimization and grid point adaptive fusion, it is ensured that the reconstruction result conforms to the continuity of the battery internal resistance distribution and the conduction characteristics of the temperature field, improving the physical rationality of the reconstruction result. The error feedback iteration mechanism ensures that the reconstruction accuracy meets the requirements, and finally realizes the accurate characterization of the performance distribution on the battery surface, providing a reliable basis for battery performance evaluation and quality control.

[0045] In an alternative embodiment, an abnormal area of the battery cell is identified in the performance distribution map, and according to the distribution characteristics and temperature change data of the abnormal area, the types of performance defects corresponding to the abnormal area include: Calculate the Gaussian curvature value of each detection point in the performance distribution map. When the Gaussian curvature value exceeds the mean value of the background field, mark the corresponding point as a performance abnormal feature point. Based on the position information of the performance abnormal feature point, expand the abnormal area with the continuity of the performance value as the criterion. When the difference in performance values between adjacent detection points is detected to exceed the expansion threshold, mark it as a regional boundary point, and connect the regional boundary points to obtain a performance abnormal area; Extract the area, roundness and irregularity information of the performance abnormal area to generate a morphological feature vector, calculate the shape tensor of the performance abnormal area based on the morphological feature vector, and obtain the expansion feature of the performance abnormal area; Collect temperature time-series data based on the boundary range of the performance abnormal area, calculate the temperature rise rate and temperature fluctuation frequency of the temperature time-series data, and obtain the temperature change feature of the performance abnormal area; Combine the morphological feature vector, expansion feature and temperature change feature of the performance abnormal area to construct a defect feature matrix of the performance abnormal area, and divide the performance abnormal area into defect areas to be classified according to the defect feature matrix; Match the defect areas to be classified with a pre-established battery cell defect feature library through differential feature coding and a double-layer locally sensitive hashing index structure, and determine the type of battery cell defect corresponding to the performance abnormal area based on the matching result.

[0046] In this embodiment, the Gaussian curvature values of each detection point in the performance distribution map are first calculated. The Gaussian curvature calculation is achieved by analyzing the performance value change trends of each detection point and its surrounding points. Specifically, for each point (i, j) in the performance distribution map, the points within its surrounding 3×3 region are selected to construct a local surface, and then the Gaussian curvature value K(i, j) of this point is calculated. In practical applications, when the Gaussian curvature value K(i, j) of a certain point exceeds the mean value Kmean of the background field, this point is marked as a performance anomaly feature point. For example, in the above 18650 battery case, the mean value Kmean of the Gaussian curvature of the background field is 0.0023. When the value of K(i, j) of a certain point is detected to be 0.0089, this point is marked as an anomaly feature point.

[0047] Based on the marked performance anomaly feature points, the anomaly region is expanded. Starting from each anomaly feature point, expand in all directions and check the performance values of adjacent points. When the difference in performance values between adjacent points exceeds the preset expansion threshold, this point is marked as a region boundary point. In an actual case, the expansion threshold is set to 2 times the standard deviation of the performance value, which is approximately 0.15. By connecting all boundary points, a closed contour of the performance anomaly region is formed. In the above case, three obvious anomaly regions are identified, which are located in the upper, middle, and lower regions of the battery respectively.

[0048] Extract the morphological features of the performance anomaly region, and calculate the area S, roundness C, and irregularity I of each anomaly region to form a morphological feature vector [S, C, I]. The area S is obtained by calculating the number of pixel points within the region; the roundness C is calculated by comparing the ratio of the perimeter of the region to the perimeter of a circle with the same area; the irregularity I is represented by the standard deviation of the curvature change of the region boundary. In the case, the morphological feature vector of the upper anomaly region is [324, 0.78, 0.35], indicating an area of 324 pixel points, a roundness of 0.78, and an irregularity of 0.35.

[0049] Based on the morphological feature vector, calculate the shape tensor of the performance anomaly region to obtain the expansion feature. The shape tensor describes the main expansion direction and expansion degree of the anomaly region. By analyzing the distribution of points within the region, calculate the eigenvalues and eigenvectors of the shape tensor to obtain the principal axis direction and the ratio of the long axis to the short axis of the region. In the case, the principal axis direction of the middle anomaly region is 67 degrees, and the ratio of the long axis to the short axis is 2.3, indicating that this region expands in an elliptical shape along the 67-degree direction.

[0050] Collect temperature time series data based on the boundary range of the performance anomaly region. Continuously monitor the anomaly region using an infrared thermal imager and record the temperature data every 10 seconds within 30 minutes. Calculate the temperature rise rate dT / dt and the temperature fluctuation frequency f as the temperature change characteristics. In the case, the temperature rise rate of the lower anomaly region is 0.15 °C / minute, and the temperature fluctuation frequency is 0.033 Hz, indicating a slow but continuous heating phenomenon in this region. Combine the morphological feature vector, extended features, and temperature change features to construct the defect feature matrix M of the performance anomaly region. Matrix M contains the geometric morphology, expansion trend, and thermodynamic characteristics of the anomaly region, providing multi-dimensional feature information for defect type identification. In the case, the defect feature matrix of the upper anomaly region is [324, 0.78, 0.35, 43, 1.8, 0.08, 0.012], where the first three values are morphological features, the middle two values are extended features, and the last two values are temperature change features. Finally, through differential feature encoding and a two-layer locally sensitive hashing index structure, match the defect region to be classified with the pre-established battery cell defect feature library. The defect feature library contains common battery defect types such as lithium dendrites, SEI film anomalies, active material shedding, etc., and each defect type has a corresponding feature matrix range. Determine the most matching defect type by calculating the similarity between the feature matrix of the region to be classified and the feature matrices of each type in the library. In the case, the upper anomaly region is identified as a lithium dendrite growth region, the middle anomaly region is identified as an SEI film anomaly region, and the lower anomaly region is identified as an active material shedding region, with similarity rates of 92.3%, 88.7%, and 95.1% respectively.

[0051] In this embodiment, by calculating the Gaussian curvature value of the performance distribution map to identify abnormal feature points and expanding the region based on the continuity of performance values, the precise positioning of the anomaly region is achieved. By extracting the morphological features, extended features, and temperature change features of the anomaly region, a multi-dimensional defect feature matrix is constructed, and an objective defect characterization system is established. Differential feature encoding and a two-layer locally sensitive hashing index structure are used for feature matching, improving the accuracy and efficiency of defect type identification. This solution breaks through the limitations of traditional manual judgment, realizes the automatic identification and classification of battery performance defects, provides reliable technical support for battery quality control and defective product analysis, and at the same time, the established defect feature library can be continuously accumulated and updated to continuously improve the defect identification ability.

[0052] In an alternative embodiment, the battery cell defect type corresponding to the performance anomaly region is determined based on the matching result by matching the defect region to be classified with the pre-established battery cell defect feature library through differential feature encoding and a two-layer locally sensitive hashing index structure, including: Obtain the morphological features, extended features, and temperature features of the defect region to be classified and perform feature encoding to generate the feature fingerprint of the defect region to be classified; Construct a two - layer locally sensitive hashing index structure. Use a random projection hashing function to perform the first - layer matching between the feature fingerprint and a pre - established battery cell defect feature library to determine candidate defect types. Then use an angular - sensitive hashing function to perform the second - layer matching on the candidate defect types to obtain a preliminary matching result; Calculate the discriminant information entropy of the morphological features, extended features, and temperature features at different time scales. Determine the feature importance index according to the discriminant information entropy, and dynamically adjust the feature weights of the preliminary matching result based on the feature importance index to obtain an optimized matching result; Construct a feature association graph of the optimized matching result. The feature association graph includes a feature node set, a feature association edge set, and an association weight matrix. Use a spectral clustering method to decompose the feature association graph to obtain feature sub - graphs of different defect patterns; Perform repeated feature - matching sampling on the defect region to be classified, calculate the feature - matching probability distribution, and determine the battery cell defect type corresponding to the defect region to be classified when the highest matching probability is greater than a preset probability threshold.

[0053] In this embodiment, first, obtain the morphological features, extended features, and temperature features of the defect region to be classified, and perform feature encoding to generate the feature fingerprint of the defect region to be classified. Specifically, the morphological features include geometric parameters such as the area, perimeter, and roundness of the defect region; the extended features include the gray - value distribution, texture features, etc. of the defect region; the temperature features include the highest temperature, average temperature, temperature gradient, etc. of the defect region. When encoding these features, a 256 - bit binary encoding method is used, where the morphological features account for 64 bits, the extended features account for 128 bits, and the temperature features account for 64 bits. For example, the feature fingerprint of a certain defect region to be classified may be: 1010...1101 (morphological features) 1100...0011 (extended features) 0101...1010 (temperature features).

[0054] Construct a two - layer locally sensitive hashing index structure. In the first layer, use a random projection hashing function to match the feature fingerprint with a pre - established battery cell defect feature library to determine candidate defect types. Specifically, when implementing, use 20 8 - bit random projection hashing functions to map the 256 - bit feature fingerprint into 20 8 - bit hash values. Then, search for samples with the same hash value in the feature library as candidate defect types. In the second layer, use an angular - sensitive hashing function to further match the candidate defect types to obtain a preliminary matching result. The angular - sensitive hashing function uses 30 hyper - planes to divide the feature space into different regions, and calculates the probability that the sample to be classified and the candidate sample fall into the same region. The higher the probability, the higher the matching degree.

[0055] Calculate the discriminant information entropy of morphological features, expansion features, and temperature features at different time scales, and determine the feature importance index according to the discriminant information entropy. Specifically, calculate the discriminant information entropy of each feature at time scales of 1 minute, 5 minutes, 30 minutes, and 1 hour. The larger the discriminant information entropy, the stronger the discrimination ability of the feature at that time scale. The discriminant information entropy at different time scales is weighted and averaged to obtain the feature importance index. For example, the importance index of the morphological feature is 0.4, the expansion feature is 0.35, and the temperature feature is 0.25. Dynamically adjust the feature weights of the preliminary matching results based on the feature importance index to obtain the optimized matching results.

[0056] Construct a feature association graph for the optimized matching results. The feature association graph includes a feature node set, a feature association edge set, and an association weight matrix. The feature node set contains all the matched features. The feature association edge set represents the association relationship between features, and the association weight matrix describes the association strength between features. For example, the feature association graph of a certain matching result may contain 10 feature nodes, 30 association edges, and the association weight ranges from 0 to 1. Use the spectral clustering method to decompose the feature association graph to obtain the feature subgraphs of different defect patterns. Spectral clustering first calculates the Laplacian matrix of the graph, then performs eigenvalue decomposition on this matrix, and finally uses the K-means algorithm to cluster the eigenvectors to obtain the feature subgraphs.

[0057] Perform repeated feature matching sampling on the defect area to be classified, and calculate the feature matching probability distribution. Specifically, perform 100 repeated samplings. Each time, randomly select 80% of the features for matching, count the number of times each defect type is matched, and calculate the matching probability. When the highest matching probability is greater than the preset probability threshold (such as 0.8), determine the battery cell defect type corresponding to the defect area to be classified.

[0058] Suppose a suspected defect area is detected on a certain battery production line, and its defect type needs to be determined. First, obtain the features of this area: in terms of morphological features, the area is 2.5 square centimeters, the perimeter is 6.8 centimeters, and the roundness is 0.85; in terms of expansion features, the average gray value is 128, and the texture complexity is 0.6; in terms of temperature features, the highest temperature is 45°C, the average temperature is 42°C, and the temperature gradient is 3°C / cm. Encode these features into a 256-bit binary feature fingerprint. Then use the two-layer locally sensitive hashing index structure for matching. After the first-layer random projection hashing matching, 5 candidate defect types are obtained: crack, bubble, foreign object, wrinkle, and deformation. After the second-layer angular sensitive hashing matching, the preliminary matching results are obtained: crack (matching degree 0.8), bubble (matching degree 0.6), foreign object (matching degree 0.4).

[0059] Next, calculate the feature importance index, obtaining a morphological feature of 0.45, an extended feature of 0.3, and a temperature feature of 0.25. Adjust the matching results based on these weights to obtain the optimized matching results: crack (matching degree 0.75), bubble (matching degree 0.65). Construct a feature association graph, which contains 8 feature nodes and 20 association edges. Use the spectral clustering method to decompose it into 2 feature subgraphs, corresponding to two defect modes of crack and bubble respectively. Finally, perform 100 repeated feature matching samplings to obtain the matching probability distribution: crack 85%, bubble 12%, others 3%. Since the matching probability of the crack, 85%, is greater than the preset threshold of 80%, the type of the defect area is determined to be a crack.

[0060] Figure 3 This is a performance comparison graph of the double-layer locality-sensitive hashing index in the embodiment of the present invention. As Figure 3 shown, this graph shows the query time comparison of different feature matching schemes in the defect recognition of battery cells through four performance curves. The horizontal axis in the graph represents the feature library size, ranging from 1,000 to 100,000; the vertical axis represents the query time, with a range of 0 - 120 milliseconds. Four different schemes are distinguished by marked points in the shapes of square, circle, diamond, and triangle. The double-layer locality-sensitive hashing index structure proposed in this paper adopts a combination strategy of "random projection hashing + angular sensitive hashing", which is represented by triangle marks in the graph. Its query time only increases from 8.2 milliseconds when the feature library has 1,000 entries to 23.8 milliseconds when it has 100,000 entries, and the growth curve is the flattest. The traditional single-layer LSH scheme is shown by square marks, and its query time increases from 12.5 milliseconds to 67.3 milliseconds. The scheme based on cosine similarity is represented by circle marks, and its query time rises from 18.7 milliseconds to 98.6 milliseconds. The scheme based on Euclidean distance is represented by diamond marks, and its query time increases from 21.4 milliseconds to 105.2 milliseconds. The performance curves clearly reflect that the double-layer locality-sensitive hashing index structure significantly improves the query efficiency while ensuring the matching accuracy through the differential encoding of features and the double-layer matching mechanism. Especially in the scenario of a large-scale feature library, this scheme shows obvious performance advantages, with a slow increase in query time and good scalability, providing efficient technical support for the rapid recognition of battery cell defects.

[0061] In the prior art, the matching of battery defect features usually adopts single feature comparison or simple similarity calculation, without considering the temporal variation and correlation of features, resulting in low defect recognition accuracy and susceptibility to noise interference. In this application, by constructing a two-layer locally sensitive hashing index structure, fast coarse matching and accurate matching of features are achieved, greatly improving the matching efficiency. By calculating the discriminant information entropy of features at different time scales to dynamically adjust the feature weights, the matching process can adapt to the change of the importance degree of features. By constructing a feature correlation graph and performing spectral clustering decomposition, the deep correlation relationship between defect features is revealed, improving the accuracy of defect pattern recognition. By adopting the repeated feature matching sampling and probability distribution calculation method, the influence of random error is effectively reduced, enhancing the reliability of defect type determination. This solution breaks through the limitations of traditional single feature matching, establishes a complete multi-feature fusion recognition mechanism, significantly improves the accuracy and robustness of battery defect type recognition, and provides more reliable technical support for battery quality control.

[0062] In an alternative embodiment, a pulse discharge test is performed on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculating the capacity loss rate of the abnormal area based on the dynamic response characteristics and the performance defect type includes: Perform a pulse discharge test on the abnormal area to obtain a voltage response curve, perform wavelet transform on the voltage response curve to obtain a time-frequency domain feature map, extract the frequency feature, energy feature and phase feature in the time-frequency domain feature map, and construct the dynamic response characteristics of the abnormal area; Perform parameter identification based on the electrochemical impedance spectroscopy data to obtain the polarization resistance and capacitance parameters of the abnormal area, and combine the polarization resistance and capacitance parameters with the dynamic response characteristics to form an abnormal area feature parameter set; Project the abnormal area feature parameter set into a high-dimensional feature space, adopt non-linear feature decomposition to obtain the coupling features between parameters, set a coupling strength threshold according to the performance defect type, extract the coupling features exceeding the coupling strength threshold, and construct a parameter coupling matrix; Determine the capacity attenuation characteristic curve as the reference benchmark according to the performance defect type, fit the parameter coupling matrix with the capacity attenuation characteristic curve to obtain the actual capacity attenuation trend curve, calculate the deviation of the actual capacity attenuation trend curve relative to the reference benchmark, set different penalty weights for the positive deviation and the negative deviation respectively to construct an asymmetric loss function, and calculate the capacity loss rate of the abnormal area based on the asymmetric loss function.

[0063] Exemplarily, first, a pulsed discharge test is performed on the abnormal area to obtain a voltage response curve. During the test, a constant-current pulsed discharge method can be used. The pulsed current amplitude is set to 1C, the pulse duration is 10 ms, the pulse interval is 100 ms, and 100 pulses are continuously applied. The voltage response throughout the process is recorded by a high-precision voltage acquisition device with a sampling frequency not lower than 1 kHz to obtain a curve of voltage varying with time. The obtained voltage response curve is subjected to wavelet transform to obtain a time-frequency domain characteristic spectrogram. The db4 wavelet basis function can be selected for 5-layer wavelet decomposition. The time-frequency domain characteristic spectrogram obtained after decomposition contains the energy distribution information of the voltage response at different frequencies and time scales.

[0064] Frequency features, energy features, and phase features are extracted from the time-frequency domain characteristic spectrogram. For the frequency features, the center frequencies of each frequency band can be selected; for the energy features, the energy percentages of each frequency band can be calculated; for the phase features, the phase angles of the signals in each frequency band can be extracted. For example, the center frequencies of 5 frequency bands can be obtained as 10 Hz, 50 Hz, 100 Hz, 500 Hz, and 1 kHz respectively, the corresponding energy percentages are 15%, 25%, 30%, 20%, and 10%, and the phase angles are 30°, 45°, 60°, 90°, and 120° respectively. These features together constitute the dynamic response features of the abnormal area. Then, parameter identification is performed based on the electrochemical impedance spectroscopy data to obtain the polarization resistance and capacitance parameters of the abnormal area. The equivalent circuit fitting method can be used, the Randles equivalent circuit model is selected, and the impedance spectroscopy data is fitted using the least squares method to obtain the values of the polarization resistance Rp and the double-layer capacitance Cdl. For example, the fitted results are Rp = 0.5 Ω and Cdl = 10 mF.

[0065] The polarization resistance, capacitance parameters, and dynamic response features are combined to form a characteristic parameter set for the abnormal area. This parameter set includes the above-extracted frequency features, energy features, phase features, as well as the polarization resistance and capacitance parameters, totaling 13 parameters.

[0066] The characteristic parameter set of the abnormal area is projected into a high-dimensional feature space. The kernel principal component analysis method can be used, and the Gaussian kernel function is selected to map the 13-dimensional parameters to a 100-dimensional feature space. In the high-dimensional feature space, a non-linear feature decomposition method is used to obtain the coupling features between the parameters. The kernelized independent component analysis algorithm can be used to extract the non-linear correlations between the parameters.

[0067] The coupling strength threshold is set according to the type of performance defect. For example, for the problem of lithium-ion battery capacity decay, the coupling strength threshold can be set to 0.7. The coupling features exceeding this threshold are extracted to construct a parameter coupling matrix. This matrix reflects the strong correlations between the parameters, and its dimension may be smaller than the number of original parameters.

[0068] Determine the capacity attenuation characteristic curve as the reference benchmark according to the type of performance defect. For lithium-ion batteries, the capacity attenuation curve under standard cycle life tests can be selected as the benchmark, which usually shows an exponential decay form. Fit the parameter coupling matrix with the capacity attenuation characteristic curve. The support vector regression method can be used, and the radial basis kernel function can be selected to obtain the actual capacity attenuation trend curve.

[0069] Calculate the deviation of the actual capacity attenuation trend curve relative to the reference benchmark. The capacity difference between the two curves can be calculated at multiple cycle number points to obtain a series of deviation values. Different penalty weights are set for the positive and negative deviations respectively to construct an asymmetric loss function. For example, the weight can be set to 0.8 for the positive deviation (the actual capacity is higher than the benchmark) and 1.2 for the negative deviation (the actual capacity is lower than the benchmark), reflecting a conservative estimate of the capacity loss. Calculate the capacity loss rate of the abnormal area based on the asymmetric loss function. Specifically, the weighted average of each deviation point can be used as the final capacity loss rate. For example, if the calculated weighted average deviation is 5%, it means that the capacity loss rate of this abnormal area is 5%.

[0070] In this embodiment, the dynamic response characteristics are obtained through pulse discharge tests, and parameter identification is carried out in combination with electrochemical impedance spectroscopy data to comprehensively characterize the dynamic performance characteristics of the abnormal area. The nonlinear feature decomposition method is used to reveal the coupling relationship between parameters, overcoming the limitation that traditional linear analysis methods are difficult to express complex parameter coupling effects. Select an appropriate capacity attenuation characteristic curve as the reference benchmark according to the type of performance defect, and evaluate the capacity loss through an asymmetric loss function, effectively distinguishing the influence degree of different types of defects on capacity attenuation. The accurate quantitative evaluation of the capacity loss in the abnormal area of the battery is realized, providing a reliable basis for battery performance prediction and life assessment. At the same time, the established evaluation method has strong versatility and can be applied to the analysis of different types of battery defects.

[0071] In practical applications, the parameters in each step can be appropriately adjusted according to the specific battery type and usage scenario. For example, the parameters of the pulse discharge test can be set according to the rated capacity of the battery and the maximum allowable discharge rate; the number of layers of wavelet transform can be selected according to the required frequency resolution; the parameter coupling strength threshold can be determined according to different types of performance defects. Through these flexible adjustments, this method can be applied to various types of batteries and diverse application scenarios.

[0072] Extract the historical data of these two parameters in the recent 8 hours from the process parameter records of the production management system, with a recording interval of 1 minute. For the roll pressure parameter, its set value is 15 MPa, and the actual fluctuation range is from 14.5 MPa to 15.5 MPa; the set value of the coating thickness parameter is 80 μm, and the actual fluctuation range is from 78 μm to 82 μm. By analyzing these historical data, calculate the change trend of the process parameters.

[0073] Pair the change data of the process parameters with the data of the battery capacity loss rate produced during the corresponding period in time series, and establish the corresponding relationship between the parameter change amount and the capacity loss rate change amount. Take the difference between the parameter value at each time point and its previous time point as the parameter change amount, and take the difference between the capacity loss rate of the corresponding batch of batteries and the reference value as the capacity loss rate change amount. Calculate the sensitivity coefficient through these paired data. If the capacity loss rate changes by about 0.5% when the roll pressure changes by 0.1 MPa, the sensitivity coefficient is 5; if the capacity loss rate changes by about 0.3% when the coating thickness changes by 1 μm, the sensitivity coefficient is 0.3. Based on these sensitivity coefficients, combined with the deviation between the current capacity loss rate and the target value, calculate the adjustment amount of the process parameters. For example, when the capacity loss rate is 8% higher than the target value, it is recommended to reduce the roll pressure by 0.2 MPa and increase the coating thickness by 2 μm.

[0074] To achieve smooth adjustment of the process parameters, a PID controller is used for dynamic optimization. The proportional coefficient of the PID controller is set to 0.8, the integral time constant is set to 120 seconds, and the derivative time constant is set to 30 seconds. Take the deviation between the real-time collected capacity loss rate and the target value as the input signal of the PID controller, and the adjustment signal output by the controller is used as the correction amount of the process parameters after amplitude limiting processing. Among them, the correction amount of the roll pressure is limited within the range of ±0.5 MPa, and the correction amount of the coating thickness is limited within the range of ±3 μm.

[0075] In the actual application process, the control system collects the capacity loss rate data every 1 minute. After calculating the correction amount through the PID controller, the corrected parameters are sent to the corresponding device controller through the industrial control computer. After receiving the new parameters, the device controller adjusts the device action through the frequency converter or servo system to achieve real-time update of the process parameters. At the same time, the system will record the change of the capacity loss rate before and after the parameter adjustment for evaluating the adjustment effect.

[0076] Figure 4 This is the kernel density estimation analysis chart of the capacity loss rate in the embodiment of the present invention, as Figure 4As shown, this figure presents the analysis results of the distribution characteristics of the capacity loss rate of defective batteries using the kernel density estimation method. It is clearly shown in the figure that by processing 100 defective sample data with a Gaussian kernel function (bandwidth 0.8), two significant peaks are identified in the capacity loss rate distribution, located at 7% and 12% respectively, and the peak probability densities reach 0.19 and 0.18 respectively. This bimodal distribution characteristic indicates the existence of two main defective patterns, which are difficult to accurately capture by traditional histogram statistical methods. It can be seen from the figure that compared with the stepped distribution presented by the traditional histogram statistical method (using discrete interval statistics), the kernel density estimation method of this technical solution can more accurately describe the continuous distribution characteristics of the data, providing a reliable mathematical basis for the subsequent identification of abnormal process parameters. This figure verifies that the kernel density estimation method in this technical solution can accurately identify the distribution pattern of the capacity loss rate, providing a data basis for locating abnormal process parameters.

[0077] In this embodiment, by analyzing the distribution characteristics of the capacity loss rate through kernel density estimation, accurate identification of defective patterns and accurate positioning of abnormal process parameters are achieved. By establishing the correspondence between the change amount of process parameters and the change amount of the capacity loss rate, calculating the sensitivity coefficient, a quantitative basis for parameter adjustment is established. Using a PID controller for dynamic optimization, real-time automatic adjustment of process parameters is achieved, avoiding the lag and subjectivity of manual adjustment. Adaptive optimization of process parameters based on product quality characteristics is realized, significantly improving the product quality stability, reducing the defective rate, and at the same time, this method has strong versatility and can be extended and applied to the optimization and adjustment of other process parameters.

[0078] In the second aspect of the embodiment of the present invention, a process intelligent detection system for an energy storage battery production line is provided, and the system includes: a first unit for obtaining the electrical performance parameters of battery cells on the energy storage battery production line; a second unit for arranging a plurality of detection points along the length direction and the circumferential direction of the battery cell, collecting the AC impedance data and temperature data of the battery cell at each detection point, and generating a performance distribution map of the battery cell; a third unit for charging the battery cell with constant current and constant voltage to a fully charged state, recording the temperature change data of the battery cell during the charging process, identifying the abnormal area of the battery cell in the performance distribution map, and determining the performance defect type corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data; a fourth unit for selecting corresponding pulse discharge test parameters according to the performance defect type, performing a pulse discharge test on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculating the capacity loss rate of the abnormal area based on the dynamic response characteristics and the performance defect type; The fifth unit is used to mark the corresponding battery cell as a non-conforming product when the capacity loss rate exceeds a preset loss rate threshold, control the manipulator to transfer the non-conforming product to the defective product collection area, and adjust the process parameters according to the distribution characteristics of the capacity loss rate, and send the adjusted process parameters to the production line control system.

[0079] In the third aspect of the embodiments of the present invention, A kind of electronic device is provided, including: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to call the instructions stored in the memory to execute the method described above.

[0080] In the fourth aspect of the embodiments of the present invention, A computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0081] The present invention can be a method, a device, a system and / or a computer program product. The computer program product can include a computer-readable storage medium, on which computer-readable program instructions for executing various aspects of the present invention are carried.

[0082] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent detection method for the process of an energy storage battery production line, characterized in that, Including: Obtain the electrical performance parameters of battery cells on the energy storage battery production line; Arrange multiple detection points along the length direction and circumferential direction of the battery cell, collect the AC impedance data and temperature data of the battery cell at each detection point, and generate a performance distribution map of the battery cell; Charge the battery cell with constant current and constant voltage until it reaches a fully charged state, record the temperature change data of the battery cell during the charging process, identify the abnormal area of the battery cell in the performance distribution map, and determine the performance defect type corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data; Select the corresponding pulse discharge test parameters according to the performance defect type, perform a pulse discharge test on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculate the capacity loss rate of the abnormal area based on the dynamic response characteristics and the performance defect type; When the capacity loss rate exceeds the preset loss rate threshold, mark the corresponding battery cell as a defective product, control the manipulator to transfer the defective product to the defective product collection area, and at the same time adjust the process parameters according to the distribution characteristics of the capacity loss rate, and send the adjusted process parameters to the production line control system.

2. The method according to claim 1, wherein Arranging multiple detection points along the length direction and circumferential direction of the battery cell, and collecting the AC impedance data and temperature data of the battery cell at each detection point to generate a performance distribution map of the battery cell includes: Set multiple detection rings along the length direction of the battery cell, evenly set multiple detection points in the circumferential direction of each detection ring, construct a detection point array on the surface of the battery cell, apply an AC signal to the detection point array, collect the voltage signal and current signal of the detection point array, and calculate the AC impedance data of the detection point array; Collect the temperature data of the detection point array, establish the corresponding relationship between the AC impedance data and temperature data of the detection point array at each detection point, perform digital filtering on the AC impedance data and temperature data, and obtain the filtered AC impedance data and filtered temperature data of the detection point array; Based on the distribution coordinates of the detection point array, use the interpolation algorithm to reconstruct the filtered AC impedance data and filtered temperature data, and generate the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell; Normalize the AC impedance distribution surface and temperature distribution surface to obtain the normalized AC impedance distribution data and normalized temperature distribution data on the surface of the battery cell, and calculate the AC impedance weight coefficient and temperature weight coefficient of the battery cell; Weight and superimpose the normalized AC impedance distribution data and normalized temperature distribution data according to the AC impedance weight coefficient and temperature weight coefficient to generate a performance distribution map of the battery cell.

3. The method according to claim 2, wherein Based on the distribution coordinates of the detection point array, using the interpolation algorithm to reconstruct the filtered AC impedance data and filtered temperature data to generate the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell includes: Calculate the Euclidean distance between adjacent detection points in the detection point array to generate a distance matrix, calculate the detection point density distribution value based on the distance matrix, compare the detection point density distribution value with the density threshold for region division, and generate detection region identification data; Perform wavelet transform filtering on the AC impedance data and temperature data, and store the filtered AC impedance data and filtered temperature data in partitions according to the detection area identification data; Calculate the coefficient of variation of each partition, and allocate an interpolation algorithm according to the comparison result between the coefficient of variation and the preset variation threshold. When the coefficient of variation is less than the variation threshold, allocate the cubic spline interpolation algorithm. When the coefficient of variation is greater than or equal to the variation threshold, allocate the Kriging interpolation algorithm; Execute the allocated interpolation algorithm to perform reconstruction operations to obtain partition reconstruction data, and construct a physical constraint optimization function based on the internal resistance continuity and temperature conduction characteristics on the surface of the battery cell. Perform conjugate gradient optimization on the partition reconstruction data, calculate the spatial gradient value of the optimized data, determine the grid point fusion window parameter for adaptive fusion, and obtain the fusion reconstruction data; Calculate the error between the data at the verification point and the fused reconstruction data. When the error value exceeds the preset error threshold, update the interpolation algorithm parameters and return to execute interpolation reconstruction until the error value does not exceed the preset error threshold, and generate the AC impedance distribution surface and temperature distribution surface on the surface of the battery cell.

4. The method according to claim 1, characterized in that Identify the abnormal area of the battery cell in the performance distribution map, and determine the performance defect types corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data, including: Calculate the Gaussian curvature value of each detection point in the performance distribution map. When the Gaussian curvature value exceeds the average value of the background field, mark the corresponding point as a performance abnormal feature point. Based on the position information of the performance abnormal feature point, expand the abnormal area with the continuity of the performance value as the criterion. When the performance value difference between adjacent detection points is detected to exceed the expansion threshold, mark it as a regional boundary point, and connect the regional boundary points to obtain the performance abnormal area; Extract the area, roundness, and irregularity information of the performance abnormal area to generate a morphological feature vector, calculate the shape tensor of the performance abnormal area based on the morphological feature vector, and obtain the expansion feature of the performance abnormal area; Collect temperature time-series data based on the boundary range of the performance abnormal area, calculate the temperature rise rate and temperature fluctuation frequency of the temperature time-series data, and obtain the temperature change characteristics of the performance abnormal area; Combine the morphological feature vector, expansion feature, and temperature change feature of the performance abnormal area to construct a defect feature matrix of the performance abnormal area, and divide the performance abnormal area into defect areas to be classified according to the defect feature matrix; Match the defect areas to be classified with the pre-established battery cell defect feature library through differential feature coding and a double-layer locally sensitive hashing index structure, and determine the battery cell defect type corresponding to the performance abnormal area based on the matching result.

5. The method according to claim 4, characterized in that, Match the defect areas to be classified with the pre-established battery cell defect feature library through differential feature coding and a double-layer locally sensitive hashing index structure, and determine the battery cell defect type corresponding to the performance abnormal area, including: Obtain the morphological features, expansion features, and temperature features of the defect areas to be classified and perform feature coding to generate the feature fingerprints of the defect areas to be classified; Construct a two - layer locally sensitive hashing index structure. Use a random projection hashing function to perform the first - layer matching between the feature fingerprints and a pre - established battery cell defect feature library to determine candidate defect types. Then use an angular - sensitive hashing function to perform the second - layer matching on the candidate defect types to obtain a preliminary matching result; Calculate the discriminant information entropy of the morphological features, extended features, and temperature features at different time scales. Determine the feature importance index according to the discriminant information entropy, and dynamically adjust the feature weights of the preliminary matching result based on the feature importance index to obtain an optimized matching result; Construct a feature association graph for the optimized matching result. The feature association graph includes a feature node set, a feature association edge set, and an association weight matrix. Use a spectral clustering method to decompose the feature association graph to obtain feature sub - graphs of different defect patterns; Perform repeated feature - matching sampling on the defect region to be classified, calculate the feature - matching probability distribution, and determine the battery cell defect type corresponding to the defect region to be classified when the highest matching probability is greater than a preset probability threshold.

6. The method according to claim 1, characterized in that Perform a pulse - discharge test on the abnormal region to obtain the dynamic response characteristics of the abnormal region. Calculate the capacity loss rate of the abnormal region based on the dynamic response characteristics and the performance defect type, including: Perform a pulse - discharge test on the abnormal region to obtain a voltage response curve. Perform wavelet transform on the voltage response curve to obtain a time - frequency domain feature map. Extract the frequency features, energy features, and phase features from the time - frequency domain feature map to construct the dynamic response characteristics of the abnormal region; Perform parameter identification based on electrochemical impedance spectroscopy data to obtain the polarization resistance and capacitance parameters of the abnormal region. Combine the polarization resistance and capacitance parameters with the dynamic response characteristics to form an abnormal region feature parameter set; Project the abnormal region feature parameter set into a high - dimensional feature space, use non - linear feature decomposition to obtain the coupling features between parameters. Set a coupling strength threshold according to the performance defect type, extract the coupling features that exceed the coupling strength threshold, and construct a parameter coupling matrix; Determine the capacity decay characteristic curve as a reference benchmark according to the performance defect type. Fit the parameter coupling matrix with the capacity decay characteristic curve to obtain the actual capacity decay trend curve. Calculate the deviation of the actual capacity decay trend curve relative to the reference benchmark, set different penalty weights for the positive and negative deviations respectively to construct an asymmetric loss function, and calculate the capacity loss rate of the abnormal region based on the asymmetric loss function.

7. The method according to claim 1, characterized in that Adjust the process parameters according to the distribution characteristics of the capacity loss rate, including: Perform statistical analysis on the capacity loss rate data of unqualified products. Use the kernel density estimation method to obtain the probability density function of the capacity loss rate, extract the peak distribution characteristics of the probability density function, and determine the abnormal process parameters according to the number of peak distribution characteristics; Extract the historical data of the abnormal process parameters from the process parameter records corresponding to the unqualified products. Calculate the change amount of the abnormal process parameters according to the historical data, calculate the sensitivity coefficient according to the corresponding relationship between the change amount and the change amount of the capacity loss rate, and calculate the adjustment amount of the process parameters based on the sensitivity coefficient; The adjustment amount of the process parameters is dynamically optimized by using a PID controller, where the real-time change amount of the capacity loss rate is used as the input signal of the PID controller, the output signal of the PID controller is used as the correction amount of the process parameters, and the process parameters are updated based on the correction amount, so as to realize the adjustment of the process parameters according to the distribution characteristics of the capacity loss rate.

8. Energy storage battery production line process intelligent detection system, used to implement the method described in any one of the foregoing claims 1-7, characterized in that, It includes: The first unit is used to obtain the electrical performance parameters of battery cells on the energy storage battery production line; The second unit is used to arrange a plurality of detection points along the length direction and circumferential direction of the battery cell, collect the AC impedance data and temperature data of the battery cell at each detection point, and generate a performance distribution map of the battery cell; The third unit is used to charge the battery cell with constant current and constant voltage to the full charge state, record the temperature change data of the battery cell during the charging process, identify the abnormal area of the battery cell in the performance distribution map, and determine the performance defect type corresponding to the abnormal area according to the distribution characteristics of the abnormal area and the temperature change data; The fourth unit is used to select the corresponding pulse discharge test parameters according to the performance defect type, perform a pulse discharge test on the abnormal area to obtain the dynamic response characteristics of the abnormal area, and calculate the capacity loss rate of the abnormal area based on the dynamic response characteristics and the performance defect type; The fifth unit is used to mark the corresponding battery cell as a non-conforming product when the capacity loss rate exceeds the preset loss rate threshold, control the manipulator to transfer the non-conforming product to the defective product collection area, and at the same time adjust the process parameters according to the distribution characteristics of the capacity loss rate, and send the adjusted process parameters to the production line control system.

9. An electronic device, characterized in that, It includes: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, the method according to any one of claims 1 to 7 is realized.

Images