Battery pack body detection imaging method and device based on circular trajectory scanning

Abstract

The invention discloses a battery pack body detection imaging method and device based on circular trajectory scanning. The method comprises the following steps: firstly, by inputting the three-dimensional size and key detection area information of a battery pack, generating a segmented circular trajectory in multi-circle-center distribution, enabling a radiation source and a detector to cooperatively move according to the planned circular trajectory, and completing local circular trajectory projection data acquisition segment by segment; and carrying out local reconstruction on the acquired projection data by using an improved filtering back projection algorithm, and finally splicing cross-sectional images of all local areas to form a complete battery pack body detection image. According to the method, the problem that the large battery pack body exceeds the effective imaging range of the detector is solved through a multi-circle-center segmented circular trajectory scanning mode, and the image reconstruction efficiency and quality are remarkably improved in combination with the optimized filtering back projection algorithm.

Description

Technical Field

[0001] The present invention relates to the field of detection technologies, and particularly to a method and device for detecting and imaging a battery pack body based on circular trajectory scanning. Background Art

[0002] X-ray Computed Tomography (CT) is widely used in the field of industrial inspection, and has the advantages of high spatial resolution, short scanning time, large scanning range, easy operation, etc. With the rapid development of the new energy vehicle industry, as the core safety component of the whole vehicle, the reliability detection requirements of power battery packs are increasing day by day, showing three core trends of high precision, high efficiency, and full life cycle management. However, the internal structure of the battery pack is complex and the materials are diverse (metals, polymers, liquid electrolytes, etc.), which brings great challenges to traditional detection technologies. For example, welding defects (false welding, cracks) may lead to an increase in the contact resistance of the module, and then cause thermal runaway, accounting for 37% of the causes of battery accidents; leakage or blockage of the cooling pipeline will cause a local temperature gradient ≥ 15°C, accelerating the aging of the battery cells and reducing the cycle life by 30%.

[0003] Although the existing CL scanning technology has sub-micron spatial resolution, second-level single-layer scanning speed, and full-frame non-destructive imaging characteristics, there are still limitations in the detection of large battery pack bodies, such as exceeding the effective imaging range of the detector, low calculation efficiency, noise sensitivity, edge artifacts, sensitivity to motion artifacts, etc., and it is difficult to meet the actual industrial application requirements. Summary of the Invention

[0004] Based on this, the embodiments of the present application provide a method and device for detecting and imaging a battery pack body based on circular trajectory scanning. This method uses the Filtered Back Projection (FBP) algorithm and combines the improvement of calculation efficiency, the improvement of the image quality of the battery pack, and practical applications.

[0005] In a first aspect, a method for detecting and imaging a battery pack body based on circular trajectory scanning is provided, and the method includes:

[0006] Input the three-dimensional size and key detection area information of the battery pack to generate a segmented circular trajectory with a multi-center distribution;

[0007] The radiation source and the detector move cooperatively according to the planned circular trajectory to complete the acquisition of local circular trajectory projection data section by section;

[0008] Use the improved filtered back projection algorithm to perform local reconstruction on the acquired projection data, and splice the tomographic images of each local area to form a complete battery pack body detection image.

[0009] Optionally, a segmented circular trajectory with a multi-center distribution is generated, including calculating the center coordinates, rotation angle range, and translation compensation amount of each segment of the trajectory.

[0010] Optionally, in the step of the collaborative movement of the ray source and the detector, the position information is fed back in real time through an encoder to ensure the time-space synchronization accuracy of the data.

[0011] Optionally, the acquisition of the projection data of the local circular trajectory is completed segment by segment, including:

[0012] At each scanning angle, the ray intensity distribution after penetrating the sample is acquired, and normalization is performed based on the reference intensity of the empty field to obtain the linear attenuation integral projection;

[0013] According to the distance between the ray source and the object center and the detector pixel coordinates, a distance-related gain adjustment is applied to the projection data;

[0014] The projection data is filtered, including one-dimensional Fourier transform and frequency-domain filtering operations, to correct the spectral distribution of the projection signal.

[0015] Optionally, at each scanning angle, the ray intensity distribution after penetrating the sample is acquired, and normalization is performed based on the reference intensity of the empty field to obtain the linear attenuation integral projection, specifically including:

[0016] The projection data I at each angle is acquired, and the attenuation projection is calculated. At each scanning angle θ, the ray intensity distribution I θ (x, y) after penetrating the sample is acquired, and normalization is performed based on the reference intensity I0(x, y) of the empty field to obtain the linear attenuation integral projection:

[0017]

[0018] Optionally, according to the distance between the ray source and the object center and the detector pixel coordinates, a distance-related gain adjustment is applied to the projection data, specifically including:

[0019] According to the distance D2 between the ray source and the object center and the detector pixel coordinates (D i , D j ), a distance-related gain adjustment is applied to the projection data:

[0020]

[0021] Optionally, after obtaining the distance-related gain adjustment of the projection data, the method further includes:

[0022] Filter the projection data matrix at each scanning angle; specifically, the filtering process includes one-dimensional Fourier transform, which includes performing discrete Fourier transform on each row of the projection data to convert it to the frequency domain; and frequency domain filtering operation, which includes multiplying the transformed data by a filter function in the frequency domain to correct the spectral distribution of the projection signal, and the filter function uses a Hamming window.

[0023] Optionally, the step of performing local reconstruction using the improved filtered backprojection algorithm includes:

[0024] Calculate the projection coordinates of the reconstructed pixel points at each angle.

[0025] Calculate the pixel values through bilinear interpolation, and gradually construct the image information of the local area.

[0026] Optionally, calculating the projection coordinates of the reconstructed pixel points at each angle and calculating the pixel values through bilinear interpolation specifically includes:

[0027] Calculate the projection coordinates (m, n) in the detector of the reconstructed pixel point f(x, z, y) at the angle θ between the detector and the ray source connection line:

[0028]

[0029] where D2 represents the ray source-object center distance;

[0030] Finally, calculate the value P(u, v, θ) of the point (u, v) through bilinear interpolation to calculate the pixel values of each reconstructed pixel point:

[0031]

[0032] Traverse and update all angles θ i After that, obtain the final reconstruction result.

[0033] In a second aspect, a battery pack body detection imaging device based on circular trajectory scanning is provided, and the device includes:

[0034] A trajectory generation module, configured to input the three-dimensional size of the battery pack and the information of the key detection area, and generate a segmented circular trajectory with a multi-center distribution;

[0035] A data acquisition module, configured to make the ray source and the detector move cooperatively according to the planned circular trajectory, and complete the acquisition of local circular trajectory projection data segment by segment;

[0036] A reconstruction and stitching module, configured to perform local reconstruction on the acquired projection data using the improved filtered backprojection algorithm, and stitch the tomographic images of each local area to form a complete battery pack body detection image.

[0037] The circular trajectory scanning of the present invention only requires the ray source and the detector to rotate synchronously around the center of the object, without the need for a multi-axis translation mechanism, reducing the equipment complexity and manufacturing cost. Reliable motion control: The rotation motion error can be corrected through the encoder closed-loop control, avoiding the cumulative error during the splicing of multiple straight lines, and improving the system repeatability accuracy (typical error < 0.01°). The 360° full-coverage scanning of the circular trajectory ensures the completeness of the Radon space projection data, avoiding truncation artifacts and meeting the mathematical conditions for accurate reconstruction. High signal-to-noise ratio data acquisition: Continuous rotation scanning can extend the single-angle integration time, improving the signal-to-noise ratio of the projection data (SNR increased by more than 30%), especially suitable for low-contrast defect detection. Efficiency of the FBP algorithm: Filtered back projection (FBP) is much lower than iterative reconstruction and is suitable for real-time imaging (< 1 second / layer). The back projection step of FBP can be decomposed into independent pixel calculations, and it is easy to achieve a hundred-fold efficiency improvement through GPU acceleration. Description of the Drawings

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, without creative efforts, other implementation drawings can also be obtained based on the provided drawings.

[0039] Figure 1 It is a step flow chart of a battery pack body detection imaging method based on circular trajectory scanning provided by an embodiment of the present application;

[0040] Figure 2 It is a mechanical system diagram of the implementation scenario provided by an embodiment of the present application;

[0041] Figure 3 It is a schematic diagram of the CL scanning method provided by an embodiment of the present application;

[0042] Figure 4 It is a reconstruction result diagram of local scanning provided by an embodiment of the present application. Detailed Embodiments

[0043] In order to make the purpose, technical solutions and advantages of the present application more clear, the following further details the present application in combination with the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.

[0044] In the description of the present invention, the terms "include", "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may also include other steps or units that are inherent to these processes, methods, products or devices although not clearly listed, or steps or units added based on further optimization solutions conceived in the present invention.

[0045] In the prior art, based on the Algebraic Reconstruction Technique (ART), the technical solution is as follows:

[0046] 1. Data acquisition: The X-ray source scans the object to be detected along a circular trajectory, and projects data at multiple angles are collected.

[0047] 2. Mathematical modeling: The imaging problem is transformed into a linear equation system Ax = b.

[0048] 3. Iterative solution: Initialize an assumed image (such as a matrix of all zeros); update the pixel values for each projection angle, approximate the true solution by minimizing the residual ||Ax = b||, and use a relaxation factor (such as 0.1 - 0.5) to control the convergence rate in each iteration.

[0049] The disadvantages of the prior art one include that although the ART algorithm can process incomplete projection data (such as limited angle or truncated data), it has the following key defects:

[0050] 1. Low computational efficiency: The number of iterations needs to be ≥100 times to converge, and the time-consuming for a single reconstruction can reach several hours (taking 512×512 pixels as an example), which cannot meet the requirements of real-time detection.

[0051] 2. Noise sensitivity: In the case of low-dose projection data (such as SNR < 20dB), noise is easily amplified during the iterative process, and additional regularization (such as Tikhonov regularization) is required, but this will sacrifice the image resolution.

[0052] 3. Edge artifacts: The initial value assumption (such as all zeros) results in "trailing" artifacts (error ≥ 15%) at the edges of the reconstructed image, and prior constraints need to be selected manually.

[0053] 4. Limitations in industrial applications: Sensitive to motion artifacts (such as battery pack vibration), strict mechanical fixation (positioning accuracy ≤ 10μm) is required, which increases the equipment cost and operation complexity.

[0054] However, in the existing technology, large battery packs are large and heavy in size, and mechanical structures such as gantry are required for hoisting and loading. Moreover, since large battery packs exceed the effective imaging range of the detector, the present invention proposes a multi-region circular trajectory scanning method, which utilizes the Filtered Back Projection (FBP) algorithm and combines the improvement of computational efficiency, the improvement of battery pack image quality and practical applications.

[0055] Please refer to Figure 1 , which shows a flowchart of a battery pack body detection imaging method based on circular trajectory scanning provided by an embodiment of the present application, and may include the following steps:

[0056] S1. Input the three-dimensional size of the battery pack and the information of key detection regions, and generate a segmented circular trajectory with a multi-center distribution.

[0057] In this step, first, obtain the three-dimensional size of the battery pack and the information of the key regions that need to be focused on for detection. Based on these parameters, through a specific algorithm or computational model, generate a segmented circular trajectory with a multi-center distribution. Specifically, calculate the center coordinates of each segment of the trajectory, determine the rotation angle range of each segment of the trajectory, and the translation compensation amount required to ensure the accuracy and continuity of the trajectory. These calculation results will provide an accurate path planning for the subsequent movement of the ray source and the detector, so as to achieve a comprehensive and effective scan of the battery pack.

[0058] S2. The ray source and the detector move collaboratively according to the planned circular trajectory, and complete the acquisition of local circular trajectory projection data segment by segment.

[0059] In this step, the ray source and the detector move collaboratively according to the circular trajectory generated in step S1. During the movement, the ray source emits X-rays, which penetrate the battery pack and are received by the detector. The position information is fed back in real time through an encoder to ensure that the ray source and the detector maintain an accurate time-space synchronization accuracy during the movement, and avoid inaccurate data acquisition caused by movement errors. At each preset scanning angle, the detector acquires the ray intensity distribution after penetrating the sample, and performs normalization processing based on the reference intensity of the empty field to obtain a linear attenuation integral projection. In addition, according to the distance between the ray source and the object center and the detector pixel coordinates, a distance-related gain adjustment is applied to the projection data to compensate for the inconsistent signal attenuation caused by the distance difference, so as to improve the quality and accuracy of the projection data and provide a reliable data basis for subsequent image reconstruction.

[0060] S3. Use the improved filtered back projection algorithm to perform local reconstruction on the acquired projection data, and splice the tomographic images of each local region to form a complete battery pack body detection image.

[0061] In this step, after the projection data acquisition is completed, the improved filtered back-projection algorithm is used to perform local reconstruction on the projection data collected for each trajectory. This algorithm first performs a one-dimensional Fourier transform on the projection data to convert it to the frequency domain, and then multiplies it by a filter function using a Hamming window in the frequency domain to correct the spectral distribution of the projection signal, enhance the edge information of the image, and suppress noise. Next, the projection coordinates of the reconstructed pixel points at each angle are calculated, and the pixel values are calculated using bilinear interpolation to gradually construct the image information of the local area. After traversing and updating all angles, the final reconstruction result of the local area is obtained. Finally, using image stitching technology, the tomographic images of each local area are stitched together to form a complete battery pack body detection image, thereby achieving a comprehensive and accurate detection of the internal structure of the entire battery pack body, and providing an intuitive and reliable basis for the quality assessment and defect detection of the battery pack.

[0062] In summary, it can be seen that in this application, the three-dimensional size of the battery pack and the key detection areas are first input, a segmented circular trajectory with a multi-center distribution is generated, and the center coordinates, rotation angle range, and translation compensation amount of each trajectory are calculated. For data acquisition and synchronization, the ray source and the detector move collaboratively along the planned path to complete the local circular trajectory projection data acquisition segment by segment; the position information is fed back in real time through the encoder to ensure the time-space synchronization accuracy of the data. For image reconstruction, the improved filtered back-projection algorithm (FBP) is used to perform local reconstruction on each segment of data. Finally, using image stitching technology, the tomographic images are stitched together, as Figure 2 shown in the mechanical system diagram of the implementation scenario.

[0063] In a specific embodiment of this application, as Figure 3 shown in the schematic diagram of the circular trajectory CL scan, it is specifically executed according to the following steps:

[0064] Set the scanning parameters: the number of projection acquisition angles M, the moving radius S of the ray source R , the moving radius D of the detector R , the included angle between the connection line of the detector and the ray source is θ, and the ray source and the detector both move counterclockwise, and the connection line passes through the center of a single area to complete the circular motion.

[0065] The rotation angle is:

[0066]

[0067] The position relationship between the x-axis and z-axis of the ray source relative to the origin is:

[0068] x S = S R * sin(θ)

[0069] z S = -S R * cos(θ)

[0070] The positional relationship between the x-axis and z-axis of the detector relative to the origin is:

[0071] x S = -D R *sin(θ)

[0072] z S = D R *cos(θ)

[0073] Configure the operating parameters of the X-ray tube according to the detection requirements, including tube current (typical range 0.1 - 5 mA) and tube voltage (typical range 200 - 450 kV); keep the sample stage in an unloaded state, and trigger the detector to record the intensity distribution I0(x,y) of the unattenuated X-ray background radiation.

[0074] Collect the projection data I at each angle. For attenuation projection calculation, at each scanning angle θ, collect the intensity distribution I θ (x,y) of the rays after penetrating the sample, and perform normalization based on the empty-field reference intensity I0(x,y) to obtain the linear attenuation integral projection:

[0075]

[0076] According to the source-object center distance D2 and the detector pixel coordinates (D i , D j ), apply a distance-related gain adjustment to the projection data:

[0077]

[0078] For the projection data matrix P θ (i,j) at each scanning angle θ, its filtering process is as follows:

[0079] (1) One-dimensional Fourier transform

[0080] Perform a discrete Fourier transform (DFT) on each row of the projection data (along the detector pixel direction) to convert it to the frequency domain:

[0081]

[0082] (2) Frequency-domain filtering operation

[0083] In the frequency domain, multiply the transformed data by the filter function H(w) to correct the spectral distribution of the projection signal:

[0084]

[0085] The filter function H(w) uses a Hamming window:

[0086]

[0087] Calculate the projection coordinates (m, n) of the reconstructed pixel point f(x, z, y) in the detector at angle θ:

[0088]

[0089] Finally, use bilinear interpolation to calculate the value P(u, v, θ) at point (u, v) and calculate the pixel values of each reconstructed pixel point:

[0090]

[0091] After traversing and updating all angles θ i After that, the final reconstruction result is obtained. Figure 4 The reconstruction result diagram of the local scan is shown.

[0092] In an alternative embodiment of the present application, the scanning trajectory can be attempted with equally spaced translational scanning, equal angle rotational scanning, and hybrid trajectory scanning. The reconstruction algorithm can be attempted to use the algebraic reconstruction technique (ART), the simultaneous iterative reconstruction technique (SIRT), and the conjugate gradient least squares method (CGLS).

[0093] In summary, it can be seen that in the present invention, circular trajectory scanning only requires the ray source and the detector to rotate synchronously around the center of the object, without a multi-axis translation mechanism, reducing the equipment complexity and manufacturing cost. The motion control is reliable: the rotational motion error can be corrected by encoder closed-loop control, avoiding the cumulative error during multi-segment straight line splicing, and improving the system repeatability accuracy (typical error < 0.01°). The 360° full coverage scanning of the circular trajectory ensures the completeness of the Radon space projection data, avoiding truncation artifacts and meeting the mathematical conditions for accurate reconstruction. High signal-to-noise ratio data acquisition: continuous rotational scanning can extend the single-angle integration time, improving the signal-to-noise ratio of the projection data (SNR increased by more than 30%), especially suitable for low-contrast defect detection. The efficiency of the FBP algorithm: Filtered back projection (FBP) is much lower than iterative reconstruction and is suitable for real-time imaging (< 1 second / layer). The back projection step of FBP can be decomposed into independent pixel calculations, and it is easy to achieve a hundred-fold efficiency improvement through GPU acceleration.

[0094] An imaging device for battery pack body detection based on circular trajectory scanning provided by an embodiment of the present application may include:

[0095] A trajectory generation module, configured to input the three-dimensional size and key detection area information of the battery pack and generate a segmented circular trajectory with a multi-center distribution;

[0096] A data acquisition module, configured to make the ray source and the detector move collaboratively according to the planned circular trajectory and complete the acquisition of local circular trajectory projection data segment by segment;

[0097] A reconstruction and stitching module, which is used to perform local reconstruction on the acquired projection data by using an improved filtered back-projection algorithm, and stitch the tomographic images of each local area to form a complete battery pack body detection image.

[0098] For the specific limitations of the battery pack body detection imaging device based on circular trajectory scanning, reference can be made to the limitations of the battery pack body detection imaging method based on circular trajectory scanning in the above text, which will not be elaborated here. Each module in the above-mentioned battery pack body detection imaging device based on circular trajectory scanning can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to the above-mentioned modules.

[0099] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0100] The above-described embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A battery pack body detection imaging method based on circular trajectory scanning, characterized in that The method includes: Inputting the three-dimensional dimensions of the battery pack and information on key detection regions to generate a segmented circular trajectory with a multi-center distribution; The radiation source and the detector move collaboratively along the planned circular trajectory to complete the acquisition of projection data for each local circular trajectory segment by segment; Using an improved filtered back-projection algorithm to perform local reconstruction on the acquired projection data, and stitching the tomographic images of each local region to form a complete detection image of the battery pack body.

2. The method according to claim 1, wherein Generating a segmented circular trajectory with a multi-center distribution includes calculating the center coordinates, rotation angle range, and translation compensation amount of each trajectory segment.

3. The method according to claim 1, wherein In the step of the collaborative movement of the radiation source and the detector, the position information is fed back in real time through an encoder to ensure the time-space synchronization accuracy of the data.

4. The method according to claim 1, characterized in that Completing the acquisition of projection data for each local circular trajectory segment by segment includes: At each scanning angle, acquiring the distribution of the ray intensity after penetrating the sample, and performing normalization based on the reference intensity in the empty field to obtain a linear attenuation integral projection; Applying a distance-related gain adjustment to the projection data according to the distance between the radiation source and the object center and the detector pixel coordinates; Performing filtering processing on the projection data, including one-dimensional Fourier transform and frequency-domain filtering operations to correct the spectral distribution of the projection signal.

5. The method according to claim 4, characterized in that, At each scanning angle, acquiring the distribution of the ray intensity after penetrating the sample, and performing normalization based on the reference intensity in the empty field to obtain a linear attenuation integral projection, specifically including: Collect the projection data I at each angle. For attenuation projection calculation, at each scanning angle θ, collect the ray intensity distribution I after penetrating the sample θ (x, y), and perform normalization based on the reference intensity I0(x, y) in the empty field to obtain the linear attenuation integral projection:

6. The method according to claim 5, characterized in that Applying a distance-related gain adjustment to the projection data according to the distance between the radiation source and the object center and the detector pixel coordinates, specifically including: According to the ray source-object center distance D2 and the detector pixel coordinates (D i , D j ), apply distance-dependent gain adjustment to the projection data:

7. The method according to claim 6, wherein After obtaining the distance-related gain adjustment applied to the projection data, the method further includes: Performing filtering processing on the projection data matrix at each scanning angle; wherein, the filtering processing specifically includes one-dimensional Fourier transform, including performing discrete Fourier transform on each row of the projection data to convert it to the frequency domain; and frequency-domain filtering operation, including in the frequency domain, multiplying the transformed data by a filter function to correct the spectral distribution of the projection signal, and the filter function uses a Hamming window.

8. The method according to claim 1, wherein The step of performing local reconstruction using the improved filtered back-projection algorithm includes: Calculating the projection coordinates of the reconstructed pixel points at each angle; Calculating the pixel values through bilinear interpolation and gradually constructing the image information of the local region.

9. The method according to claim 1, wherein Calculating the projection coordinates of the reconstructed pixel points at each angle and calculating the pixel values through bilinear interpolation, specifically including: Calculating the projection coordinates (m, n) in the detector at the angle θ between the line connecting the detector and the radiation source of the reconstructed pixel point f(x, z, y): Wherein, D2 represents the distance between the radiation source and the object center; Finally, calculating the value P(u, v, θ) of the point (u, v) through bilinear interpolation to calculate the pixel values of each reconstructed pixel point; After traversing and updating all angles θ i Finally, the final reconstruction result is obtained.

10. A battery pack body detection imaging device based on circular trajectory scanning, characterized in that, The device includes: A trajectory generation module for inputting the three-dimensional dimensions of the battery pack and information on key detection regions to generate a segmented circular trajectory with a multi-center distribution; A data acquisition module for the radiation source and the detector to move collaboratively along the planned circular trajectory to complete the acquisition of projection data for each local circular trajectory segment by segment; A reconstruction and stitching module is used to perform local reconstruction on the acquired projection data by using an improved filtered back-projection algorithm, and stitch the tomographic images of each local area to form a complete detection image of the battery pack body.

Images