Solid-solid interface structure consistency evaluation method of all-solid-state battery

Abstract

The application relates to the technical field of solid-state battery structure evaluation, in particular to a solid-solid interface structure consistency evaluation method of a full solid-state battery, which comprises the following steps: scanning a full solid-state battery sample based on a CT device, obtaining a two-dimensional slice sequence, and sequentially numbering; calculating the definition index and the brightness uniformity index of each two-dimensional slice, and generating a quality score; selecting an effective analysis slice to calculate the average value of each row of gray scale and the gray scale gradient curve, determining the gradient peak position; and calculating the battery layer thickness and the average thickness of the battery layer; calculating the column direction average gray scale distribution curve of each effective analysis slice, and determining the boundary of the electrode area and the rectangular area surrounded by the boundary by using a threshold-gradient joint determination method; calculating the average consistency score of all effective analysis slices for the rectangular area of each effective analysis slice, and generating a structured evaluation report. The application realizes quantitative analysis and hierarchical evaluation of the internal interface structure of the battery.

Description

Technical Field

[0001] This application relates to the field of solid-state battery structure evaluation technology, specifically to a method for evaluating the consistency of the solid-solid interface structure of an all-solid-state battery. Background Technology

[0002] All-solid-state batteries are considered the main development direction for next-generation power batteries due to their high energy density and safety. However, the electrochemical performance of all-solid-state batteries largely depends on the tight contact between the electrode and the solid electrolyte at the solid-solid interface, and the interface consistency has a significant impact on it. Currently, the characterization of the solid-solid interface mainly relies on scanning electron microscopy (SEM) or transmission electron microscopy (TEM). However, these techniques are complex to prepare, can only obtain local information, and cannot be used to monitor the evolution of the solid-solid interface as the battery ages.

[0003] Therefore, there is a need to develop new methods for solid-solid interface characterization based on non-destructive testing techniques. Computed tomography (CT) can achieve non-destructive three-dimensional imaging, allowing observation of the layered arrangement and contact uniformity of internal components. However, it lacks standardized image screening and consistency analysis procedures. Current CT result processing is only qualitative, not quantitative, and the measurement results are highly subjective and difficult to use for process optimization or quality evaluation.

[0004] Therefore, it is necessary to design a solid-solid interface structure consistency evaluation method for all-solid batteries, and establish a complete method system from CT image acquisition to quantitative consistency indicators, so as to achieve an objective and highly accurate evaluation of the solid-solid interface adhesion, flatness and stacking uniformity. Summary of the Invention

[0005] In view of this, this application proposes a solid-solid interface structure consistency evaluation method for all-solid batteries, aiming to solve the problems of lack of standardized image screening and consistency analysis process, CT result processing being only qualitative and not quantitative, and measurement results being highly subjective and difficult to use for process optimization or quality evaluation.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] This application provides a method for evaluating the solid-solid interface structure consistency of all-solid-state batteries, including:

[0008] The solid-state battery sample was scanned using a CT scanner to obtain a two-dimensional slice sequence along the thickness direction; and the two-dimensional slice sequence was sequentially numbered.

[0009] The sharpness index and brightness uniformity index of each two-dimensional slice are calculated based on the two-dimensional slice sequence, and the weighted sum is used to generate a quality score and select valid analysis slices.

[0010] The average grayscale curve of the image rows is acquired for the effective analysis slice, and the average grayscale value and grayscale gradient curve of each row are calculated to determine the gradient peak position; and the battery layer thickness and the average battery layer thickness are calculated.

[0011] Calculate the column-direction average gray-scale distribution curve for each effective analysis slice, and use the threshold-gradient joint judgment method to determine the boundary of the electrode region and the rectangular region enclosed by the boundary.

[0012] For each valid analytical slice, the rectangular region is calculated for average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity; thereby, the average consistency score of all valid analytical slices is calculated; the scoring level is determined based on the average consistency score, and a structured evaluation report is generated.

[0013] Optionally, when calculating the sharpness index and the brightness uniformity index, the following are included:

[0014] Image sharpness was evaluated using Laplacian variance to obtain a sharpness index, which was then normalized to generate a sharpness score. Brightness uniformity was obtained by calculating the standard deviation of image grayscale values ​​and normalized to generate a uniformity score. Statistical analysis was then performed on the normalized sharpness and uniformity scores.

[0015] Optionally, when selecting valid analysis slices, the following are included:

[0016] The slice sequence was evenly divided into several equal-length segments according to the battery thickness. Each equal-length segment contained an equal number of two-dimensional slices. The four slices with the highest quality scores in each equal-length segment were selected, for a total of 20 slices as valid analysis samples.

[0017] Optionally, determining the location of the gradient peak includes:

[0018] The first gradient peak corresponds to the upper boundary line of the battery, and the last gradient peak corresponds to the lower boundary line of the battery. Moving average filtering is used to smooth the upper and lower boundary lines, eliminating random fluctuations in the boundary lines and preserving structural features.

[0019] Optionally, when using the threshold-gradient joint determination method to determine the boundary of the electrode region and the rectangular region enclosed by the boundary, the following steps are included:

[0020] Based on the column-direction average gray-scale distribution curve, identify the left and right edge regions where the brightness drops to a preset brightness threshold, define the bright region as the effective range of the image, and determine the boundary and the rectangular region enclosed by the boundary.

[0021] Specifically, when the grayscale gradient is less than the set gradient threshold and the average grayscale is consistently less than 80% of the average of the entire image, the region is determined to be an invalid edge region.

[0022] Optionally, when calculating the average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity for each rectangular region of the effective analysis slice, the following steps are included:

[0023] Calculate the arithmetic mean of the gray values ​​of all pixels within the rectangular area of ​​each effective analysis slice to obtain the average gray value;

[0024] The standard deviation of grayscale values ​​is obtained by calculating the dispersion of grayscale values ​​of all pixels within the rectangular region of each effective analysis slice from their average value.

[0025] The degree of dispersion of the battery layer thickness value within the rectangular region of each effective analysis slice is calculated to obtain the thickness fluctuation;

[0026] The connectivity coefficient is obtained by calculating the ratio of the area of ​​the largest connected region within the rectangular region of each effective analysis slice to the total area of ​​the entire interface region.

[0027] The porosity is calculated by counting the proportion of pixels with gray values ​​less than the gray value threshold within the rectangular area of ​​each effective analysis slice to the total number of pixels.

[0028] Optionally, when calculating the average consistency score for all the valid analysis slices, the following steps are included:

[0029] The calculated average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity are normalized to obtain normalized parameters.

[0030] The normalized parameters are weighted and fused to generate a consistency score. The average consistency score is generated by calculating the average of the consistency scores of all the valid analysis slices.

[0031] Optionally, when classifying rating levels based on the average consistency score, the following steps are included:

[0032] The rating levels include A, B, C, D, and E; the average consistency score is a natural number from 0 to 1.

[0033] When 1 ≥ the average consistency score > 0.9, it is classified as Grade A;

[0034] When 0.9 ≥ the average consistency score > 0.8, it is classified as Grade B;

[0035] When 0.8 ≥ the average consistency score > 0.7, it is classified as Grade C;

[0036] When 0.7 ≥ the average consistency score > 0.6, it is classified as Grade D;

[0037] When 0.6 ≥ the average consistency score ≥ 0, it is classified as Grade E.

[0038] Optionally, the method further includes marking the region corresponding to the valid analysis slice as a structural anomaly region if the consistency score deviates from the average consistency score by more than 0.1; and spatially locating the structural anomaly region to determine its specific location within the battery.

[0039] Optionally, when generating a structured assessment report, the following may be included:

[0040] The consistency score, rating level, interface structure parameters, and abnormal area identifiers are integrated into a structured evaluation report; the structured evaluation report includes text descriptions and visual charts.

[0041] According to the specific embodiments provided in this application, this application has the following technical effects:

[0042] This application provides a solid-solid interface structure consistency evaluation method for all-solid-state batteries. A unified algorithm is used for slice selection, boundary detection, and parameter extraction, avoiding subjective human error. Interface smoothness, density, uniformity, and defect ratio are comprehensively converted into a consistency score, reflecting the dynamic evolution of the internal structure during battery cycling. A unified grade and trend curve are output, facilitating batch comparison and standard establishment. Battery samples are scanned using CT equipment with a resolution not exceeding 5 micrometers, ensuring sufficiently fine slice images and providing high-quality data for structural consistency analysis. By calculating multiple indicators such as sharpness and brightness uniformity, a comprehensive analysis of the solid-solid interface structure of the battery can be performed, ensuring consistency and stability in different regions. Weighted summation generates quality scores, and parameters such as gradient, porosity, and thickness fluctuation of each slice are analyzed, forming a multi-dimensional quality assessment system. Accurate calculation of the battery layer thickness and its fluctuations, and determination of the left and right boundaries of the electrode region, improves the accuracy of the evaluation. For potentially abnormal areas in the battery structure, consistency score deviation analysis identifies and locates structural anomalies, revealing potential structural problems. By combining consistency scores, rating levels, and structural parameters, an easy-to-understand and structured evaluation report is generated. The use of clearly defined rating levels (A to E) provides a quantitative standard for battery quality control and production. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart of a solid-solid interface structure consistency evaluation method for all-solid-state batteries provided in this application embodiment;

[0045] Figure 2 A schematic diagram illustrating the upper and lower boundary identification of the solid-solid interface structure consistency evaluation method for all-solid batteries provided in this application embodiment;

[0046] Figure 3 A schematic diagram of left and right boundary identification for the solid-solid interface structure consistency evaluation method of the all-solid battery provided in the embodiments of this application;

[0047] Figure 4 A schematic diagram of brightness curve recognition for the solid-solid interface structure consistency evaluation method of the all-solid battery provided in the embodiments of this application. Detailed Implementation

[0048] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0049] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] See Figure 1 As shown, this application proposes a method for evaluating the solid-solid interface structure consistency of all-solid-state batteries, including:

[0051] S100: Scan the all-solid-state battery sample using a CT scanner to obtain a two-dimensional slice sequence along the thickness direction; and sequentially number the two-dimensional slice sequence.

[0052] S200: Calculate the sharpness index and brightness uniformity index of each two-dimensional slice based on the two-dimensional slice sequence, and generate a quality score by weighted summation and select valid analysis slices.

[0053] S300: Acquire the average grayscale curve of the image rows for the effective analysis slice, calculate the average grayscale value and grayscale gradient curve of each row, determine the gradient peak position, and calculate the battery layer thickness and the average thickness of the battery layer.

[0054] S400: Calculate the column-direction average gray-scale distribution curve for each effective analysis slice, and use the threshold-gradient joint judgment method to determine the boundary of the electrode region and the rectangular region enclosed by the boundary.

[0055] S500: For each rectangular region of the effective analysis slice, calculate the average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity; thereby calculate the average consistency score of all the effective analysis slices; classify the scoring level according to the average consistency score, and generate a structured evaluation report.

[0056] The formula for calculating the quality score using weighted summation is as follows: ,in, Rate the clarity. For uniformity scoring. The formula for calculating the cell layer thickness is: The formula for calculating the average thickness of the battery layer is: ;in, For the upper boundary line and This is the lower boundary line.

[0057] Specifically, a CT scanner with a resolution of no more than 5 micrometers was used to scan the battery sample along its thickness, obtaining at least 80 two-dimensional slice sequences. The slice sequences were then subjected to grayscale normalization, and a pixel-to-length conversion coefficient was calculated based on the image scale. The calculation formula is as follows: ; The actual length marked on the ruler. The pixel length corresponding to the scale is used to convert the pixel scale into the actual physical length. The slices are also sequentially numbered. The sharpness index and luminance uniformity index are calculated for each slice, normalized to generate sharpness and uniformity scores, and then applied according to the formula... A weighted summation is performed. The entire sequence is then divided into several equal-length segments, and the highest-quality images within each segment are selected as valid analysis slices. The average grayscale curve of each row is extracted from the valid slices, and the average grayscale value and grayscale gradient curve are calculated. The upper and lower boundaries of the image are determined based on the gradient peaks, with the first peak corresponding to the upper boundary and the last peak corresponding to the lower boundary. The extracted boundary lines are then smoothed according to the formula... Calculate the local thickness of the battery layer, and further... The average thickness of the battery layer is determined. The average grayscale distribution along the column direction of each valid slice is calculated, and the electrode region boundaries are identified using a threshold-gradient joint judgment method, resulting in rectangular analysis regions for the electrode regions. Within each rectangular region, structural consistency indicators such as average grayscale, grayscale standard deviation, thickness fluctuation, connectivity coefficient, and porosity are calculated. The results of all valid slices are averaged to obtain the overall consistency score of the entire cell. Based on this score, the cells are classified into levels, and a structured evaluation report is automatically generated, achieving a quantitative and standardized assessment of the solid-solid interface structural consistency of all-solid-state batteries.

[0058] This application addresses the challenges of quantitatively characterizing the solid-solid interface structure of all-solid-state batteries, inconsistent slice quality, difficulties in interface identification, and the lack of unified standards for evaluation metrics. It provides a complete and repeatable structural consistency evaluation process, resolving the issue of inconsistently evaluating the interface structure of all-solid-state batteries.

[0059] The working process and principle of this application are as follows: A CT scanner with a resolution ≤5μm is used to scan the all-solid-state battery layer by layer along its thickness direction to obtain a numbered two-dimensional slice sequence. A scale bar is read from the image to calculate the pixel-to-length conversion coefficient, converting the pixel scale to the actual physical length. Gray-level normalization is performed on all slices to eliminate deviations under different scanning conditions. For each slice, image sharpness (using the variance of the Laplacian operator response to characterize sharpness) and brightness uniformity (gray-level standard deviation to characterize uniformity) are calculated. These two indicators are linearly normalized into sharpness and uniformity scores (inverse normalization is used for uniformity to make "higher values ​​better"), and a quality score is synthesized according to weights. The slice sequence is uniformly divided into several equal-length segments according to thickness. Within each segment, several images with the highest quality scores (e.g., 4 images per segment, 20 images in total) are selected as effective analysis slices to suppress errors caused by low-quality images. For each effective slice, the row mean curve is calculated along the column direction, and its discrete gradient is calculated. The upper and lower boundaries are located by identifying the first and last significant peaks on the gradient curve (see details). Figure 2 , Figure 2 To utilize CT scans of cross-sectional images of the internal structure of solid-state batteries and to illustrate the identification of upper and lower boundaries (the red and green lines in the image represent the computer's identification results according to the method of this invention), a moving average smoothing process is applied to the boundary function to remove random noise and retain structural features. Local thickness is calculated column by column, and the average thickness is calculated for the entire image or all valid slices to obtain thickness statistics and fluctuation indicators. A threshold-gradient joint judgment method is adopted, combining gradient thresholds and grayscale thresholds along the row direction: when a certain grayscale gradient is low and the average grayscale is consistently lower than 80% of the average of the entire image, it is judged as an invalid edge region. After exclusion, the left and right boundaries of the electrodes are determined in the remaining valid columns through gradient abrupt changes or brightness decrease points (see details). Figure 3 , Figure 3To utilize cross-sectional images of the internal structure of a solid-state battery obtained from CT scans and to illustrate the identification of its left and right boundaries (the blue lines in the image represent the computer's identification results according to the method of this invention), a rectangular analysis area is defined. Within each rectangular area, a set of structural parameters are calculated: average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient (the ratio of the area of ​​the largest connected region to the total area), and porosity (based on a gray level threshold to determine the number of pore pixels). These raw parameters are normalized and (if necessary, the "smaller the better" items are inverted) and weighted according to preset weights to obtain the consistency score for each slice. The average of all valid slices is then taken as the overall consistency score of the sample. Based on the threshold range, the slices are classified into levels, and any slice with a deviation exceeding 0.1 is marked as a structural anomaly area. Three-dimensional spatial positioning is performed based on the slice index and pixel-length coefficient. Finally, a structured evaluation report containing a numerical table, boundary and thickness curves, porosity and connectivity diagrams, and classification conclusions is automatically generated, thus achieving a closed-loop evaluation from image acquisition, quality screening, boundary identification to multi-index quantification and anomaly localization.

[0060] As a preferred embodiment, the specific implementation of this application is as follows: In a consistency evaluation of an all-solid-state battery sample, a sulfide-based all-solid-state battery with a size of approximately 20mm × 10mm × 1mm was selected as the test object. The sample was scanned using an industrial CT scanner with a resolution of 3μm, obtaining 120 two-dimensional slices along the thickness direction, meeting the requirement of at least 80 slices. All slices were subjected to grayscale normalization, and the pixel-to-length conversion coefficient was calculated to be 0.003mm / pixel (i.e., 1mm corresponds to 333 pixels) using a 1mm scale in the CT image. All slices were numbered according to the scanning order. When evaluating the slice quality, the sharpness index and brightness uniformity index of each slice were automatically calculated. For example, the sharpness score of the 15th slice was 0.82, and the uniformity score was 0.76. According to the formula Quality Score = 0.6·Sharpness + 0.4·Uniformity, the quality score is 0.79. After dividing all 120 slices into 5 segments, the 4 highest-quality slices from each segment, totaling 20 slices, were selected as valid slices for subsequent analysis. During the structural boundary identification stage, the row average grayscale curves were extracted from these valid slices, and the grayscale gradient was calculated. Taking slice 42 as an example, the first peak of its row gradient curve appeared at pixel row 58, corresponding to the upper boundary line; the last peak appeared at pixel row 496, corresponding to the lower boundary line. After smoothing the two boundary lines, the local thickness of this slice was calculated to be 438 pixels (approximately 1.314 mm). The average thickness of the 20 slices was further calculated to be 1.307 mm, with thickness fluctuations below 1%, indicating good interlayer uniformity. The column-direction average grayscale curve for each slice was calculated, and a threshold-gradient joint judgment method was used to automatically identify positive and negative polarity regions. For example, in slice 66, the algorithm identified the positive polarity region as rectangle A (1580 pixels wide, 420 pixels high) and the negative polarity region as rectangle B (1590 pixels wide, 415 pixels high). A quantitative analysis of structural consistency was performed on each rectangular region. Taking the positive electrode region A as an example, its average gray level was 142.3, the standard deviation of gray level was 6.8, the connectivity coefficient was as high as 0.94, the porosity was only 2.1%, and the thickness fluctuation was 0.7%. After averaging all 20 valid slices, the consistency score was 0.92, and the sample was judged as "Grade A". A structured evaluation report was automatically generated, including slice illustrations, boundary curves, thickness statistics tables, gray level index curves, and consistency grading explanations.

[0061] Through the above-mentioned scheme, this application obtains sufficient slices using high-resolution CT and performs grayscale normalization and pixel-to-length conversion, enabling standardized processing of images under different sample and equipment conditions; it generates quality scores by weighting clarity and brightness uniformity, and selects the best slices by segment, solving the analytical bias caused by uneven slice quality; it extracts the upper and lower boundaries of the battery layer and quantifies the thickness and average thickness through gradient peak detection and boundary smoothing algorithms, solving the problems of large errors and unclear interfaces in traditional manual judgment; it automatically identifies electrode regions using a threshold-gradient joint judgment method, avoiding the non-repeatability of subjective division; it calculates multi-dimensional structural parameters such as grayscale, thickness fluctuation, porosity, and connectivity, and generates consistency scores and grading reports, realizing the transformation of solid-solid interface structure from "qualitative observation" to "quantitative evaluation".

[0062] This application further proposes the following methods for calculating sharpness and luminance uniformity indices:

[0063] Image sharpness was evaluated using Laplacian variance to obtain a sharpness index, which was then normalized to generate a sharpness score. Brightness uniformity was obtained by calculating the standard deviation of image grayscale values ​​and normalized to generate a uniformity score. Statistical analysis was then performed on the normalized sharpness and uniformity scores.

[0064] Specifically, the process of calculating the sharpness index and the brightness uniformity index first assesses the image sharpness using the Laplacian variance to obtain the sharpness index. The Laplacian operator calculates the second derivative of the image and its variance, reflecting the image's details and edge information. The sharpness index obtained through this calculation is normalized to ensure its range is between [0,1], where a larger value indicates a sharper image. Secondly, the brightness uniformity index reflects the dispersion of gray values ​​in the image by calculating the standard deviation of gray levels; a smaller standard deviation indicates a more uniform brightness distribution. This index is also normalized to ensure its result is within the range of [0,1], with a larger value indicating more uniform brightness. Finally, the normalized sharpness score and uniformity score are combined and weighted to obtain a comprehensive quality score for each slice.

[0065] Through the above technical solution, this application evaluates image sharpness using Laplacian variance, calculates a sharpness index, and generates a sharpness score through normalization processing. This reflects the details and edges of the battery image, improving the recognition accuracy of image quality. Calculating the grayscale standard deviation of the image as a brightness uniformity index can assess the consistency of the image brightness distribution, ensuring the uniformity of the battery structure.

[0066] This application further proposes dividing a two-dimensional slice sequence into several equal-length segments, and selecting several slices with the highest quality scores from each segment as valid analysis slices, including:

[0067] The slice sequence was evenly divided into several equal-length segments according to the battery thickness. Each equal-length segment contained an equal number of two-dimensional slices. The four slices with the highest quality scores in each equal-length segment were selected, for a total of 20 slices as valid analysis samples.

[0068] Specifically, when dividing the two-dimensional slice sequence into several equal-length segments, the sequence is uniformly divided into five segments based on the battery thickness, with each segment containing the same number of two-dimensional slices. The total thickness of the battery is measured, and the thickness range of each segment is determined accordingly. Slices are evenly distributed into these five segments along the thickness direction, ensuring that each segment contains the same number of slices. This ensures a uniform evaluation of the entire battery sample, unaffected by local thickness differences. Within each segment, the slices with the highest quality scores (such as sharpness and uniformity) are selected as valid analytical slices. The top four quality scores from each segment are selected, for a total of 20 slices, which are then chosen as valid analytical samples. This method ensures that the optimal slices in each segment are selected, thereby improving the representativeness and accuracy of the analysis.

[0069] By employing the aforementioned technical solution, this application divides the two-dimensional slice sequence into several equal-length segments and selects the slice with the highest quality score from each segment, ensuring the comprehensiveness and representativeness of the evaluation results for the all-solid-state battery samples. Uniform segmentation ensures that slices from different thicknesses of the battery can be fully analyzed, avoiding the omission of information from localized areas. Selecting the slice with the highest quality score as the effective analysis sample helps improve analytical accuracy, eliminating potential image noise or poor-quality slices, thereby enhancing the reliability and accuracy of the evaluation results.

[0070] This application further proposes methods for smoothing the upper and lower boundary lines, including:

[0071] Moving average filtering is used to smooth the upper and lower boundary lines, eliminating random fluctuations in the boundary lines while preserving structural features.

[0072] Specifically, the upper and lower boundary lines are defined as the contours of the battery edges in the image. These edges may exhibit random fluctuations due to image noise, resolution limitations, or differences in shooting angle. To reduce the impact of this noise on the analysis results, the boundary lines need to be smoothed. Moving average filtering is a commonly used smoothing method that reduces local fluctuations by calculating the average of data points near the boundary lines. For each data point on the upper and lower boundary lines (such as the edge position of each slice), a sliding window of fixed size is used to calculate the average of that point and its neighbors. This process typically involves averaging several neighboring points (e.g., 5 or 7 points) around each point to reduce the impact of single outliers; using this average to replace the original data point, thereby smoothing the entire boundary line and eliminating short-term or local random fluctuations; this process is repeated for all boundary points, smoothing the entire upper and lower boundary lines point by point.

[0073] Through the above technical solution, this application removes minor fluctuations caused by image noise or resolution limitations through smoothing processing, ensuring the stability and consistency of the boundary lines, thereby reflecting the actual structure of the battery. This process maintains the overall characteristics of the battery interface while avoiding the impact of local anomalies on battery thickness and electrode area division.

[0074] This application further proposes a threshold-gradient joint determination method for determining the boundary of the electrode region and the rectangular region enclosed by the boundary, including:

[0075] Based on the average gray-scale distribution curve in the column direction Identify the left and right edge regions where brightness decreases significantly, define the bright areas as the effective range of the image, and determine the boundaries and the rectangular regions (ROI, Region of Interest) enclosed by the boundaries.

[0076] Among them, when the gray-level gradient When the gradient value is less than the set gradient threshold and the average gray value is consistently less than 80% of the average value of the entire image, the region is determined to be an invalid edge region.

[0077] Specifically, see Figure 4To identify the brightness curves of the left and right boundaries of the battery, this invention utilizes the analysis curves used in the boundary identification process. The left and right boundaries are determined based on abrupt change points in the curves. The average grayscale distribution curve in the column direction of the image is analyzed to identify areas where grayscale values ​​decrease significantly. These areas typically represent the left and right boundaries of the electrode region, as the electrode region may have different brightness characteristics compared to other parts. By identifying these regions, the left and right boundaries of the electrode region can be preliminarily determined. A "bright area" is defined as the effective range in the image. Bright areas typically represent electrode regions, where higher grayscale values ​​reflect the presence of electrodes. Using this feature, combined with the image's gradient information, the boundary localization is further refined, forming the ROI (Region of Interest). The boundary changes are judged by calculating the grayscale gradient (i.e., the rate of change of grayscale values). When the grayscale gradient in the image is less than a preset gradient threshold, and the average grayscale value of this region is consistently lower than 80% of the overall image average, this region can be determined as an invalid edge region, i.e., not part of the electrode region.

[0078] In determining the left and right boundaries of the electrode region and the rectangular region using the threshold-gradient joint judgment method, the gradient threshold plays a crucial role. The gradient threshold is a preset critical value for the rate of gray-level change, used to determine whether the change in image gray-level in the horizontal direction is significant. When the local gradient value of the average gray-level distribution curve in the column direction is higher than this threshold, it indicates a large gray-level change, possibly corresponding to the boundary of the electrode region; conversely, when the gradient value is lower than the threshold and the average gray-level of this region is consistently lower than 80% of the average of the entire image, the region is determined to be an invalid edge region, i.e., a non-electrode region. The gradient threshold can be set according to the dynamic range and noise level of the image. By statistically analyzing the mean and standard deviation of the gray-level gradient of the entire image, an appropriate critical value can be selected to balance the sensitivity and robustness of boundary detection. Using this threshold, the left and right edge regions with significantly decreased brightness can be accurately identified, the bright area in the middle can be defined as the effective electrode range, and a rectangular analysis area can be formed.

[0079] By combining the column-direction average grayscale distribution curve and grayscale gradient information, this application identifies edge regions with significantly decreased brightness and eliminates invalid edge regions with low grayscale gradients and persistently low brightness, thus avoiding the influence of noise or background interference on electrode region localization. The bright areas determined in this way serve as the effective range of the image, reflecting the actual electrode structure and ensuring that the rectangular analysis area covers the entire electrode region.

[0080] This application further proposes that, for each valid analytical slice's rectangular region, when calculating the average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity, the following should be included:

[0081] Calculate the arithmetic mean of the gray values ​​of all pixels within the rectangular region of each valid analysis slice to obtain the average gray value. ; , used to reflect density and compactness;

[0082] The standard deviation of grayscale values ​​is obtained by calculating the dispersion of grayscale values ​​of all pixels within the rectangular region of each effective analysis slice from their mean value. ; Used to reflect uniformity;

[0083] The dispersion of the battery layer thickness values ​​within the rectangular region of each effective analysis slice is calculated to obtain the thickness fluctuation. That is, to calculate each effective analytical slice. The average thickness of the battery layer; used to reflect the flatness of the stacking.

[0084] The connectivity coefficient is obtained by calculating the ratio of the area of ​​the largest connected region within the rectangular region of each valid analysis slice to the total area of ​​the entire interface region. ; ,in, The area of ​​the maximum connected region. This represents the total area of ​​the interface region; it is used to describe the degree of interface continuity.

[0085] The porosity is calculated by counting the proportion of pixels with gray values ​​less than a gray value threshold within the rectangular region of each valid analysis slice out of the total number of pixels. ; ;in, It is the number of pixels whose grayscale value is less than the grayscale threshold (the grayscale threshold is generally taken as 60-70% of the average brightness of the rectangular area). It represents the total number of pixels; used to reflect the proportion of interface defects.

[0086] Specifically, when analyzing the rectangular region of each effective analysis slice, the following five key indicators are calculated to comprehensively evaluate the structural characteristics and quality of the battery interface.

[0087] Average grayscale: The average grayscale is calculated by averaging the grayscale values ​​of all pixels within a rectangular area. This metric reflects the density and compactness of the battery interface. A higher average grayscale value indicates a denser material, which helps improve battery stability and performance.

[0088] Gray-scale standard deviation: This metric calculates the dispersion of gray-scale values ​​of all pixels within a rectangular region from their average gray-scale value. It is used to assess the uniformity of an interface. A smaller gray-scale standard deviation indicates a more uniform brightness distribution at the interface, stable material distribution, and the absence of significant light spots or uneven areas, thus reducing potential defects or performance fluctuations.

[0089] Thickness fluctuation: This metric calculates the dispersion of battery layer thickness values ​​within a rectangular region, using variance to determine thickness fluctuation. It reflects the flatness and consistency of the battery layer stack. Smaller thickness fluctuations indicate more uniform battery layer thickness, thus contributing to improved overall battery performance and lifespan.

[0090] Connectivity coefficient: The connectivity coefficient is calculated by dividing the area of ​​the largest connected region within a rectangular area by the total area of ​​the entire interface region. This index describes the continuity of the interface and reflects whether there are breaks or cracks in the battery interface. A higher connectivity coefficient indicates a more complete battery interface structure, which helps improve the battery's mechanical strength and long-term stability.

[0091] Pore ​​percentage: This is the percentage of pixels with gray values ​​less than a gray value threshold (usually set to 60%-70% of the average brightness of the rectangular area) within a rectangular area, out of the total number of pixels. This metric reflects the proportion of defects at the interface. A higher porosity percentage indicates that there may be more gaps or defects at the interface, which may lead to decreased battery performance, reduced efficiency, or structural instability.

[0092] By combining these five indicators, the quality, uniformity, stacking flatness, continuity, and defects of the battery interface can be comprehensively evaluated.

[0093] Through the above technical solutions, this application can reflect the density and compactness of battery materials by calculating the average grayscale. A higher average grayscale value indicates a denser material, reducing internal defects and porosity, thereby improving the safety and stability of the battery. The grayscale standard deviation can reveal the uniformity of brightness distribution on the battery interface; a smaller standard deviation indicates good uniformity of the battery interface structure, avoiding local performance differences that may be caused by uneven material distribution. The calculation of thickness fluctuation can reflect the flatness of the battery layer stack. Smaller thickness fluctuation indicates a more uniform thickness of the battery layers, avoiding the impact of irregularities in the battery structure on performance and lifespan. The calculation of connectivity coefficient helps to assess whether there are breaks or cracks at the battery interface; a higher connectivity coefficient indicates a more complete and coherent battery interface structure, which helps to improve the mechanical strength and fatigue resistance of the battery. The porosity ratio can reveal the proportion of defects on the interface; areas with grayscale values ​​below a set threshold correspond to pores or cracks; a lower porosity ratio indicates fewer interface defects and a more complete structure. The comprehensive analysis of these indicators can provide important reference data for battery design optimization and quality control.

[0094] This application further proposes a method for calculating the average consistency score of all valid analytical slices, including:

[0095] The calculated average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity are normalized to obtain normalized parameters.

[0096] The normalized parameters are weighted and fused to generate a consistency score. The average consistency score is generated by averaging the consistency scores of all valid analysis slices.

[0097] The formula for calculating the consistency score is as follows: ;in, These are weighting factors, with weights of respectively =0.25、 =0.2、 =0.25、 =0.15、 =0.15; The standard deviation of grayscale; The connectivity coefficient; For thickness fluctuations; Average gray level; The porosity is the percentage of voids; the formula for calculating the average consistency score is: .

[0098] Specifically, when calculating the average consistency score for all valid analytical slices, five key indicators (average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity) for each valid analytical slice need to be normalized. The purpose of normalization is to convert indicators of different dimensions and scales into uniform standardized values, allowing for comparison and weighted calculation within a unified range. The normalized result of each indicator serves as a normalization parameter, ensuring the comparability of indicators in the consistency score calculation. A weighted summation method is used to fuse these normalized parameters into the final consistency score. During weighted fusion, different indicators are assigned different weights based on their importance to the consistency of the battery structure. Average gray level and thickness fluctuation are assigned higher weights (…). =0.25 and =0.25, because they directly reflect the compactness and structural stability of the battery; the weights of the grayscale standard deviation and connectivity coefficient are slightly lower ( =0.2 and =0.15), while the weight of the porosity ratio is the smallest ( =0.15), because although porosity is important, it is relatively minor in the overall consistency evaluation. The average consistency score for each sample is obtained by averaging the consistency scores of all valid analytical slices. The average consistency score can comprehensively reflect the structural consistency of the battery slices in multiple dimensions, helping to determine whether there are potential structural problems at the battery interface.

[0099] Through the above technical solution, this application normalizes the average gray level, gray level standard deviation, thickness fluctuation, connectivity coefficient, and porosity of all valid analytical slices, and then applies them according to preset weights ( =0.25、 =0.2、 =0.25、 =0.15、 =0.15) is used to perform weighted fusion to generate a consistency score. Then, the consistency scores of all valid slices are averaged to obtain the average consistency score, which quantifies the structural consistency of the battery interface. It can uniformly evaluate the multi-dimensional structural features, reflect the compactness, uniformity, flatness, continuity and defect ratio of the battery layer, and realize an objective evaluation of the overall quality of the battery interface.

[0100] This application further proposes, when classifying rating levels based on average consistency scores, the following:

[0101] The rating levels are A, B, C, D, and E; the average consistency score is a natural number from 0 to 1.

[0102] When 1 ≥ average consistency score > 0.9, it is classified as Grade A; indicating that the interface is flat, dense and uniform.

[0103] When 0.9 ≥ average consistency score > 0.8, it is classified as Grade B; indicating slight fluctuations in local areas.

[0104] When 0.8 ≥ average consistency score > 0.7, it is classified as Grade C; indicating slight thickness fluctuation.

[0105] When 0.7 ≥ average consistency score > 0.6, it is classified as Grade D; indicating significant unevenness.

[0106] When 0.6 ≥ average consistency score ≥ 0, it is classified as Grade E, indicating the presence of delamination or porosity defects.

[0107] Specifically, when scoring the interface of all-solid-state battery slices based on the average consistency score, the score is divided into five levels, which can intuitively reflect the structural quality and uniformity of the interface. Specifically, when the average consistency score is between 0.9 and 1, it is classified as Grade A. At this level, the interface is flat, dense, and uniform, the battery layers are stacked smoothly, defects are minimal, material distribution is balanced, and the overall structural quality is optimal. When the average consistency score is between 0.8 and 0.9, it is classified as Grade B. The interface is generally good, but there are slight local undulations or minor thickness fluctuations, which have a limited impact on battery performance. When the average consistency score is between 0.7 and 0.8, it is classified as Grade C. This indicates slight thickness fluctuations, slightly poor interface uniformity, and the possibility of a few local defects, but the overall performance is still acceptable. When the average consistency score is between 0.6 and 0.7, it is classified as Grade D. The interface is significantly uneven, the thickness varies considerably, and the local structure may be loose or have slight cracks, which may have some impact on battery performance. When the average consistency score is below 0.6 but greater than or equal to 0, it is classified as Grade E. At this level, there are obvious delamination or porosity defects at the interface, uneven thickness, and poor connectivity, which may lead to a significant decrease in battery performance or safety hazards.

[0108] Using the above technical solution, this application scores the interface of all-solid-state batteries based on the average consistency score, classifying the sliced ​​interface into five levels from A to E. Each level intuitively reflects the uniformity and integrity of the interface structure. Level A (score 0.9–1) indicates a smooth, dense, and uniform interface with stable material distribution, very few defects, and the best overall structural quality; Level B (score 0.8–0.9) indicates a generally good interface with only slight local fluctuations, having limited impact on performance; Level C (score 0.7–0.8) indicates slight thickness fluctuations, slightly poor interface uniformity, and possibly a few local defects; Level D (score 0.6–0.7) indicates a significantly uneven interface with large thickness variations, potentially loose local structures, and performance may be affected; Level E (score 0–0.6) indicates the presence of delamination or porosity defects, poor connectivity, which may lead to a significant decrease in battery performance or safety hazards. This classification method is beneficial for quickly and intuitively evaluating the quality of the battery interface.

[0109] This application further proposes that if there is a consistency score that deviates from the average consistency score by more than 0.1, the region corresponding to the valid analysis slice is marked as a structural anomaly region; and the structural anomaly region is spatially located to determine the specific location of the anomaly region in the battery.

[0110] Specifically, in the interface consistency evaluation process of all-solid-state batteries, if the consistency score of a valid analytical slice deviates from the calculated average consistency score by more than 0.1, it indicates that the structural characteristics of the region where the slice is located differ significantly from the overall interface, and this region may contain local defects or anomalies. To accurately identify and analyze such anomalous regions, the region corresponding to the slice is marked as a structural anomaly area, and further spatial positioning is performed. By using the slice's sequence number in the thickness direction and its position coordinates in a two-dimensional image, the specific three-dimensional spatial location of the anomalous region within the battery is determined.

[0111] Through the above technical solution, this application quantifies the severity of the abnormal region and intuitively displays its distribution inside the battery.

[0112] This application further proposes methods for generating structured evaluation reports, including:

[0113] The consistency score, rating level, interface structure parameters, and abnormal area identifiers are integrated into a structured evaluation report; the structured evaluation report includes text descriptions and visual charts.

[0114] Specifically, when generating the structured evaluation report, all analysis results are integrated to comprehensively reflect the structural consistency and potential anomalies of the all-solid-state battery interface. The report includes the consistency score for each valid analysis slice and the overall average consistency score, and provides intuitive quality evaluation descriptions based on the average consistency score's classification (Grade A to Grade E). The report lists key interface structural parameters for each slice, such as average grayscale, grayscale standard deviation, thickness fluctuation, connectivity coefficient, and porosity, to quantitatively describe the density, uniformity, flatness, coherence, and defect ratio of the battery layer. For identified structural anomaly areas, the report will mark their spatial location within the battery and provide a 3D or 2D location diagram, making the distribution of anomaly areas visually apparent. The report includes visualizations such as interface thickness curves, grayscale distribution curves, connectivity heatmaps, and porosity distribution maps to help users quickly understand interface structural characteristics and potential problems. The textual descriptions detail the meaning of each parameter, the evaluation results, and their potential impact on battery performance.

[0115] Through the above technical solutions, this application, by combining text and charts, provides a structured evaluation report that not only offers a scientific basis for battery manufacturing and quality control, but also supports defect analysis, process optimization, and product reliability assessment, achieving a comprehensive, intuitive, and quantitative evaluation of the interface structure of all-solid-state batteries.

[0116] In summary, a unified algorithm was used for slice selection, boundary detection, and parameter extraction, avoiding subjective human error. Interface smoothness, density, uniformity, and defect ratio were comprehensively converted into a consistency score, reflecting the dynamic evolution of the internal structure during battery cycling. A unified grade and trend curve were output, facilitating batch comparison and standard establishment. Using CT equipment to scan battery samples with a resolution of no more than 5 micrometers ensured sufficiently detailed slice images, providing high-quality data for structural consistency analysis. By calculating multiple indicators such as sharpness and brightness uniformity, a comprehensive analysis of the solid-solid interface structure of the battery was achieved, ensuring consistency and stability in different regions. Weighted summation generated quality scores, and analysis of parameters such as gradient, porosity, and thickness fluctuation of each slice formed a multi-dimensional quality assessment system. Precise calculation of the battery layer thickness and its fluctuations, along with determination of the left and right boundaries of the electrode region, improved the accuracy of the evaluation. For potentially abnormal areas in the battery structure, consistency score deviation analysis identified and located structural anomalies, revealing potential structural problems. By combining consistency scores, rating levels, and structural parameters, an easy-to-understand and structured evaluation report was generated. The use of clear rating levels (from A to E) provides a quantitative standard for battery quality control and production.

[0117] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0118] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0119] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for evaluating the consistency of a solid-solid interface structure of an all-solid-state battery, characterized by, The solid-solid interface structure consistency evaluation method of the all-solid-state battery comprises: The CT device is used to scan the all-solid-state battery sample to obtain a two-dimensional slice sequence along the thickness direction; and the two-dimensional slice sequence is sequentially numbered; The definition index and the brightness uniformity index of each two-dimensional slice are calculated according to the two-dimensional slice sequence, and a weighted sum is generated to generate a quality score and select effective analysis slices; The image row average gray curve of the effective analysis slice is collected, the average gray value and the gray gradient curve of each row are calculated, the gradient peak position is determined, and the battery layer thickness and the average thickness of the battery layer are calculated; The column direction average gray distribution curve of each effective analysis slice is calculated, and the threshold-gradient joint determination method is used to determine the boundary of the electrode region and the rectangular region surrounded by the boundary; The average gray value, the gray standard deviation, the thickness fluctuation, the connectivity coefficient and the pore ratio of each effective analysis slice are calculated, and the average consistency score of all the effective analysis slices is calculated; the average consistency score is divided into score levels, and a structured evaluation report is generated.

2. The method of claim 1, wherein the method is a method of evaluating the consistency of the solid-solid interface structure of the all-solid-state battery, characterized by, When calculating the definition index and the brightness uniformity index, the following steps are included: The Laplace variance is used to evaluate the image sharpness to obtain the definition index, and the definition score is generated by normalization; the image gray standard deviation is calculated to obtain the brightness uniformity index, and the uniformity score is generated by normalization; the normalized definition score and the uniformity score are statistically analyzed.

3. The method according to claim 2, wherein When selecting the effective analysis slices, the following steps are included: The slice sequence is evenly divided into several equal length sections according to the uniformity of the battery thickness, each equal length section contains an equal number of two-dimensional slices; in each equal length section, the top 4 slices with the highest quality score are selected, totaling 20 as effective analysis samples.

4. The method of claim 3, wherein the method is a method of evaluating consistency of a solid-solid interface structure of an all-solid battery, characterized by, When determining the gradient peak position, the following steps are included: The first gradient peak corresponds to the upper boundary line of the battery, and the last gradient peak corresponds to the lower boundary line of the battery; the moving average filter is used to smooth the upper boundary line and the lower boundary line to eliminate the random fluctuations of the boundary line and retain the structural characteristics.

5. The method of claim 4, wherein the method is a method of evaluating consistency of a solid-solid interface structure of an all-solid battery, characterized by, When the threshold-gradient joint determination method is used to determine the boundary of the electrode region and the rectangular region surrounded by the boundary, the following steps are included: Based on the column direction average gray distribution curve, the left and right edge regions where the brightness decreases to the preset brightness threshold are identified, the bright region is defined as the effective range of the image, the boundary and the rectangular region surrounded by the boundary are determined; When the gray gradient is less than the set gradient threshold and the average gray value is less than 80% of the average value of the entire image, the region is determined as an invalid edge region.

6. The method of claim 5, wherein the method is a method of evaluating the consistency of the solid-solid interface structure of the all-solid-state battery, characterized by, When calculating the average gray value, the gray standard deviation, the thickness fluctuation, the connectivity coefficient and the pore ratio of each effective analysis slice, the following steps are included: The arithmetic mean of the gray values of all pixel points in the rectangular region of each effective analysis slice is calculated to obtain the average gray value; The dispersion degree of the gray values of all pixel points in the rectangular region of each effective analysis slice from the average value is calculated to obtain the gray standard deviation; The dispersion degree of the battery layer thickness values in the rectangular region of each effective analysis slice is calculated to obtain the thickness fluctuation; A connectivity coefficient is obtained by calculating a ratio of a maximum connected region area in a rectangular region of each of the effective analysis slices to a total area of the entire interface region; A pore ratio is obtained by counting a proportion of a number of pixel points with a gray value less than a gray threshold value in a total number of pixel points in the rectangular region of each of the effective analysis slices.

7. The method according to claim 6, wherein The average consistency score of all the effective analysis slices is calculated, including: The average gray value, the gray standard deviation, the thickness fluctuation, the connectivity coefficient, and the pore ratio are normalized to obtain normalized parameters; The normalized parameters are fused to generate a consistency score, and an average value of the consistency scores of all the effective analysis slices is calculated to generate the average consistency score. 8.The method of claim 7, wherein the method is characterized by, The average consistency score is divided into score levels, including: The score levels include A, B, C, D, and E, and the average consistency score is a natural number between 0 and 1; When 1>the average consistency score>0.9, the score is A; When 0.9>the average consistency score>0.8, the score is B; When 0.8>the average consistency score>0.7, the score is C; When 0.7>the average consistency score>0.6, the score is D; When 0.6>the average consistency score>0, the score is E. 9.The method of claim 8, wherein the method is characterized by, If the consistency score deviates from the average consistency score by more than 0.1, the region corresponding to the effective analysis slice is marked as a structural abnormal region, and the structural abnormal region is spatially located to determine the specific position of the abnormal region in the battery. 10.The method of claim 9, wherein the method is characterized by, The structured evaluation report is generated, including: The consistency score, the score level, the interface structure parameter, and the abnormal region identifier are integrated into a structured evaluation report, and the structured evaluation report includes a textual description and a visual chart.

Images