A method for automatically calibrating and correcting errors of CT / DR imaging precision of an electric core

Abstract

The present application relates to the technical field of non-destructive X-ray imaging algorithm of lithium battery industry, and particularly discloses a cell CT / DR imaging precision automatic calibration and error correction method, comprising the following steps: S1, composite calibration body synchronous imaging preprocessing, S2, multi-error coupling decoupling decomposition, S3, adaptive optimal compensation amount solving, S4, multi-dimensional collaborative correction, S5, dynamic calibration threshold determination, S6, calibration self-verification closed-loop reinspection. The cell CT / DR imaging precision automatic calibration and error correction method solves the problems that the three types of imaging errors of geometry, gray scale and detector are mutually coupled, multiple calibration is required, and cross residual error cannot be separated, forms a complete automatic calibration algorithm system, does not require manual intervention, greatly improves the imaging gray scale uniformity and size measurement precision of internal defects of the cell, takes into account the production line detection efficiency, and is suitable for offline calibration and online real-time calibration of cylindrical, soft package and hard shell cells.

Description

Technical Field

[0001] This invention relates to the field of non-destructive X-ray imaging algorithm technology in the lithium battery industry, and in particular to an automatic calibration and error correction method for CT / DR imaging accuracy of battery cells. Background Technology

[0002] Lithium-ion battery production widely uses DR planar imaging and CT tomography to detect defects such as misaligned electrode plates, poor soldering of terminal posts, and internal foreign objects. Imaging accuracy directly determines the defect detection rate. Existing imaging calibration technologies suffer from four major shortcomings: 1. Multiple errors are strongly coupled and cannot be separated and quantized synchronously: Traditional single standard samples can only calibrate geometric or grayscale errors separately, requiring multiple standard sample replacements and multiple imaging operations; the three types of errors, namely detector pixel non-uniformity, grayscale nonlinearity, and geometric distortion, are coupled with each other, and the cross residuals continuously affect the imaging measurement accuracy, resulting in long calibration time and low production line uptime.

[0003] 2. The compensation amount lacks a standardized solution logic and relies on manual trial and error: The displacement of the X-ray source and detector, as well as the grayscale correction parameters, depend on repeated adjustments by engineers. There is no globally optimal mathematical model, which makes it very easy for overcompensated image distortion or undercompensated residual errors to occur.

[0004] 3. Fixed and rigid calibration thresholds and trigger cycles: The allowable error range was not dynamically adjusted based on ambient temperature, continuous operating time of the equipment, and cumulative cell detection volume; thermal drift amplifies the error under high temperature and long-term operation, and the fixed standard leads to loss of control over the accuracy of batch cell detection.

[0005] 4. Lack of calibration closed-loop self-verification mechanism: There is no automatic re-inspection process after parameter adjustment is completed, making it impossible to determine whether the calibration has met the standard. Equipment with excessive error will continue to produce unqualified images, resulting in batch missed inspections and potential quality risks.

[0006] Existing technologies lack a fully automated algorithm system integrating multi-error synchronous decoupling, adaptive optimal compensation, dynamic calibration threshold, and closed-loop self-verification, thus failing to meet the high-precision unmanned inspection requirements of high-end lithium battery production lines. To address these issues, this invention proposes an automatic calibration and error correction method for CT / DR imaging accuracy of battery cells. Summary of the Invention

[0007] The main objective of this invention is to provide an automatic calibration and error correction method for the accuracy of battery cell CT / DR imaging, which can effectively solve the problems in the background art.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An automatic calibration and error correction method for CT / DR imaging accuracy of battery cells, which is executed fully automatically by the industrial control computer of the equipment equipped with the imaging algorithm, includes the following closed-loop execution steps: S1. Synchronous imaging preprocessing of composite calibration body: Place a composite standard calibration body with integrated dot matrix geometric calibration area, multi-level gray scale area and uniform reference plate area, perform single panoramic X-ray imaging, use adaptive median filtering for noise reduction, and extract dot matrix coordinates through sub-pixel algorithm. S2. Multi-error coupling decoupling decomposition: A multi-error joint decoupling algorithm is adopted to independently separate three types of uncoupled errors from a single calibration image: pixel response unevenness error, gray-level nonlinearity error, and geometric distortion error, and then quantize them in the whole domain. S3. Solving for the optimal compensation amount: Construct a multi-objective minimization model with mechanical travel and grayscale coefficient boundary constraints, and use sequential quadratic programming (SQP) to iteratively solve for all the optimal compensation amounts of the ray source displacement, detector XY displacement, pixel matrix, and grayscale coefficient. S4. Multi-dimensional collaborative correction: Simultaneously perform mechanical position correction of the X-ray source and detector, and image pixel and grayscale channel correction; S5. Dynamic calibration threshold determination: Real-time acquisition of ambient temperature, continuous operating time of equipment, and cumulative number of battery cells detected, dynamically updating the imaging allowable error threshold; S6. Calibration self-verification closed-loop re-inspection: After calibration, the body image is recalibrated and three indicators are extracted: pixel mean square error, grayscale correlation coefficient, and geometric mean offset. If any indicator exceeds the dynamic threshold, it will automatically return to S2 for secondary decoupling compensation. The loop is limited to 3 times. If the standard is not met after 3 times, a hardware warning will be output. After the standard is met, the entire compensation parameter library is stored and the calibration can be called in real time for batch imaging of battery cells.

[0009] Preferably, step S2, multi-error coupling decoupling, includes three sets of independent quantization calculation formulas, and the three types of errors are completely separated without cross-coupling: (1) Pixel response unevenness error:

[0010] In the formula: coordinates Pixel response deviation; The original image grayscale; The average gray level of pixels in a uniform flat area; (4) Gray-scale nonlinear error:

[0011] In the formula: This refers to the nonlinear deviation of grayscale attenuation. The actual thickness of the device is mapped to grayscale. This is the standard homogeneous grayscale decay function; (5) Two-dimensional geometric distortion offset field:

[0012] In the formula: , Extract the actual dot matrix coordinates for sub-pixels; , These are the coordinates of the theoretical standard lattice. Independent error matrices are generated separately for compensation solution.

[0013] Preferably, the adaptive optimal compensation solution model in step S3 includes a multi-objective objective function and a set of boundary constraint equations: Comprehensive residual minimization objective function:

[0014] Constraints:

[0015] In the formula: For the comprehensive imaging residuals; The error weights are for the three categories, and their sum is 1. For pixel compensation matrix; This is the grayscale correction factor; For geometric offset compensation; To compensate for the displacement of the radiation source; , This is the XY compensation amount for the detector; , This is the upper limit of the mechanical travel. , Upper and lower limits of grayscale coefficients; the SQP algorithm is used to solve the global optimal compensation vector to avoid overcompensation and undercompensation.

[0016] Preferably, step S4, the collaborative correction, is executed in two layers: Mechanical correction layer: based on , , The two-dimensional motion platform driving the X-ray source and detector completes physical position compensation; Image correction layer: Load the detector acquisition channel and correct the grayscale pixel by pixel, then substitute it into... Globally corrected X-ray grayscale attenuation curve.

[0017] Preferably, the formula for calculating the dynamic threshold in step S5 is:

[0018] In the formula: This is a real-time dynamic error threshold; This is the factory-set threshold value for the equipment. This is a temperature correction factor; This is the cumulative cell detection coefficient; The correction factor is the continuous operation factor; the higher the temperature, the larger the total amount of detection, and the longer the continuous operation, the smaller the correction factor and the stricter the judgment criteria.

[0019] Preferably, the three judgment indicators for self-verification in step S6 are pixel mean square error, gray-level linear correlation coefficient, and average geometric distortion offset; if any one of the three indicators exceeds... Automatically initiates a secondary calibration cycle; if three consecutive calibrations fail, an early warning of a hardware failure in the X-ray source or detector is issued.

[0020] Preferably, the composite standard calibration body is an integrated three-layer partition: First layer of lattice geometric calibration area: multi-scale, multi-spatial frequency metal lattice array; The second layer is a multi-level grayscale gradient area: homogeneous aluminum and graphite gradients of different thicknesses; The third layer is a uniform reference plate area: a low-attenuation homogeneous plate; image data required for three types of errors are acquired simultaneously in a single imaging session.

[0021] Preferably, the formula for extracting sub-pixel coordinates in step S1 is:

[0022] In the formula: The x-coordinate of the subpixel of the dot matrix; Integer pixel coordinates; To improve the decoupling accuracy of geometric distortion by using grayscale values ​​at corresponding locations.

[0023] Preferably, after the calibration in step S6 is successful, the product is permanently stored. , , The parameters are then accessed from the local parameter library; all subsequent CT / DR batch imaging of battery cells can directly call the parameters for real-time correction without repeating the calibration process.

[0024] Compared with the prior art, the present invention has the following beneficial effects: 1. Single-shot imaging synchronous decoupling of three types of coupling errors: The integrated composite calibration block acquires all calibration data in one imaging, eliminating the need to replace the standard sample multiple times, completely eliminating cross residuals, and reducing calibration downtime; 2. Mathematical model adaptively solves for optimal compensation: It abandons manual trial and error, and uses constrained multi-objective optimization to globally minimize imaging residuals, thus avoiding overcompensation distortion and undercompensation residual errors; 3. Dynamic threshold balance between accuracy and production line efficiency: The error judgment standard is adaptively adjusted based on temperature, running time and detection volume. The calibration requirements are tightened under high temperature conditions and the downtime calibration is reduced under stable conditions, thus balancing imaging accuracy and production capacity. 4. Fully automated closed-loop self-verification to eliminate batch quality risks: Automatic re-inspection after calibration, automatic secondary calibration for non-compliance, hardware warning for exceeding limits, preventing low-precision images from entering the inspection process in batches, and significantly reducing the rate of missed detection of minor defects; 5. High versatility and adaptability to all scenarios: Compatible with cylindrical, soft-pack, and square hard-shell battery cell CT / DR equipment, supports offline laboratory calibration and online automatic cyclic calibration on the production line in dual modes, and the algorithm can be directly integrated into the software without modifying the hardware. Attached Figure Description

[0025] Figure 1 This is a flowchart of the overall method for automatic calibration and error correction of CT / DR imaging accuracy of battery cells according to the present invention; Figure 2 This is a schematic diagram of the three-layer partitioned composite standard calibration body structure of the automatic calibration and error correction method for CT / DR imaging accuracy of battery cells according to the present invention; Figure 3 This is a logic diagram for solving the multi-error coupling decoupling and compensation of an automatic calibration and error correction method for CT / DR imaging accuracy of a battery cell according to the present invention. Detailed Implementation

[0026] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0027] like Figures 1-3 As shown, an automatic calibration and error correction method for CT / DR imaging accuracy of battery cells is automatically executed by the imaging algorithm embedded in the equipment's industrial control computer. The overall closed loop includes six core steps, and is equipped with an integrated three-layer partitioned composite standard calibration body to realize the calibration of all data sources in a single imaging acquisition. Specifically: S1. Composite calibration volume imaging and image preprocessing: The composite calibration body, integrating the dot matrix geometric calibration area, multi-level grayscale step area, and uniform reference plate area, is placed at the standard station of the equipment to perform a single panoramic X-ray imaging, outputting the original grayscale image. Adaptive median filtering is used for noise reduction, and sub-pixel edge extraction algorithm is used to improve the accuracy of dot matrix coordinate extraction.

[0028] Subpixel coordinate calculation formula:

[0029] In the formula: : Precise horizontal coordinates of subpixel matrix; : Original integer pixel coordinates of the dot matrix; : The grayscale value of the original image at the corresponding coordinate position.

[0030] Composite calibration body partitioning function: Lattice geometric calibration area: Multi-scale, multi-spatial frequency metal lattice, used for geometric distortion coordinate acquisition; Multi-level grayscale gradient region: Homogeneous aluminum / graphite gradients of different thicknesses are used to construct standard X-ray grayscale attenuation curves; Uniform reference plate area: a low-attenuation homogeneous plate used to solve for the pixel response error of the global detector.

[0031] S2. Decoupling and decomposition of multiple error couplings: Three types of non-cross-coupled errors are independently separated and quantized from a single preprocessed calibration image, and independent error matrices are output for each.

[0032] Formula for detector pixel response non-uniformity error:

[0033] In the formula: Pixel coordinates Response deviation; : Original image pixel grayscale; The average grayscale value of all pixels in the uniform reference plate area represents the standard ideal response of the detector.

[0034] Gray-scale nonlinear mapping error formula:

[0035] In the formula: : Nonlinear deviation in grayscale attenuation; : Actual thickness of equipment - grayscale mapping function; : The grayscale decay function of a theoretical standard homogeneous material.

[0036] Formula for two-dimensional geometric distortion migration error field:

[0037] In the formula: : Pixel 2D distortion offset vector; , Sub-pixel algorithm extracts the actual coordinates of the dot matrix; , : Standard coordinates of the lattice theory.

[0038] The three types of errors are completely decoupled and generated separately. , , The error matrix is ​​used for compensation calculations.

[0039] S3. Solving for the adaptive optimal compensation amount: A multi-objective minimization model with mechanical travel and grayscale parameter boundary constraints is constructed. Sequential quadratic programming (SQP) is used to iteratively solve for the global optimal compensation vector, avoiding overcompensation and undercompensation.

[0040] Comprehensive residual objective function:

[0041] Constraint equations:

[0042] In the formula: The smaller the value of the comprehensive residual after compensation, the higher the calibration accuracy. Pixel, grayscale, and geometric error weights ; : Pixel-wise gain compensation matrix; Global grayscale correction factor; Pixel-level geometric offset compensation; : One-dimensional displacement compensation of the X-ray source; : Detector displacement compensation in the X and Y directions; The maximum adjustable stroke of the X-ray source mechanism; : Maximum travel distance of the detector in one direction; The upper and lower limits of the grayscale coefficient are used to prevent image grayscale saturation distortion.

[0043] Solving for the optimal compensation vector using SQP: .

[0044] S4, Multi-dimensional Collaborative Correction: Perform simultaneous mechanical hardware calibration and image channel calibration: 1. Mechanical geometric correction: based on The motion platform of the driving radiation source and detector completes physical position compensation; 2. Image pixel compensation: Load the detector acquisition channel and correct the original grayscale pixel by pixel; 3. Global grayscale correction: Substitute Correct the X-ray grayscale attenuation curve of the entire image.

[0045] S5. Dynamic threshold calibration determination: Real-time collection of ambient temperature Continuous operating time of equipment Total number of battery cells tested The upper limit of the allowable error is dynamically updated.

[0046] Formula for calculating dynamic error threshold:

[0047] In the formula: Real-time dynamic error allowable threshold; : Fixed threshold values ​​at the equipment's factory specifications; Temperature correction factor; the higher the temperature, the smaller the value. : Cumulative detection correction coefficient; the larger the detection volume, the smaller the value. : Continuous operation correction factor, the longer the working time, the smaller the value.

[0048] High temperature and long-term production conditions Automatic narrowing and stricter calibration criteria; relaxed threshold for normal temperature and low load conditions, reducing the frequency of downtime calibration.

[0049] S6. Calibration self-verification closed-loop re-inspection: After calibration, the composite calibration body is placed again for imaging, and three verification indicators, namely pixel mean square error, gray-level linear correlation coefficient, and geometric mean offset, are calculated simultaneously. Judgment Logic: Any Indicator The system automatically jumps to S2 to re-execute the decoupling and compensation calculation, with a maximum of 3 iterations. If the calculation fails after 3 iterations, a hardware fault warning for the X-ray source / detector is issued. Once all indicators meet the requirements, the calculation is stored. , , The parameters are accessed from the parameter library, and subsequent batch imaging of battery cells can directly call the parameters for real-time correction without repeated calibration.

[0050] In summary, this invention provides an automatic calibration and error correction method for CT / DR imaging accuracy of battery cells. Addressing the shortcomings of existing technologies—namely, the inability to simultaneously decouple three types of coupling errors (geometric distortion, grayscale nonlinearity, and detector response), manual adjustment of compensation parameters, fixed calibration cycles, and lack of closed-loop verification—this invention employs a three-layer partitioned composite calibration body to acquire all calibration information in a single imaging session; designs a multi-error decoupling algorithm to achieve independent quantification of the three types of errors; constructs a constrained multi-objective optimization model to solve for the globally optimal compensation amount; introduces a multi-condition coupled dynamic threshold adjustment calibration standard; and adds a calibration re-inspection closed-loop self-verification process. This invention forms a complete automated calibration algorithm system, requiring no manual intervention, significantly improving the grayscale uniformity and dimensional measurement accuracy of internal defect imaging in battery cells, while also considering production line inspection efficiency. It is suitable for offline calibration and real-time online production line calibration of cylindrical, pouch, and rigid battery cells.

[0051] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatic calibration and error correction of cell-based CT / DR imaging accuracy, characterized in that, The imaging algorithm, implemented by the equipment's industrial control computer, is executed fully automatically, including the following closed-loop execution steps: S1. Synchronous imaging preprocessing of composite calibration body: Place a composite standard calibration body with integrated dot matrix geometric calibration area, multi-level gray scale area and uniform reference plate area, perform single panoramic X-ray imaging, use adaptive median filtering for noise reduction, and extract dot matrix coordinates through sub-pixel algorithm. S2. Multi-error coupling decoupling decomposition: A multi-error joint decoupling algorithm is adopted to independently separate three types of uncoupled errors from a single calibration image: pixel response unevenness error, gray-level nonlinearity error, and geometric distortion error, and then quantize them in the whole domain. S3. Solving for the optimal compensation amount: Construct a multi-objective minimization model with mechanical travel and grayscale coefficient boundary constraints, and use sequential quadratic programming (SQP) to iteratively solve for all the optimal compensation amounts of the ray source displacement, detector XY displacement, pixel matrix, and grayscale coefficient. S4. Multi-dimensional collaborative correction: Simultaneously perform mechanical position correction of the X-ray source and detector, and image pixel and grayscale channel correction; S5. Dynamic calibration threshold determination: Real-time acquisition of ambient temperature, continuous operating time of equipment, and cumulative number of battery cells detected, dynamically updating the imaging allowable error threshold; S6. Calibration self-verification closed-loop re-inspection: After calibration, re-image the calibration body and extract three indicators: pixel mean square error, gray-level correlation coefficient, and geometric mean offset. If any indicator exceeds the dynamic threshold, it will automatically return to the S2 secondary decoupling compensation, with a maximum cycle of 3 times. If the standard is not met after 3 cycles, a hardware warning will be output. After the standard is met, the entire compensation parameter library will be stored, and the battery cell batch imaging will call up the correction in real time.

2. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 1, characterized in that: Step S2, multi-error coupling decoupling, involves three independent quantization calculation formulas, ensuring complete separation and no cross-coupling of the three types of errors: (1) Pixel response unevenness error: In the formula: coordinates Pixel response deviation; The original image grayscale; The average gray level of pixels in a uniform flat area; (2) Gray-scale nonlinear error: In the formula: This refers to the nonlinear deviation of grayscale attenuation. The actual thickness of the device is mapped to grayscale. This is the standard homogeneous grayscale decay function; (3) Two-dimensional geometric distortion offset field: In the formula: , Extract the actual dot matrix coordinates for sub-pixels; , These are the coordinates of the theoretical standard lattice. Independent error matrices are generated separately for compensation solution.

3. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 2, characterized in that: Step S3, the adaptive optimal compensation solution model, includes a multi-objective objective function and a set of boundary constraint equations: The objective function for minimizing the overall residual is: Constraints: In the formula: For the comprehensive imaging residuals; The error weights are for the three categories, and their sum is 1. This is the pixel compensation matrix; This is the grayscale correction factor; For geometric offset compensation; To compensate for the displacement of the radiation source; , This is the XY compensation amount for the detector; , This is the upper limit of the mechanical travel. , Upper and lower limits of grayscale coefficients; the SQP algorithm is used to solve for the global optimal compensation vector to avoid overcompensation and undercompensation.

4. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 3, characterized in that: Step S4, the collaborative correction, is executed in two layers: Mechanical correction layer: based on , , The two-dimensional motion platform driving the X-ray source and detector completes physical position compensation; Image correction layer: Load the detector acquisition channel and correct the grayscale pixel by pixel, then substitute it into... Globally corrected X-ray grayscale attenuation curve.

5. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 4, characterized in that: The formula for calculating the dynamic threshold in step S5 is as follows: In the formula: This is a real-time dynamic error threshold; This is the factory-set threshold value for the equipment. This is a temperature correction factor; This is the cumulative cell detection coefficient; The correction factor is the continuous operation factor; the higher the temperature, the larger the total amount of detection, and the longer the continuous operation, the smaller the correction factor and the stricter the judgment criteria.

6. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 5, characterized in that: Step S6 self-verification uses three criteria: pixel mean square error, grayscale linear correlation coefficient, and average geometric distortion offset. If any one of these criteria exceeds... Automatically initiates a secondary calibration cycle; if three consecutive calibrations fail, an early warning of a hardware failure in the X-ray source or detector is issued.

7. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 1, characterized in that: The composite standard calibration body is integrated into a three-layer partition: First layer of lattice geometric calibration area: multi-scale, multi-spatial frequency metal lattice array; The second layer is a multi-level grayscale gradient area: homogeneous aluminum and graphite gradients of different thicknesses; The third layer of uniform reference plate area: low-attenuation homogeneous plate; A single imaging session simultaneously acquires image data required for three types of errors.

8. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 1, characterized in that: Step S1 Sub-pixel coordinate extraction formula: In the formula: The x-coordinate of the subpixel of the dot matrix; Integer pixel coordinates; To improve the decoupling accuracy of geometric distortion by using grayscale values ​​at corresponding locations.

9. The method for automatic calibration and error correction of cell CT / DR imaging accuracy according to claim 6, characterized in that: After calibration in step S6 is successful, store permanently. , , The parameters are then accessed from the local parameter library; all subsequent CT / DR batch imaging of battery cells can directly call the parameters for real-time correction without repeating the calibration process.

Images