Local dynamic enhanced visual muon imaging method based on magnetic focusing

Abstract

The invention provides a local dynamic enhanced visual muon imaging method based on magnetic focusing. The method comprises the following steps: after muons are focused, receiving muon signal data of a small area through a detector; preprocessing the muon signal data of different angles in the same area to obtain a projection data set; performing imaging reconstruction on the projection data set by adopting a three-dimensional reconstruction technology; and obtaining the three-dimensional density distribution of the target area. According to the invention, the multi-stage superconducting solenoids or quadrupole iron are utilized to actively regulate and focus muons, so that the signal intensity is greatly improved, the imaging period is greatly shortened, and the imaging quality is improved; by controlling the magnetic field intensity and direction of the multistage superconducting solenoids, the muon flux from a specific direction can be selectively enhanced, directional imaging is realized, and the imaging speed is improved by 5-8 times; by alternately irradiating the target, multiple groups of complementary projection data are obtained, the problem of projection missing in traditional muon imaging is improved, and the sampling rate of three-dimensional reconstruction is improved by more than three times.

Description

Technical Field

[0001] The present invention relates to the technical field of muon imaging, and in particular to a local dynamic enhanced visualization muon imaging method based on magnetic focusing. Background Art

[0002] In large, complex structures like nuclear power plants and skyscrapers, which are expected to operate for extended periods of time, assessing damage and defects in key internal components is crucial. For example, during the operation of nuclear reactors, core fuel assemblies can develop defects such as stress deformation, fractures, and cladding cracks. These defects can impact the lifespan and safety of nuclear power plants, necessitating frequent inspections.

[0003] In related technologies, fuel assembly damage is typically detected through visual inspection, temperature anomaly detection, and radiation anomaly detection. Visual inspection is a general check of the fuel assembly's external integrity; temperature anomaly detection measures the temperature at key locations within the core and cooling circuit to determine whether the core is operating normally; and radiation anomaly detection measures radiation levels at key locations within the containment vessel to determine whether the fuel cladding and cooling circuit are damaged.

[0004] However, conventional techniques typically rely on a few non-visual measurements combined with existing experience to assess damage within the reactor core. This observational approach is less intuitive and prone to misjudgment. Furthermore, these methods cannot accurately determine the location of damage, requiring downtime and inspection, resulting in long maintenance cycles and high costs.

[0005] In order to solve the above technical problems, the existing technology uses cosmic muon imaging for detection. There are two main principles of muon imaging commonly used, namely Muon Scattering Tomography (MST) and Muon Transmission Imaging (MTI). However, this technology still faces problems such as low detection efficiency, low imaging clarity, and environmental interference caused by the low natural muon flux. The imaging resolution of existing cosmic ray muon imaging technology is usually in the meter to ten-meter level, which is much lower than the millimeter-level imaging resolution of X-ray imaging technology, and the imaging time also takes several days to several months. Moreover, since high-energy particles are incident from all angles in a completely random manner, their flux, energy, and direction cannot be artificially controlled, resulting in fundamental passive limitations in the entire imaging process. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides a local dynamic enhanced visualization muon imaging method based on magnetic focusing. The present invention can improve the imaging quality and speed, and realize effective dynamic enhanced visualization muon imaging monitoring of large structures.

[0007] The technical solution of the present invention is: a local dynamic enhanced visualization muon imaging method based on magnetic focusing, comprising the following steps:

[0008] S1), after focusing the muons, the detector receives the muon signal data of a small area;

[0009] S2), acquiring muon signal data at different angles in the same area according to the method of step S1), and preprocessing the acquired muon signal data to obtain a projection data set;

[0010] S3) Using three-dimensional reconstruction technology to reconstruct the projection data set to obtain the three-dimensional density distribution of the target area.

[0011] Preferably, in step S1), muons are focused using a multi-stage superconducting solenoid or a quadrupole.

[0012] Preferably, in step S1), the diameter of the multi-stage superconducting solenoid is gradually reduced along the muon incident direction; thus achieving a high muon flux in a small area.

[0013] Preferably, in step S2), the preprocessing includes noise filtering, detector response non-uniformity correction, hardening effect compensation, and logarithmic transformation.

[0014] Preferably, in step S3), three-dimensional reconstruction of the projection data set is performed using three-dimensional reconstruction technology, specifically comprising the following steps:

[0015] S31), applying a ramp filter or an improved window function filter to the projection data at each angle to perform frequency domain convolution to eliminate star-shaped artifacts and enhance high-frequency details;

[0016] S32), projecting the filtered projection data back along the ray path into the two-dimensional voxel space for accumulation to obtain a two-dimensional slice;

[0017] S33), achieving layer-by-layer reconstruction of the tomographic image sequence by sequential stacking to obtain a three-dimensional density distribution map of the target area;

[0018] S34) splicing the three-dimensional density distribution maps at different times according to the position information to obtain the overall three-dimensional density distribution map of the target area.

[0019] Preferably, in step S31), a ramp filter or an improved window function filter is applied to the projection data at each angle to perform frequency domain convolution, specifically comprising the following steps:

[0020] S311) After the muon passes through the object, the signal received by the detector is recorded as p θ(u), where θ is the current scanning angle; u is the position coordinate of the signal on the detector;

[0021] S312), the received signal p is transformed by Fourier transform θ (u) Convert from spatial domain to frequency domain signal P θ (ω), that is:

[0022] P θ (ω)=F{p θ (u)};

[0023] Where F represents Fourier transform; ω is a frequency domain variable, representing the frequency component of the signal;

[0024] S313), introduce the ramp filter function H Ram-Lak (ω) for frequency domain P θ (ω) performs filtering operation to sharpen the high-frequency edge of the signal; wherein the ramp filter function H Ram-Lak The expressions of (ω) and filtering operation are:

[0025]

[0026] Where |ω| is the ramp function; is a rectangular window; ω c To limit the cutoff frequency;

[0027] is the frequency domain signal after filtering;

[0028] S314), the filtered frequency domain signal The filtered spatial signal is obtained by inverse Fourier transform

[0029] Preferably, in step S32), the filtered projection data is reversely projected along the ray path into the two-dimensional voxel space for accumulation to obtain a two-dimensional slice; specifically, the steps include:

[0030] S321), the object area to be reconstructed is divided into two-dimensional voxel grids of equal size, and the coordinate of each voxel grid i is (x i ,y i ), the voxel grid size matches the detector resolution;

[0031] S322) Calculate the trajectory of each muon signal and determine whether each voxel grid i is passed by the current muon trajectory; if the voxel grid i is passed by the ray, the spatial domain signal of the voxel grid i after filtering is converted to The weight is accumulated to the voxel grid i according to the length of the ray passing through the voxel; that is:

[0032]

[0033] Where ΔL i is the path length of the muon track through voxel grid i; is the voxel value of voxel grid i, and its initial value is 0;

[0034] S323) After all angle projections are completed, the voxel values ​​are normalized to eliminate the accumulated errors and obtain a two-dimensional slice.

[0035] Preferably, in step S33), layer-by-layer reconstruction of the tomographic image sequence is achieved by sequential stacking, specifically comprising the following steps:

[0036] S331), the two-dimensional slice obtained in step S32) is z i The coordinates are stacked sequentially to generate the initial three-dimensional volume data V(x i ,y i ,z i );

[0037] S332) Apply Gaussian smoothing or non-local mean filtering to the initial three-dimensional volume data to obtain a continuous single-moment three-dimensional volume density distribution map.

[0038] The beneficial effects of the present invention are:

[0039] 1. The present invention uses multi-stage superconducting solenoids or quadrupole iron to actively control and focus muons, concentrating incident muons from a larger range into a smaller area to achieve rapid muon imaging enhancement in a small area. This not only significantly increases signal intensity, but also significantly shortens the imaging cycle and improves imaging quality.

[0040] 2. By controlling the magnetic field strength and direction of the multi-stage superconducting solenoid, the present invention can selectively enhance the muon flux from a specific direction, thereby achieving directional imaging. This directional imaging increases the imaging speed of a specific area by 5-8 times;

[0041] 3. The present invention utilizes multiple sets of multi-stage superconducting solenoids in conjunction with corresponding detectors to alternately illuminate the target at different angles, obtaining multiple sets of complementary projection data. This improves the projection loss problem in traditional muon imaging and increases the sampling rate of three-dimensional reconstruction by more than three times. Furthermore, data from different illumination groups can be mutually verified, effectively reducing the uncertainty of the reconstruction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Schematic diagram of the process of the present invention;

[0043] Figure 2 Schematic diagram of the layout of the multi-stage superconducting solenoid of the present invention;

[0044] Figure 3 Schematic diagram of the structure of the superconducting solenoid of the present invention;

[0045] Figure 4 Schematic diagram of the structure of the quadrupole iron of the present invention. DETAILED DESCRIPTION

[0046] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0047] like Figure 1 As shown, this embodiment provides a local dynamic enhanced visualization muon imaging method based on magnetic focusing, comprising the following steps:

[0048] S1), after focusing the muons, the detector receives the muon signal data of a small area;

[0049] This embodiment utilizes multi-stage superconducting solenoids or quadrupoles to focus muons. Furthermore, one or more multi-stage superconducting solenoids or quadrupoles and their corresponding detectors are combined into an illumination group. This embodiment allows for cross-illumination of the same area in different directions based on the illumination group. Furthermore, the number of detectors can be far greater than the number of solenoids or quadrupoles. This allows for simultaneous transmission and scattering imaging of muons to numerically reconstruct the interior of the material / structure / structure being inspected. Combined with the locally enhanced imaging of the solenoids and quadrupoles, this results in a complete, non-destructive, dynamically enhanced visualization.

[0050] S2), acquiring muon signal data at different angles in the same area according to the method of step S1), and preprocessing the acquired muon signal data to obtain a projection data set;

[0051] S3) Using three-dimensional reconstruction technology to reconstruct the projection data set to obtain the three-dimensional density distribution of the target area.

[0052] As a priority of this embodiment, in step S1), the diameter of the multi-stage superconducting solenoid is gradually reduced along the muon incident direction; a high muon flux is achieved in a small area, such as Figure 2 shown.

[0053] As a priority of this embodiment, in step S1), if Figure 3 As shown, the superconducting solenoid is composed of a long straight spiral coil. When energized, a uniform axial magnetic field is generated inside the coil; it has a weak focusing effect on the particle beam and can suppress lateral divergence.

[0054] like Figure 4As shown, the quadrupole iron is composed of four groups of symmetrically arranged magnetic poles, and the polarities of adjacent magnetic poles are opposite; for example, NSNS; while playing a focusing role on the X axis, it plays a defocusing role on the Y axis. F represents the magnetic field force exerted on the muon. In this embodiment, simultaneous focusing on the X axis and the Y axis can be achieved by cooperating with multiple groups of quadrupole iron.

[0055] As a priority in this embodiment, in step S2), the preprocessing includes noise filtering, detector response non-uniformity correction, hardening effect compensation, and logarithmic transformation. The detector response non-uniformity correction refers to optimizing and adjusting each detector in the detector array through time and energy spectrum response correction methods to eliminate interference from the detector's own non-ideal characteristics on the measurement results.

[0056] The hardening effect compensation mentioned above refers to the use of machine deep learning methods to compare multiple different muon signal data and infer the true linear attenuation coefficient and density distribution in the detection area to achieve correction of image cupping artifacts and streak artifacts;

[0057] The logarithmic transformation process is to convert the exponential attenuation of the ray intensity into linear projection data so that the attenuation value is proportional to the thickness / density of the object.

[0058] As a preferred embodiment of this invention, in step S3), the projection data set is image reconstructed using a three-dimensional reconstruction technology, which specifically includes the following steps:

[0059] S31), applying a ramp filter or an improved window function filter to the projection data at each angle to perform frequency domain convolution to eliminate star-shaped artifacts and enhance high-frequency details; specifically comprising the following steps:

[0060] S311) After the muon passes through the object, the signal received by the detector is recorded as p θ (u), where θ is the current scanning angle; u is the position coordinate of the signal on the detector;

[0061] S312), the received signal p is transformed by Fourier transform θ (u) Convert from spatial domain to frequency domain signal P θ (ω), that is:

[0062] P θ (ω)=F{p θ (u)};

[0063] Where F represents Fourier transform; ω is a frequency domain variable, representing the frequency component of the signal;

[0064] S313), introduce the ramp filter function H Ram-Lak (ω) for frequency domain P θ(ω) performs filtering operation to sharpen the high-frequency edge of the signal; wherein the ramp filter function H Ram-Lak The expressions of (ω) and filtering operation are:

[0065]

[0066] Where |ω| is the ramp function; is a rectangular window; ω c To limit the cutoff frequency;

[0067] is the frequency domain signal after filtering;

[0068] S314), the filtered frequency domain signal The filtered spatial signal is obtained by inverse Fourier transform

[0069] S32), projecting the filtered projection data back along the ray path into the three-dimensional voxel space for accumulation to obtain a two-dimensional slice; specifically comprising the following steps:

[0070] S321), the object area to be reconstructed is divided into two-dimensional voxel grids of equal size, and the coordinate of each voxel grid i is (x i ,y i ), the voxel grid size matches the detector resolution;

[0071] S322) Calculate the trajectory of each muon signal and determine whether each voxel grid i is passed by the current muon trajectory; if the voxel grid i is passed by the ray, the spatial domain signal of the voxel grid i after filtering is converted to The weight is accumulated to the voxel grid i according to the length of the ray passing through the voxel; that is:

[0072]

[0073] Where ΔL i is the path length of the muon track through voxel grid i; is the voxel value of voxel grid i, and its initial value is 0;

[0074] S323) After all angle projections are completed, the voxel values ​​are normalized to eliminate the accumulated errors and obtain a two-dimensional slice.

[0075] S33) reconstructing the tomographic image sequence layer by layer by sequential stacking to obtain a three-dimensional density distribution map of the target area; specifically comprising the following steps:

[0076] S331), the two-dimensional slice obtained in step S32) is z i The coordinates are stacked sequentially to generate the initial three-dimensional volume data V(x i,y i ,z i );

[0077] S332) Apply Gaussian smoothing or non-local mean filtering to the initial three-dimensional volume data to obtain a continuous single-moment three-dimensional volume density distribution map.

[0078] S34) splicing the three-dimensional density distribution maps at different times according to the position information to obtain the overall three-dimensional density distribution map of the target area.

[0079] The above embodiments and descriptions are only for explaining the principles and best embodiments of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, which shall fall within the scope of the invention to be protected.

Claims

1. A local dynamic enhanced visualization muon imaging method based on magnetic focusing, characterized in that: The following steps are involved: S1), after focusing the muons, the detector receives the muon signal data of a small area; S2), acquiring muon signal data at different angles in the same area according to the method of step S1), and preprocessing the acquired muon signal data to obtain a projection data set; S3) Using three-dimensional reconstruction technology to reconstruct the projection data set to obtain the three-dimensional density distribution of the target area.

2. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 1, characterized in that: In step S1), muons are focused using a multi-stage superconducting solenoid or a quadrupole.

3. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 2, characterized in that: In step S1), the diameter of the multi-stage superconducting solenoid is gradually reduced along the muon incident direction; a high muon flux is achieved in a small area.

4. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 3, characterized in that: The superconducting solenoid is composed of a long straight spiral coil, and a uniform axial magnetic field is generated inside the coil when electricity is applied.

5. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 2, characterized in that: The quadrupole iron is composed of four groups of symmetrically arranged magnetic poles, and the polarities of adjacent magnetic poles are opposite.

6. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 1, characterized in that: In step S3), the projection data set is image reconstructed using a three-dimensional reconstruction technique, specifically comprising the following steps: S31), applying a ramp filter or an improved window function filter to the projection data at each angle to perform frequency domain convolution to eliminate star-shaped artifacts and enhance high-frequency details; S32), projecting the filtered projection data back along the ray path into the two-dimensional voxel space for accumulation; S33), achieving layer-by-layer reconstruction of the tomographic image sequence by sequential stacking to obtain a three-dimensional density distribution map of the target area; S34) splicing the three-dimensional density distribution maps at different times according to the position information to obtain the overall three-dimensional density distribution map of the target area.

7. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 6, characterized in that: In step S31), a ramp filter or an improved window function filter is applied to the projection data of each angle to perform frequency domain convolution, which specifically includes the following steps: S311) After the muon passes through the object, the signal received by the detector is recorded as p θ (u), where θ is the current scanning angle; u is the position coordinate of the signal on the detector; S312), the received signal p is transformed by Fourier transform θ (u) Convert from spatial domain to frequency domain signal P θ (ω), that is: P θ (ω)=F{p θ (u)}; Where F represents Fourier transform; ω is a frequency domain variable, representing the frequency component of the signal; S313), introduce the ramp filter function H Ram-Lak (ω) for frequency domain P θ (ω) performs filtering operation to sharpen the high-frequency edge of the signal; wherein the ramp filter function H Ram-Lak The expressions of (ω) and filtering operation are: Where |ω| is the ramp function; is a rectangular window; ω c To limit the cutoff frequency; is the frequency domain signal after filtering; S314), the filtered frequency domain signal The filtered spatial signal is obtained by inverse Fourier transform 8. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 7, characterized in that: In step S32), the filtered projection data is projected back along the ray path into the two-dimensional voxel space for accumulation to obtain a two-dimensional slice. The specific steps include: S321), the object area to be reconstructed is divided into two-dimensional voxel grids of equal size, and the coordinate of each voxel grid i is (x i ,y i ), the voxel grid size matches the detector resolution; S322) Calculate the trajectory of each muon signal and determine whether each voxel grid i is passed by the current muon trajectory; if the voxel grid i is passed by the ray, the spatial domain signal of the voxel grid i after filtering is converted to The weight is accumulated to the voxel grid i according to the length of the ray passing through the voxel; that is: Where ΔL i is the path length of the muon track through voxel grid i; is the voxel value of voxel grid i, and its initial value is 0; S323) After all angle projections are completed, the voxel values ​​are normalized to eliminate the accumulated errors and obtain a two-dimensional slice.

9. The method for local dynamic enhanced visualization muon imaging based on magnetic focusing according to claim 6, characterized in that: In step S33), the layer-by-layer reconstruction of the tomographic image sequence is achieved by sequential stacking, which specifically includes the following steps: S331), the two-dimensional slice obtained in step S32) is z i The coordinates are stacked sequentially to generate the initial three-dimensional volume data V(x i ,y i ,z i ); S332) Apply Gaussian smoothing or non-local mean filtering to the initial three-dimensional volume data to obtain a continuous single-moment three-dimensional volume density distribution map.

Images