CT (Computed Tomography) detection method for local accelerator of large-diameter rotary part shell
Through the multiple local source straight line scanning CT method, local CT imaging detection of large-diameter sysfunction parts is solved, and the insufficient projection coverage angle caused by the limitation of the radiation tension angle and the detector size is achieved, and a relatively complete reconstruction of defect structures and High-quality imaging.
Patent Information
- Application Number
- CN202410913593.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-09
- Publication Date
- 2024-11-29
AI Technical Summary
The prior art is difficult to conduct comprehensive local CT imaging detection of large-diameter slewing body parts when the ray tension angle and detector size are limited, resulting in insufficient projection coverage angle and inability to meet actual detection needs.
Multiple local source linear scanning CT (mL-STCT) method is used to rotate the detection object and biased ray source and detector to scan the estimated defects of the shell of the detection object, and each scan is different. The data is projected at the angle and image reconstruction is performed using the SIRT algorithm to scan multiple times.
Through multiple scans, the projection coverage angle is increased, and the relatively complete defect structure is reconstructed to meet actual detection needs and improve imaging quality, especially when the ray tension angle and detector size are limited.
Smart Images

Figure CN119044217A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of scanning detection and analysis, and relates to a local accelerator CT detection method for a large-diameter rotating body component shell. Background Art
[0002] Computed tomography (CT) can perform three-dimensional imaging of the internal structure of an object without damaging the object, and is widely used in fields such as materials science, medical diagnosis, and industrial testing. Traditional CT technology is usually a circular scan, in which the radiation source and detector rotate relative to the object to be tested to obtain complete tomographic scan data and achieve accurate reconstruction of the object to be tested.
[0003] Large-diameter rotating parts have important applications in real life, and nondestructive testing can avoid disasters and reduce economic losses. In actual testing, large-diameter rotating parts are large in size and contain high-density materials, which cannot be penetrated by conventional X-rays. Electron linear accelerators are required as ray sources, and the ray angle is usually no more than 30°. Based on the existing digital imaging (DR) system and existing detection motion, full-section scanning imaging of the detection object can be achieved, but the detection time is long and the cost of achieving precise rotation of the detection object is high. Usually, the defects of the shell part of the detection object are concerned, such as strip defects, which are distributed along the circumference tangentially. It can be considered to perform local CT imaging of the shell part. In response to the above detection needs, based on the existing accelerator imaging system, an imaging method is studied to realize local CT imaging detection of the shell of large-diameter rotating parts, which has important practical application value.
[0004] For the inspection of large solid rocket engines, Shi Yuanyuan proposed a tangential CT (TCT) technology. By offsetting the object to be tested, the main X-ray beam is tangent to the shell, and the object to be tested is rotated within a certain angle range to detect debonding defects; Chen Qinggui used narrow-angle fan beam industrial CT detection technology and proposed a translation / rotation scanning method based on TCT technology. For external CT imaging detection of objects, Zeng established a two-dimensional external fan beam CT scanning model. During the scanning process, the object to be tested is fixed, and the ray source and detector rotate along a circular orbit; Riis proposed an eccentric scanning method for the inspection of underwater oil pipelines. The detector and ray source are offset on both sides of the pipeline to be tested, and rotated around the center of the pipeline to be tested, which can realize the detection of defects such as underwater pipeline cracks. The above scanning imaging method obtains projection data at different angles through the relative rotation of the object, the ray source and the detector, and has high requirements for rotation accuracy.
[0005] In recent years, linear scanning CT has received widespread attention. It has the advantages of simple mechanical structure, easy implementation, and fast scanning speed. It is more suitable for unconventional detection scenarios such as fixed pipelines and industrial water flow detection. For the detection of fixed pipelines, Liu studied a scanning method in which the ray source and detector move synchronously in a straight line. For the detection of cables in corners, Schon proposed a linear scanning CT in which the movement trajectory of the ray source is perpendicular to the detector. Liu proposed an ultra-low-cost relatively parallel linear scanning CT, in which the detection object is fixed and the ray source and detector move relative to each other in a straight line. Zhang designed a scanning imaging system in which the ray source array and the detector are linearly symmetrically distributed. The ray source and detector do not need to be moved during the scanning process. In 2021, Yu proposed a source linear scanning CT (STCT) imaging method, in which the object is placed close to the ray source, the detector is fixed during the scanning process, and the ray source moves in a straight line parallel to the detector to collect projection data at different angles, and achieves at least 180° projection coverage angle through multi-segment scanning. For local defect detection of in-service high-voltage cables, Liao Mingjuan proposed a local source linear scanning CT (L-STCT) imaging method, which places the local imaging center at the maximum projection coverage angle. The scanning method is similar to STCT, except that the L-STCT scanning detector is placed close to the cable under test. The L-STCT scanning motion method is simple and the imaging speed is fast. It can realize local CT imaging detection of the outside of the object. However, when the ray angle is small and the diameter of the object under test is much larger than the detector size, the L-STCT scanning projection coverage angle is limited, which is difficult to meet the actual detection needs.
[0006] Therefore, a new scanning imaging method is urgently needed to realize the inspection of large-diameter rotating parts. Summary of the invention
[0007] In view of this, the purpose of the present invention is to provide a local accelerator CT detection method for the shell of a large-diameter rotating body component, so as to solve the problem of increasing the projection coverage angle through multiple scans when the ray angle and the detector size are limited, reconstruct a more complete defect structure, and meet actual detection needs.
[0008] In order to achieve the above object, the present invention provides the following technical solutions:
[0009] A local accelerator CT detection method for the shell of a large-diameter rotating body component, in view of the characteristics that large-diameter rotating body components have a large diameter and high-density materials, and defects are usually located in the shell, an electron linear accelerator is used as a ray source, and a multiple local source linear scanning CT (mL-STCT) imaging method is designed. The method specifically includes the following steps:
[0010] S1: Establishing the mL-STCT geometric model, where mL-STCT stands for multiple local source linear scanning CT. Specifically, by rotating the inspection object and offsetting the ray source and detector, the estimated defective part of the inspection object shell is scanned multiple times, and the ray source is moved each time to collect projection data at different angles;
[0011] S2: Analyze the different distributions of mL-STCT projection data in Radon space, and determine the rotation angle and related imaging parameters of each scanned object based on the geometric relationship, then rotate the detected object and move the radiation source and detector downward to realize mL-STCT scanning;
[0012] S3: Use SIRT (simultaneous iterative reconstruction technology) algorithm to reconstruct the image of the projection data of multiple scans.
[0013] Further, in step S1, an mL-STCT geometric model is established, which specifically includes: using an electron linear accelerator as a radiation source, the radiation source and the flat-panel detector are offset and located on both sides of the object to be inspected, during the scanning process, the object to be inspected and the detector are fixed, and the radiation source moves linearly from bottom to top in a direction parallel to the detector; in a single scan, the center of the imaging area is located at the intersection of a line connecting the upper end of the radiation source trajectory and the lower end of the detector and a line connecting the lower end of the radiation source trajectory and the upper end of the detector, so as to obtain the best imaging quality; after a single scan is completed, multiple scans are achieved by rotating the object to be inspected and moving the detector and the radiation source downward; an o-xyz coordinate system is established with the center of the object as the origin, the x-axis is parallel to the radiation source movement trajectory, the radiation source movement direction is the positive direction, the y-axis is perpendicular to the detector plane, the direction pointing to the detector is the positive direction, and the z-axis is parallel to the detector row array direction;
[0014] Take the xoy plane for analysis, and assume that the number of scans is T. The two-dimensional geometric model of the tth (t=1,2,...,T) scan is: when t is an odd number, the object under test rotates clockwise around the origin o, and when t is an even number, the object under test rotates counterclockwise around the origin o; let R be the radius of the object under test; α is the ray angle of the electron linear accelerator; 2s is the length of the moving trajectory of the ray source, 2d is the width of the detector, and the line connecting the center of the moving trajectory of the ray source and the center of the detector is always parallel to the y-axis; l and h are the horizontal distances from the center o of the object under test to the moving trajectory of the ray source and the detector, respectively; after the horizontal distance (l+h) from the ray source trajectory to the detector is determined, the length of the moving trajectory of the ray source can be expressed as
[0015]
[0016] Let θ t D is the rotation angle of the object around the origin o in the tth scan, and counterclockwise rotation is defined as positive; tis the distance from the line connecting the center of the moving trajectory of the ray source and the center of the detector to the center of the object under inspection for the tth scanning. According to the geometric relationship,
[0017] D t =R cosθ t (2)
[0018] Connect the upper end of the ray source trajectory and the lower end of the detector, and the lower end of the ray source trajectory and the upper end of the detector respectively. Let the intersection of the two lines be Q. In a single scan, the estimated defect position of the object to be inspected is placed at point Q. At this time, s / d=l / h must be satisfied, and point Q is always on the x-axis.
[0019] Further, in step S2, the distribution of mL-STCT projection data in Radon space specifically includes: taking the center of the imaging area as the origin o′, establishing a new coordinate system x′o′y′; the horizontal distances from the center of the imaging area to the ray source trajectory and the detector are l′ and h′ respectively, according to the geometric relationship:
[0020] l′=lR sinθ t (3)
[0021] h′=h+R sinθ t (4)
[0022] A point in Radon space represents an incident angle of The projection data obtained by the ray at a distance r from the imaging center;
[0023] Assume that the projection data corresponds to the left intersection point L of the straight line r = 0 Value is The right intersection point R corresponds to Value is For the tth scan, we have
[0024]
[0025] For a single scan, the rotation angle θ1 = 0, l′ and h′ can be calculated by equations (3) and (4), thus: and It can be directly calculated by equations (5) and (6); for multiple scans, the rotation angle of the object for each scan must be determined first. When the projection is continuous, the left endpoints and right endpoints of two adjacent segments of projection data correspond to The values are equal, that is
[0026]
[0027] Combine equations (3) to (4), (5) to (7), and calculate the value obtained by a single scan. and Substituting, we can calculate the critical rotation angle θ of the t-th scanC ; When the tth scanning rotation angle is greater than θ C When the t-th scan rotation angle is equal to θ C , the projection is continuous.
[0028] Further, in step S2, due to the limitation of the electron linear accelerator ray angle and the detector size, despite multiple scans, the projection coverage angle is still less than 180°. The scanning parameters can be adjusted to maximize the projection data in the imaging center area and improve the imaging quality. In actual detection, the estimated defect position is placed at the intersection of the upper end of the ray source trajectory-the lower end of the detector and the lower end of the ray source trajectory-the upper end of the detector, that is, s / d=l / h is satisfied, and a single scan can obtain the maximum projection coverage angle. Under the same number of scans, the projection coverage angle of multiple scans is the largest when the projection data of each segment is continuous. At this time, the rotation angle of the object each time is scanned satisfies θ t =θ C .
[0029] Further, in step S3, the image is reconstructed using the SIRT algorithm. The reconstruction process can be expressed as solving the following equation:
[0030] p=Af (8)
[0031] Where p=[p1,p2,...,p I ] T represents projection data, I represents the total number of rays; f=[f1,f2,...,f J ] T represents the image to be reconstructed, J represents the number of pixels in the reconstructed image, A=[a ij ](i=1,2,...,I;j=1,2,...,J) is the system matrix, a ij is the weight factor, which represents the contribution of the j-th pixel to the projection value of the i-th ray;
[0032] The SIRT reconstruction algorithm expression is:
[0033]
[0034] Where k is the number of iterations; λ is the relaxation factor, which is used to suppress overcorrection and affect the convergence speed of the iterative process; p i is the projection value of the i-th ray.
[0035] The beneficial effects of the present invention are:
[0036] (1) The present invention designs a multiple local source linear scanning CT (mL-STCT) imaging method. In view of the characteristics of large-diameter rotating body parts, such as large diameter, high-density materials, and defects usually located in the shell, an electron linear accelerator is used as a radiation source. By rotating the detection object and offsetting the radiation source and detector, the estimated defective parts of the detection object shell are scanned multiple times. Each time the radiation source is moved, projection data at different angles are collected, thereby realizing local CT imaging detection of the shell of large-diameter rotating body parts.
[0037] (2) When the ray angle and detector size are limited, the mL-STCT imaging method designed in the present invention can be used to increase the projection coverage angle through multiple scans, reconstruct a more complete defect structure, and meet actual detection needs.
[0038] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:
[0040] Figure 1 Schematic diagram of mL-STCT geometric model, where (a) is a three-dimensional geometric model; (b) is a two-dimensional geometric model;
[0041] Figure 2 is the two-dimensional geometric model in the new coordinate system;
[0042] Figure 3 The distribution of mL-STCT projection data, where (a) is a single scan; (b) is discontinuous projection; (c) is continuous projection;
[0043] Figure 4 is a simulation phantom, wherein (a) is a simulation phantom; (b) is a local enlarged view;
[0044] Figure 5 The reconstruction results under different projection data distributions, where (ad) is the reconstruction result of case 1; (eh) is the reconstruction result of case 2; (il) is the reconstruction result of case 3;
[0045] Figure 6 is the ROI magnification image, where (al) is Figure 5 Middle ROI zoomed in;
[0046] Figure 7 The reconstruction results of different imaging methods, where (a) is the phantom image; (be) are the reconstruction results of L-STCT, TCT, and mL-STCT when the projection is discontinuous and continuous; (fj) are the magnified images of ROI in (ae);
[0047] Figure 8 The 26th and 44th row cross-sectional views of the reconstructed images under different imaging methods, where (a) is the 26th row cross-sectional view; (b) is the 44th row cross-sectional view;
[0048] Fig. 9 Projection data distribution of different imaging methods, where (a) is mL-STCT; (b) is TCT;
[0049] Fig.10 The reconstruction results of mL-STCT with different scanning times, where (ad) are the reconstruction results of 2-5 scanning times respectively; (eh) is the enlarged image of ROI in (ad);
[0050] Fig.11 Reconstruction results of different imaging methods, where (ad) are the reconstruction results of L-STCT, TCT and mL-STCT with discontinuous and continuous projections, respectively; (eh) is the enlarged view of the ROI in (ad). DETAILED DESCRIPTION
[0051] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.
[0052] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.
[0053] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.
[0054] See also Figures 1 to 11 In response to the inspection needs of large-diameter rotating parts, inspired by multi-segment STCT scanning and L-STCT scanning, the present invention proposes a multiple local source linear scanning CT (mL-STCT) imaging method to perform local CT imaging inspection on the shell part of large-diameter rotating parts. The mL-STCT scanning method is that the detector and the radiation source are placed away from the center of the object. During the scanning process, the detector and the object to be inspected are fixed, and the radiation source moves in a straight line parallel to the detector to collect projection data at different angles. The center of the detector is aligned with the defects in the shell part of the inspected object to scan. After a single scan is completed, the inspected object is rotated, and the detector and the radiation source are moved downward by a certain distance to achieve multiple scans.
[0055] The present invention establishes an mL-STCT imaging model, analyzes the distribution of multiple scanning projection data in Radon space, calculates the rotation angle and related parameters of the object each time it is scanned, and designs numerical simulation experiments and simulated physical scanning experiments to verify the effectiveness of the mL-STCT scanning imaging method.
[0056] 1. mL-STCT imaging method
[0057] 1) Geometry model
[0058] During the scanning process, the inspected object and the detector are fixed, and the radiation source moves in a straight line from bottom to top in a direction parallel to the detector. Since the diameter of large-diameter rotating parts is too large, it is difficult to achieve global imaging, and only local imaging is performed on the shell. In a single scan, the center of the imaging area is located at the intersection of the line connecting the upper end of the radiation source trajectory and the lower end of the detector and the line connecting the lower end of the radiation source trajectory and the upper end of the detector to obtain the best imaging quality. After a single scan is completed, multiple scans are achieved by rotating the inspected object and moving the detector and radiation source downward. With the center of the object as the origin, an o-xyz coordinate system is established, such as Figure 1As shown in (a), the x-axis is parallel to the moving trajectory of the ray source, with the moving direction of the ray source as the positive direction, the y-axis is perpendicular to the detector plane, with the direction pointing to the detector as the positive direction, and the z-axis is parallel to the detector row array direction.
[0059] Take the xoy plane for analysis, assume the number of scans is T, and the two-dimensional geometric model of the tth (t=1,2,...,T) scan is as follows Figure 1 As shown in (b), when t is an odd number, the object under test rotates clockwise around the origin o, and when t is an even number, the object under test rotates counterclockwise around the origin o. Where R is the radius of the object under test; α is the ray angle of the electron linear accelerator; 2s is the length of the moving trajectory of the ray source, 2d is the width of the detector, and the line connecting the center of the moving trajectory of the ray source and the center of the detector is always parallel to the y-axis; l and h are the horizontal distances from the center o of the object under test to the moving trajectory of the ray source and the detector, respectively. After the horizontal distance (l+h) from the ray source trajectory to the detector is determined, the length of the moving trajectory of the ray source can be expressed as
[0060]
[0061] θ t D is the rotation angle of the object around the origin o in the tth scan, and counterclockwise rotation is defined as positive; t is the distance from the line connecting the center of the moving trajectory of the ray source and the center of the detector to the center of the object under inspection for the tth scanning. According to the geometric relationship,
[0062] D t =R cosθ t (2)
[0063] Connect the upper end of the ray source trajectory and the lower end of the detector, and the lower end of the ray source trajectory and the upper end of the detector respectively. Let the intersection of the two lines be Q. In a single scan, the estimated defect position of the object to be inspected is placed at point Q. At this time, s / d=l / h must be satisfied, and point Q is always on the x-axis.
[0064] 2) Projection data
[0065] To illustrate the projection coverage of mL-STCT scanning, Radon spatial analysis projection data is introduced. Taking the center of the imaging area as the origin o′, a new coordinate system x′o′y′ is established, as follows: Figure 2 As shown. The horizontal distances from the center of the imaging area to the ray source trajectory and the detector are l′ and h′ respectively. According to the geometric relationship,
[0066] l′=lR sinθ t (3)
[0067] h′=h+R sinθ t (4)
[0068] A point in Radon space represents an incident angle of The projection data obtained by the ray with a distance r from the imaging center is
[0069]
[0070] Where, β = arctan(l' / -x' s ). Among them, x′ d , x′ s are the horizontal coordinates of the intersection of the detector and the ray source trajectory with the ray in the coordinate system x′o′y′, and x′ d ∈[-d,-d],x′ s ∈[-s,s].
[0071] The imaging field of view is defined as the maximum imaging radius R of multiple scans with o′ as the center. FOV is a circular area with a radius of
[0072] R FOV =max{R t} t=1,...,T (7)
[0073] in
[0074]
[0075] In the formula, R t is the imaging radius of the tth scan, that is, the distance from the imaging center to the line connecting the upper end of the ray source trajectory and the lower end of the detector.
[0076] In a single scan, the center of the imaging area is placed at point Q to obtain the maximum projection coverage angle. The distribution of the projection data in Radon space is as follows: Figure 3 As shown in (a), the projection coverage angle is less than 180°. To obtain sufficient projection data, the object under inspection is rotated by θ around the origin o after a single scan. t , while the detector and the radiation source track move downward (RD t ), perform the tth scan, and each scan movement is the same until multiple scans are completed.
[0077] To obtain complete projection data required for reconstruction, each point in the imaging field of view must obtain at least 180° of projection coverage. Compared with a single scan, multiple scans can obtain more projection data, but the projection coverage is still less than 180°, and the amount of projection data at points with different distances from the imaging center is different. Relatively speaking, the amount of data at the imaging center is the largest. Therefore, according to the distribution of the projection data of the imaging center in the Radon space of multiple scans, it can be divided into two cases: (1) Projection discontinuity: When there are gaps between each segment of projection data, the projection data is discontinuous, such as Figure 3(b) (2) Continuous projection: When there is overlap or no overlap or gap between the projection data segments, the projection data is continuous. Since the overlap of projection data leads to data redundancy, the number of scans under the same projection coverage angle increases, which reduces the detection efficiency. Therefore, only the situation where there is no overlap or gap between the projection data segments is considered, such as Figure 3 (c) as shown.
[0078] In Radon space, the projection data obtained by multiple scans is equivalent to the projection data of a single scan along Axis translation rotation angle θ t , so the multiple scan projection data is shown as the projection data distributed on the right side of the single projection data when the object is rotated counterclockwise, and on the left side when it is rotated clockwise. Since the rotation angle of the object is different each time, it can be seen from equations (3) to (4) that l' and h' are different. For any rotation angle θ t , imaging center projection coverage angle It can be expressed as
[0079]
[0080] The larger the counterclockwise rotation angle, the larger h', and the smaller the corresponding projection coverage angle; similarly, the larger the clockwise rotation angle, the smaller l', and the smaller the corresponding projection coverage angle. The distribution shape of the projection data is different for each scan. The larger the rotation angle, the shorter the length of the line segment where the projection data intersects with the straight line r = 0 in the Radon space. Therefore, in order to improve the quality of the reconstructed image, the present invention rotates the object to be inspected alternately counterclockwise and clockwise to obtain a larger projection coverage angle as much as possible. In actual detection, after determining the rotation angle and related imaging parameters for each scan, the object to be inspected can be rotated in a single direction to realize mL-STCT scanning.
[0081] 3) Rotation angle analysis
[0082] In order to study the influence of the distribution of multiple scanning projection data on the quality of reconstructed images, the present invention analyzes two distributions of projection data in Radon space. By analyzing the geometric relationship, the distribution of each segment of projection data and the rotation angle θ of each scan are t And the projection coverage angle.
[0083] Assume that the projection data corresponds to the left intersection point L of the straight line r = 0 Value is The right intersection point R corresponds to Value is For the tth scan, we have
[0084]
[0085] For a single scan, the rotation angle θ1 = 0, l′ and h′ can be calculated by equations (3) and (4), thus: and It can be directly calculated by equations (10) and (11). For multiple scans, the rotation angle of the object for each scan must be determined first. When the projection is continuous, Figure 3 (c) It can be seen that the left endpoints and right endpoints of two adjacent projection data segments correspond to each other. The values are equal, that is
[0086]
[0087] Combining equations (3) to (4), (10) to (12), the single scan calculation results are and Substituting, we can calculate the critical rotation angle θ of the t-th scan C When the tth scanning rotation angle is greater than θ C When the t-th scan rotation angle is equal to θ C , the projection is continuous.
[0088] Based on the above analysis, due to the limitation of the electron linear accelerator's beam angle and the size of the detector, despite multiple scans, the projection coverage angle is still less than 180°. The imaging quality can be improved by adjusting the scanning parameters to maximize the projection data in the imaging center area. In actual detection, the estimated defect position is placed at the intersection of the upper end of the beam source trajectory-the lower end of the detector and the lower end of the beam source trajectory-the upper end of the detector, that is, s / d=l / h is satisfied, and the maximum projection coverage angle can be obtained with a single scan; under the same number of scans, the projection coverage angle is the largest when the projection data of each segment is continuous, and the rotation angle of the object each time is scanned satisfies θ t =θ C .
[0089] 4) Image reconstruction algorithm
[0090] During mL-STCT scanning, ray projection can only cover part of the inspected object, and the projection data is truncated. In addition, due to the limitations of the electron linear accelerator ray angle and the size of the detector, the projection coverage angle is less than 180°, and there are also limited angle and truncation problems. The analytical reconstruction algorithm is fast, but has high requirements for data completeness. Limited angle and truncation artifacts have a greater impact on the reconstruction quality, and serious artifacts will appear. The iterative reconstruction algorithm has lower requirements for data completeness. The SIRT algorithm uses all rays passing through all angles of a certain pixel to correct the pixel in one iteration process, and can reconstruct well from missing and truncated data. Therefore, the present invention uses the SIRT algorithm to reconstruct the image.
[0091] The image reconstruction process can be expressed as solving the following equation:
[0092] p=Af (13)
[0093] Where p=[p1,p2,...,p I ] T represents the projection data, and I represents the total number of rays. f=[f1,f2,...,f J ] T represents the image to be reconstructed, J represents the number of pixels in the reconstructed image, A=[a ij ](i=1,2,...,I;j=1,2,...,J) is the system matrix, a ij is the weight factor, which represents the contribution of the j-th pixel to the projection value of the i-th ray.
[0094] The SIRT reconstruction algorithm expression is:
[0095]
[0096] In the formula, k is the number of iterations, λ is the relaxation factor, which is used to suppress over-correction and affect the convergence speed of the iterative process, and p i is the projection value of the i-th ray.
[0097] 2. Experiment
[0098] To verify the effectiveness of the mL-STCT scanning imaging method, numerical simulation experiments and simulated physical scanning experiments were designed in this experiment. The ray angle was set to 30°, and the number of sampling points for each mL-STCT scan was 501. After the scan was completed, the SIRT algorithm was used for reconstruction, and the number of iterations was set to 1000. The reconstruction platform was CPU: Intel(R) Core i5-12500H, 2.5GHz, GPU: NVIDIA GeForce RTX 3050, the programming environment was MATLAB 2017b, and the SIRT reconstruction algorithm was implemented based on ASTRAToolbox.
[0099] 1) Numerical simulation experiment
[0100] According to the structure of large diameter rotating parts, design Figure 4 The phantom shown in the figure is used for simulation experiment, and the size of the phantom is 2048 pixels × 2048 pixels. Actual detection usually focuses on the defects in the shell part, so strip and hole defects are set in the shell part, where the strip defects are distributed along the circumference tangentially to simulate the debonding defects of the solid rocket engine. During the simulation process, mL-STCT scanning and reconstruction are performed on the defective parts of the phantom. The simulation experiment parameters are set according to the actual detection parameters, as shown in Table 1, and the size of the local reconstructed image is 200 pixels × 400 pixels.
[0101] Table 1 Simulation experiment parameters
[0102]
[0103]
[0104] The simulation experiment analyzes the reconstruction results of different projection data distributions, and compares the reconstruction results of local source linear scanning CT (L-STCT), tangential CT (TCT) and the mL-STCT imaging method proposed in this invention for local detection of large-diameter rotating body parts shells. Objective evaluation indicators such as root mean square error (RMSE), information fidelity criterion (IFC) and structural similarity (SSIM) are introduced to evaluate the quality of reconstructed images. Generally speaking, the smaller the RMSE, the larger the IFC and SSIM, the higher the quality of the reconstructed image.
[0105] 1.1) Simulation experiments on different projection data distributions
[0106] Firstly, the different distributions of multiple scanning projection data in Radon space are simulated. When the projection angle range does not exceed 180°, two groups of rotation angles of the inspected object are set when the projection is discontinuous, and the gaps between each segment of projection data are 10° and 5°, respectively, which are recorded as case 1 and case 2. When the projection is continuous, the rotation angle of the inspected object is the critical rotation angle, and the gap between each segment of projection data is 0°, which is recorded as case 3. The rotation angles of the inspected object in each scan in the three cases are shown in Table 2.
[0107] Table 2 Rotation angle of the object under each scan under different projection data distribution
[0108]
[0109] Figure 5 The reconstruction results of multiple scans in three cases. In the results of 2 to 5 scans, the projection coverage angles of case 1 are approximately 50°, 73°, 90° and 111°, the projection coverage angles of case 2 are approximately 51°, 75°, 92° and 114°, and the projection coverage angles of case 3 are approximately 52°, 77°, 95° and 117°. The shell parts are completely reconstructed. As the number of scans increases, the projection data increases, and the defect structure is restored more completely. However, the projection data is always less than 180°, and there are obvious limited angle artifacts in the image. Further analysis of the defect structure recovery, Figure 6 for Figure 5In the ROI magnified image, when the projection is discontinuous, each segment of the projection data has an artifact that is tangent to the feature boundary. The artifacts of multiple scans intersect to form the defect outline, resulting in the distortion of the defect structure. By comparing the results of case 1 and case 2, it can be seen that the larger the gap between the projection data segments, the more serious the distortion, and there are serious artifacts in the reconstructed image, which interferes with the identification of defects. When the projection is continuous, although the defect structure is limited due to the influence of limited angle artifacts, the artifacts in the reconstructed image are significantly reduced, the strip and hole defects are clearly visible, the image quality is high, and as the number of scans increases, the defect structure is restored more completely.
[0110] In order to further evaluate the restoration of defective structures under different projection data distributions, the objective evaluation indicators of the reconstructed images were compared and analyzed, and the results are shown in Table 3. Comparing the RMSE value, since the projection data is incomplete data and there is a truncation problem, it has an impact on the numerical value of the reconstruction result, and the projection coverage angle and truncation conditions are different under different projection data distributions, and there are slight differences in the results. Comparing the IFC and SSIM values, case 3 is the best, case 2 is the second, and case 1 is slightly worse, indicating that the reconstructed image is less distorted when the projection is continuous and is more similar to the true image in structure. When the projection is discontinuous, the larger the gap between each segment of projection data, the more serious the distortion of the reconstructed image. For the same projection data distribution, as the number of scans increases, the RMSE, IFC and SSIM indicators all become better. Therefore, the objective evaluation indicators show that the quality of the reconstructed image is higher when the projection is continuous, and the more scans, the better the reconstruction effect.
[0111] Table 3 Reconstructed image indicators under different projection data distributions
[0112]
[0113] 1.2) Comparative experiments of different imaging methods
[0114] To verify the effectiveness of the mL-STCT imaging method, the imaging effects of L-STCT, TCT and mL-STCT with continuous and discontinuous projections were compared under the same magnification ratio and detector size. The projection angle range and sampling points of TCT were set to be consistent with those of mL-STCT. Figure 7 As shown in the figure, the projection coverage angles of the imaging center for continuous and discontinuous projections of L-STCT, TCT and mL-STCT are 30°, 74°, 70° and 117° respectively. The shell is partially restored. Due to incomplete projection data, there are limited angle artifacts to varying degrees. Further comparison of defect structure restoration and ROI reconstruction effects, according to the visible boundary theory
[27] If a characteristic boundary of an object is tangent to a ray in a finite data set, then the boundary can be easily reconstructed from the finite data. L-STCT scanning is limited by the angle of the electron linear accelerator’s rays, resulting in serious data loss and severe limited-angle artifacts in the reconstructed image, which interferes with defect recognition. Strip defects are distributed along the circumferential tangent direction, with more tangent rays in the y-axis direction, resulting in better recovery. However, there are almost no tangent rays in the x-axis direction, so the contours of each segment of the strip defect are not clear in the x-axis direction. For hole-shaped defects, only the small segment contours tangent to the rays are restored, making it difficult to distinguish multiple defects, and the L-STCT reconstruction effect is poor. TCT scanning only affects the defect reconstruction results in a certain angle range. [5] , so the recovery of hole defects is limited, but multiple defects can be distinguished. Strip defects are distributed along the circumference tangentially and have more tangent rays, so they are basically restored. mL-STCT scanning offsets the ray source and detector to ensure that the center of the detector is aligned with the imaging center for each scan, and the projection data increases with the number of scans. When the projection is discontinuous, the number of scans decreases. Although the strip defects are basically restored, the hole defects have obvious distortion, which affects the identification of the defects; when the projection is continuous, the rays tangent to the defects increase, and both strip and hole defects are well restored. In the same projection angle range, compared with the reconstruction results of L-STCT, TCT and mL-STCT with discontinuous projection, the reconstruction results of mL-STCT with continuous projection have more complete defect structures, clearer contours, and better image reconstruction effects.
[0115] To further compare the differences in reconstruction results of different imaging methods, the gray values of the 26th row (the row where the center of the stripe defect is located) and the 44th row (the row where the center of the hole defect is located) of the phantom image, L-STCT, TCT, and mL-STCT reconstruction results are compared. The results are shown in Figure 2. Figure 8 As shown. For strip defects, the grayscale curves of different imaging methods have the same trend of change, and the continuous mL-STCT projection is closer to the model curve; for hole defects, affected by the limited angle artifact, the L-STCT and TCT grayscale curves are relatively gentle, and the discontinuous mL-STCT projection is distorted, which is quite different from the model curve. The continuous mL-STCT projection is closer to the model curve, indicating that the continuous mL-STCT projection is more accurate in reconstructing images.
[0116] The evaluation indicators of the reconstruction results of different imaging methods are shown in Table 4. Compared with the results of L-STCT, TCT and mL-STCT projection discontinuity, the RMSE, IFC and SSIM indicators of mL-STCT projection continuity are the best, indicating that its reconstructed image is closer to the true image, with less distortion and higher structural similarity. Therefore, quantitative indicators show that the imaging effect of mL-STCT projection continuity is better.
[0117] Table 4 Image indicators reconstructed by different imaging methods
[0118]
[0119] In summary, mL-STCT scans the local area multiple times to obtain projection data at different angles. The reconstruction effect is better when the projection is continuous. As the number of scans increases, the defect structure of the mL-STCT reconstructed image is restored more completely. For the local CT imaging of the shell of a large-diameter rotating body component, compared with the local reconstruction results of L-STCT and TCT scanning, the defect structure of the mL-STCT scanning reconstruction results is more complete, the contour is clearer, and the reconstructed image quality is higher when the mL-STCT projection is continuous.
[0120] 1.3) Analysis and discussion
[0121] The simulation results show that mL-STCT can effectively detect the shell of large-diameter rotating parts by performing multiple linear scans on the same local area. Limited by the ray angle and the size of the detector, the projection data has limited angle and truncation problems, but the strip and hole defects in the shell are well restored, and the defect structure is more complete as the number of scans increases. When the number of scans is the same, the distribution of projection data in Radon space shows that the smaller the gap between each segment of projection data, the more projection data there is, and the larger the equivalent projection coverage angle. Therefore, the image reconstruction effect is optimal when the projection is continuous.
[0122] In the same projection angle range, Figure 7 The simulation results shown in the figure show that both mL-STCT and TCT can recover strip defects distributed along the circumference tangentially. For hole defects, mL-STCT projection continuity is better than TCT. Fig. 9 As shown in the figure, with the increase of the number of scans of mL-STCT, the projection data of the imaging center continues to increase, and the projection coverage angle of 5 scans is about 117°; while under TCT scanning, the projection data of the imaging center only increases within a certain projection angle range, and the maximum projection coverage angle is about 74°. Therefore, under the same projection angle range, when the projection angle range exceeds the maximum projection coverage angle of TCT, the defect structure recovery of mL-STCT is better than that of TCT.
[0123] 2) Simulation of physical scanning experiment
[0124] In order to verify the actual reconstruction effect of the mL-STCT imaging method, a simulated mL-STCT experimental system was built in this experiment. Due to some limitations, the experiment simulated physical scanning imaging by proportionally reducing the object to be inspected and related imaging parameters. The simulation experimental system consists of a ray source, a flat-panel detector, two linear slides, a turntable and a computer system. The linear slides control the linear movement of the ray source and the offset of the object to be inspected, and the turntable controls the rotation of the object to be inspected. During the experiment, the object to be inspected was placed on the turntable, the flat-panel detector was fixed, and after the distance from the ray source to the detector was determined, the ray angle obtained by the detector was controlled to not exceed 30° by controlling the length of the ray source movement trajectory.
[0125] In this experiment, a sample was made to simulate the structure of a large-diameter rotating body component. The sample shell was made of PVC material and filled with flour. Strip defects with a length of 6 mm and a width of 0.3 mm and hole defects with a diameter of 2.5 mm were set in the shell. The outer diameter of the sample was 172.6 mm and the inner diameter was 160 mm. The actual diameter of the object to be tested was 1400 mm. Therefore, the relevant imaging parameters were proportionally reduced by about 8.1 times. The projection data collected by the 398×398 detectors in the center of the flat-panel detector was used. The sample was scanned 5 times based on the simulated mL-STCT experimental system. In the experiment, the center of the detector was aligned with the inner wall of the sample. The experimental parameters of the simulated physical scanning are shown in Table 5.
[0126] Table 5 Parameters of simulated physical scanning experiment
[0127]
[0128] Fig.10 The results of mL-STCT imaging reconstruction under different scanning times are shown in Figure 1. The local area of the sample shell is restored. Since the projection coverage angles of 2 to 5 scans are approximately 45°, 70°, 87° and 109°, the reconstruction results all have limited angle artifacts. Observing the ROI area, with the increase of the number of scans, the defect structure is gradually restored and the contour is gradually clear. For strip defects, the contour along the circumference is restored, the contour of the 2 scans along the circumference is unclear, and the 3 to 5 scans have a clear contour; for hole defects, multiple defects can be distinguished, but due to the limitation of the projection coverage angle, there are limited angle artifacts and only part of the contour can be restored. With the increase of the number of scans, the hole defect structure is gradually complete.
[0129] To further verify the effectiveness of mL-STCT, the reconstruction results of L-STCT, TCT and mL-STCT with discontinuous and continuous projections were compared under the same projection angle range and imaging parameters. Fig.11As shown. The L-STCT projection coverage angle is about 26°, the reconstruction result is poor, the defect outline is not clear, and the gray value is slightly different from the defect-free area, which makes it difficult to detect defects. The TCT projection coverage angle is about 71°, the strip defects are basically restored, and the hole defects have partially restored the outline. In the case of discontinuous projection, mL-STCT performs 3 scans, the projection coverage angle is about 65°, the defect structure is seriously distorted, and there is interference with the identification of defects; in the case of continuous projection, mL-STCT performs 5 scans, the projection coverage angle is about 109°, the reconstruction results show that the strip defects are clearly visible, the hole defect structure is relatively complete, and multiple strip and hole defects can be distinguished, which proves that the mL-STCT imaging method can effectively detect defects in the shell of large-diameter rotating parts in the case of continuous projection.
[0130] 3. Conclusion
[0131] In response to the inspection needs of large-diameter rotating parts, the present invention proposes an mL-STCT imaging method based on the existing accelerator imaging system. By establishing an mL-STCT geometric model, the distribution of multiple scan projection data in Radon space is analyzed, and the critical rotation angle of multiple scans is calculated in combination with geometric relationships. The above experiment compares and analyzes the simulation experiment results of multiple scans under different projection data distribution conditions. When each segment of projection data is continuous, the image reconstruction quality is higher. The results of simulation experiments and simulated physical scanning experiments show that compared with the reconstruction results of L-STCT, TCT and mL-STCT projection discontinuity, the strip and hole defects of the mL-STCT projection continuity of the present invention are better restored, and multiple segments of strip and hole defects can be clearly distinguished, realizing local CT imaging detection of the shell of large-diameter rotating parts.
[0132] The mL-STCT imaging method of the present invention has a simple mechanical structure and is easy to implement. The rotation angle of each scan does not have high requirements on the rotation accuracy. When the ray angle and the detector size are limited, the local area of the large-diameter rotating body component can be scanned multiple times to obtain projection data at different angles. However, the projection coverage angle is less than 180°, and the reconstructed image has limited-angle artifacts. In the future, it will be considered to study the optimization of the reconstruction algorithm to achieve high-quality imaging to better meet the actual detection needs.
[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.
Claims
1. A local accelerator CT detection method for the shell of a large-diameter rotating body component, characterized in that: The method specifically comprises the following steps: S1: Establishing the mL-STCT geometric model, where mL-STCT stands for multiple local source linear scanning CT. Specifically, by rotating the inspection object and offsetting the ray source and detector, the estimated defective part of the inspection object shell is scanned multiple times, and the ray source is moved each time to collect projection data at different angles; S2: Analyze the different distributions of mL-STCT projection data in Radon space, and determine the rotation angle and related imaging parameters of each scanned object based on the geometric relationship, then rotate the detected object and move the radiation source and detector downward to realize mL-STCT scanning; S3: Use the SIRT algorithm to reconstruct images from projection data of multiple scans.
2. The local accelerator CT detection method for the shell of a large-diameter rotating body component according to claim 1 is characterized in that: In step S1, an mL-STCT geometric model is established, which specifically includes: using an electron linear accelerator as a ray source, the ray source and the flat-panel detector are offset and located on both sides of the object to be inspected, during the scanning process, the object to be inspected and the detector are fixed, and the ray source moves linearly from bottom to top in a direction parallel to the detector; during a single scan, the center of the imaging area is located at the intersection of a line connecting the upper end of the ray source trajectory and the lower end of the detector and a line connecting the lower end of the ray source trajectory and the upper end of the detector to obtain the best imaging quality; after a single scan is completed, multiple scans are achieved by rotating the object to be inspected and moving the detector and the ray source downward; an o-xyz coordinate system is established with the center of the object as the origin, the x-axis is parallel to the ray source movement trajectory, the ray source movement direction is the positive direction, the y-axis is perpendicular to the detector plane, the direction pointing to the detector is the positive direction, and the z-axis is parallel to the detector row array direction; Take the xoy plane for analysis, and assume that the number of scans is T. The two-dimensional geometric model of the tth (t=1,2,...,T) scan is: when t is an odd number, the object under test rotates clockwise around the origin o, and when t is an even number, the object under test rotates counterclockwise around the origin o; let R be the radius of the object under test; α is the ray angle of the electron linear accelerator; 2s is the length of the moving trajectory of the ray source, 2d is the width of the detector, and the line connecting the center of the moving trajectory of the ray source and the center of the detector is always parallel to the y-axis; l and h are the horizontal distances from the center o of the object under test to the moving trajectory of the ray source and the detector, respectively; after the horizontal distance (l+h) from the ray source trajectory to the detector is determined, the length of the moving trajectory of the ray source is expressed as Let θ t D is the rotation angle of the object around the origin o in the tth scan, and counterclockwise rotation is defined as positive; t is the distance from the line connecting the center of the moving trajectory of the ray source and the center of the detector to the center of the object under inspection for the tth scanning. According to the geometric relationship, D t =Rcosθ t (2) Connect the upper end of the ray source trajectory and the lower end of the detector, and the lower end of the ray source trajectory and the upper end of the detector respectively. Let the intersection of the two lines be Q. In a single scan, the estimated defect position of the object to be inspected is placed at point Q. At this time, s / d=l / h must be satisfied, and point Q is always on the x-axis.
3. The local accelerator CT detection method for the shell of a large-diameter rotating body component according to claim 2 is characterized in that: In step S2, the distribution of mL-STCT projection data in Radon space specifically includes: taking the center of the imaging area as the origin o′, establishing a new coordinate system x′o′y′; the horizontal distances from the center of the imaging area to the ray source trajectory and the detector are l′ and h′ respectively, and according to the geometric relationship, l′=l-Rsinθ t (3) h′=h+Rsinθ t (4) A point in Radon space represents an incident angle of The projection data obtained by the ray at a distance r from the imaging center; Assume that the projection data corresponds to the left intersection point L of the straight line r = 0 Value is The right intersection point R corresponds to Value is For the tth scan, we have For a single scan, the rotation angle θ1 = 0, l′ and h′ can be calculated by equations (3) and (4), thus: and It can be directly calculated by equations (5) and (6); for multiple scans, the rotation angle of the object for each scan must be determined first. When the projection is continuous, the left endpoints and right endpoints of two adjacent projection data segments correspond to The values are equal, that is Combine equations (3) to (4), (5) to (7), and calculate the value obtained by a single scan. and Substitute and calculate the critical rotation angle θ for the tth scan C ; When the tth scanning rotation angle is greater than θ C When the t-th scan rotation angle is equal to θ C , the projection is continuous.
4. The local accelerator CT detection method for the shell of a large-diameter rotating body component according to claim 3 is characterized in that: In step S2, due to the limitation of the electron linear accelerator ray angle and the size of the detector, despite multiple scans, the projection coverage angle is still less than 180°. By adjusting the scanning parameters, the imaging center area has the most projection data and the imaging quality is improved. In actual detection, the estimated defect position is placed at the intersection of the upper end of the ray source trajectory-the lower end of the detector and the lower end of the ray source trajectory-the upper end of the detector, that is, s / d=l / h is satisfied, and a single scan can obtain the maximum projection coverage angle. Under the same number of scans, the projection coverage angle of multiple scans is the largest when the projection data of each segment is continuous. At this time, the rotation angle of the object each time is scanned satisfies θ t =θ C .
5. The local accelerator CT detection method for the shell of a large-diameter rotating body component according to claim 1 is characterized in that: In step S3, the image is reconstructed using the SIRT algorithm. The reconstruction process is expressed as solving the following equation: p=Af (8) Where p=[p1,p2,...,p I ] T represents projection data, I represents the total number of rays; f=[f1,f2,...,f J ] T represents the image to be reconstructed, J represents the number of pixels in the reconstructed image, A=[a ij ](i=1,2,...,I;j=1,2,...,J) is the system matrix, a ij is the weight factor, which represents the contribution of the j-th pixel to the projection value of the i-th ray; The SIRT reconstruction algorithm expression is: Where k is the number of iterations; λ is the relaxation factor, which is used to suppress overcorrection and affect the convergence speed of the iterative process; p i is the projection value of the i-th ray.
Citation Information
Cited By
Industrial X-ray detection manipulator
CN119224016A
Detection method and detection device for crude oil output hose
CN119619187A
