Detection device and method based on muon imaging technology by utilizing muon momentum

Abstract

The invention provides a muon imaging technology-based detection device and method using muon momentum. Relates to the fields of customs and frontier freight detection technology, safety and nuclear energy and nuclear technology. The detection device comprises a muon track acquisition module, a time-of-flight spectrometer, an electronics acquisition sub-device and a data analysis system, the muon track acquisition module comprises a muon incident track acquisition sub-module and a muon emergent track acquisition sub-module; the muon incident track acquisition sub-module and the muon emergent track acquisition sub-module are at least three high-position-resolution multi-air-gap resistive plate chambers which are arranged in parallel in a sealed cavity respectively; the time-of-flight spectrometer comprises a muon incidence time detection module and a muon emergence time detection module, and the muon incidence time detection module and the muon emergence time detection module are respectively at least one high-time-resolution multi-air-gap resistive plate chamber in the sealed cavity; the detection device is good in imaging performance, short in imaging time and high in robustness.

Description

Technical Field

[0001] This invention relates to customs and border control cargo inspection technology, security, and nuclear energy and nuclear technology, and particularly to a detection device and method based on muon imaging technology that utilizes muon momentum. Background Technology

[0002] Muon imaging technology has been recognized by the international radiation detection imaging and muon detection fields as a new imaging technology with great development value and application prospects in the 21st century. It is mainly divided into two types: muon transmission imaging and muon scattering imaging. Muon transmission imaging requires long exposure times for muons and is mainly suitable for fields such as geological surveying and archaeology. Muon scattering imaging, on the other hand, obtains the atomic number information of the analyte by analyzing the multiple Coulomb scattering angles of muons before and after passing through the material, resulting in higher imaging quality and shorter imaging time. However, the scattering angle of muons is related to their momentum, and existing muon scattering imaging devices cannot effectively obtain muon momentum information, leading to a significant discrepancy between the imaging results and theoretical predictions.

[0003] Therefore, there is an urgent need to study a high-performance muon detection device that utilizes muon momentum to realize the application of muon non-destructive testing technology in customs and border cargo inspection technology, safety, and nuclear energy and nuclear technology fields. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a detection device and method based on muon imaging technology that utilizes muon momentum. This device can be placed at customs border control locations to perform minute-level non-destructive internal inspections of containers, cargo boxes, and even vehicles of different volumes to detect the presence of heavy nuclear materials. It can also perform detailed imaging of the structural or displacement behavior of certain specialized nuclear materials during transportation or storage.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] The present invention provides a detection device based on muon imaging technology, comprising:

[0007] The system comprises a muon track acquisition module, a time-of-flight spectrometer, an electronic acquisition sub-device, and a data analysis system; the muon track acquisition module and the time-of-flight spectrometer are connected to the electronic acquisition sub-device, and the electronic acquisition sub-device is connected to the data analysis system.

[0008] The muon track acquisition module includes a muon incident track acquisition submodule and a muon exit track acquisition submodule; the muon incident track acquisition submodule and the muon exit track acquisition submodule are respectively at least three high position resolution multi-gap resistive plate chambers placed in parallel in a sealed cavity;

[0009] The time-of-flight spectrometer includes a muon incident time detection module and a muon exit time detection module, wherein the muon incident time detection module and the muon exit time detection module are respectively at least one high time resolution multi-gap resistive plate chamber in a sealed cavity;

[0010] The data analysis system includes a final dual-channel physical information fusion network. The first channel processes the spatial coordinates of muon tracks, and the second channel fuses the physical constraint relationship between momentum and scattering angle. The output layer contains a joint prediction of radiation length and atomic number.

[0011] In one embodiment, the detection device further includes an adjustable frame for fixing the muon incident time detection module, the muon incident track acquisition submodule, the muon exit track acquisition submodule, and the muon exit time detection module.

[0012] In one implementation, the muon incident time detection module and the muon incident track acquisition submodule are located above the object to be detected, and the muon exit track acquisition submodule and the muon exit time detection module are located below the object to be detected.

[0013] In one implementation, the muon incident time detection module, the muon incident track acquisition submodule, the muon exit time detection module, and the muon exit track acquisition submodule are arranged in parallel.

[0014] In one implementation, the muon incident time detection module is located above the muon incident track acquisition submodule, and the muon exit time detection module is located below the muon exit track acquisition submodule.

[0015] In one embodiment, the detection device further includes an intelligent central control system and at least one of the following devices connected to the intelligent central control system: a temperature monitoring device, a pressure monitoring device, a dark current monitoring device, a detector gas supply device, and a high-pressure control device; the intelligent central control system is used to receive signals from the temperature monitoring device, the pressure monitoring device, and the dark current monitoring device, and to send instructions to the detector gas supply device and the high-pressure control device.

[0016] In one embodiment, the position resolution of the high position resolution multi-gap resistive plate chamber is ≤0.5mm; the time resolution of the high time resolution multi-gap resistive plate chamber is ≤30ps.

[0017] This invention provides a method for acquiring heavy nuclei based on muon imaging technology, comprising:

[0018] Obtain the track information of the muon passing through the object to be measured;

[0019] Obtain the flight time of the muon as it passes through the object under test;

[0020] Based on the track information and the flight time, the multiple Coulomb scattering angle and momentum information of the muon are obtained; the track information, multiple Coulomb scattering angle and momentum information of the muon are input into the final dual-channel physical information fusion network (DCNN) to output the atomic number prediction value and confidence score;

[0021] When the predicted atomic number is greater than the predicted atomic number threshold and the confidence level is greater than the confidence level threshold, it is determined that the object under test contains heavy nucleus material.

[0022] In one implementation, the muon momentum information is obtained using formulas 2 and 3:

[0023]

[0024] Where p is the momentum of the muon, and m μ Let v be the rest mass of the muon, v be the average speed, and c be the speed of light.

[0026]

[0027] Where t is the muon's flight time and L is the muon's flight distance.

[0028] In one implementation, the muon's track information, multiple Coulomb scattering angle, and momentum information are input into a final dual-channel physical information fusion network. The output atomic number prediction and confidence score include: obtaining the radiation length of the object at different locations using the multiple Coulomb scattering calculation formula (Formula 1) combined with the dual-channel physical information fusion network; and based on the radiation length, obtaining the atomic number and confidence score at different locations to determine whether heavy nuclei exist in the object under test.

[0029] Construct an initial convolutional neural network; change the single input layer of the initial convolutional neural network to a dual-channel input layer containing spatial feature channels and physical feature channels, and inject the multiple Coulomb scattering calculation formula (Formula 1) to form a convolutional layer with injected physical formula; change the mean square error loss to physical regularization loss; add an uncertainty quantization module to the output layer to obtain an improved dual-channel physical information fusion network.

[0030] The intermediate feature channel is used to process the geometric information of muon tracks (incident / exit positions, trajectory vectors), using 3D convolutional layers to extract spatial patterns, and the spatial attention module focuses on high scattering regions; the physical feature channel is used to fuse the physical relationship between momentum and scattering angle, realizing the physical constraint of the multiple Coulomb scattering calculation formula (Formula 1); the physical regularization loss is used to calculate the error between the theoretical value and the predicted value, constraining the network output to conform to the muon scattering theoretical model; the uncertainty quantification module is used for the propagation calculation of momentum uncertainty and outputs the result with an added confidence score; the multiple Coulomb scattering calculation formula is as follows.

[0031]

[0032] Where p, v, and z are the momentum, velocity, and charge number of the incident muon, respectively, and L is the length of the muon path. rad Let σ be the radial length of the object. θ For multiple Coulomb scattering angles;

[0033] Collect historical labeled muon data; the historical labeled muon data includes muon track information, multiple Coulomb scattering angle and momentum information, as well as manually labeled atomic number predictions and confidence scores;

[0034] An improved dual-channel physical information fusion network is trained using historical labeled muon data and an optimization algorithm to obtain the final dual-channel physical information fusion network.

[0035] The track information, multiple Coulomb scattering angle, and momentum information of the muon are input into a dual-channel physical information fusion network, which outputs real-time atomic number predictions and confidence scores. Based on the atomic number predictions and confidence scores, it is determined whether there is heavy nucleus material in the object under test.

[0036] Beneficial effects

[0037] The above-mentioned technical solutions adopted in this specification can achieve the following beneficial effects:

[0038] The detection device based on muon imaging technology provided by this invention has a high degree of integration and is suitable for outdoor work requirements. It can be applied to customs and border cargo inspection, security, and nuclear energy and nuclear technology fields. It can perform minute-level non-destructive internal inspection of containers, cargo boxes, and even vehicles of different volumes to detect the presence of heavy nuclear materials. It can also perform fine imaging of the structural or displacement behavior of certain special nuclear materials during transportation or storage.

[0039] The detection device based on muon imaging technology provided by this invention has high practicality and robustness. Its modular design and adjustable frame allow it to flexibly adapt to objects of different volumes, optimizing detection efficiency and measurement accuracy. The sealed cavity design significantly reduces gas consumption of the detector, resulting in low cost and eliminating the need for additional detectors (such as scintillator detectors) as triggering devices. The integrated intelligent central control system enables real-time monitoring and automatic adjustment of key parameters such as temperature, air pressure, and dark current, and can trigger protective measures, significantly improving the long-term operational stability and reliability of the system and reducing maintenance requirements. Due to the utilization of momentum information, it exhibits good imaging performance, short imaging time, and strong robustness.

[0040] This invention provides a detection method based on muon imaging technology that utilizes muon momentum. By using the scattering angle and momentum information of muons, combined with a final dual-channel physical information fusion network, it effectively integrates the precisely measured track geometry information and momentum-scattering physical relationship, and introduces physical regularization loss and uncertainty quantization. It can analyze whether there are heavy core materials in the object under test in a short time, which is safer and saves time and effort compared to open-box inspection. Attached Figure Description

[0041] Figure 1 A schematic diagram of a detection device based on muon imaging technology according to an embodiment of this application is shown, wherein 1 represents the incident muon trajectory; 2 represents the exit muon trajectory.

[0042] Figure 2A -C shows the effect of different momentum accuracy on material recognition capability in Monte Carlo simulation;

[0043] Figure 3 The figure shows the ROC curve of the results of processing the muon scattering angle and momentum information obtained from Monte Carlo simulation using the machine learning method according to an embodiment of this application;

[0044] Figure 4 The graph shows the relationship between the accuracy of the machine learning method for identifying re-core materials according to an embodiment of this application and the muon exposure time.

[0045] Figure 5 The diagram shows the scattering angle distribution for integrity testing of heavy core materials after momentum screening obtained from Monte Carlo simulation according to an embodiment of this application.

[0046] Figure 6 The paper illustrates the imaging results of a detection method based on muon momentum using muon imaging technology according to an embodiment of this application, compared with conventional methods, for the Fetter model.

[0047] Figure 7The image shows the imaging results of a slight movement test of the Fetter model using a detection method based on muon momentum and muon imaging technology according to an embodiment of this application.

[0048] Figure 8 A schematic flowchart of a detection method based on muon imaging technology according to an embodiment of this application is shown. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.

[0050] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0051] In one embodiment of this application, a detection device based on muon imaging technology, such as Figure 1 As shown, the system includes a muon track acquisition module 1, a time-of-flight spectrometer 2, an electronic acquisition sub-device, and a data analysis system. The muon track acquisition module 1 is used to acquire track information of muons passing through the object under test, such as the flight distance L of the muons. The time-of-flight spectrometer 2 is used to acquire the flight time of the muons passing through the object under test. The muon track acquisition module 1 and the time-of-flight spectrometer 2 are connected to the electronic acquisition sub-device, which is connected to the data analysis system.

[0052] Muon track acquisition module 1 includes a muon incident track acquisition submodule 11 and a muon exit track acquisition submodule 12; the muon incident track acquisition submodule 11 and the muon exit track acquisition submodule 12 are respectively at least three high position resolution multi-gap resistive plate chamber MRPCs (High Position Resolution MRPCs) placed parallel to each other in a sealed cavity. 15(03)(2020)C03012.); High position resolution multi-gap resistive plate MRPC here specifically refers to MRPC with position resolution of 0.5mm, without regard to time resolution; The sealed cavity enables the high position resolution MRPC to work in an airless environment; Each high position resolution MRPC in the sealed cavity can acquire two-dimensional position information with a position resolution better than 0.5mm; The high position resolution MRPC has extremely high muon detection efficiency, and the detection efficiency of a single MRPC can reach 97%;

[0053] The time-of-flight spectrometer 2 includes a muon incident time detection module 21 and a muon exit time detection module 22. The muon incident time detection module 21 and the muon exit time detection module 22 are respectively located within a sealed cavity and are high time-resolution multi-gap resistive plate chamber MRPCs (High Time Resolution MRPC: Y.Yu, D.Han, Y.Wang, B.Guo, F.Wang, X.Chen, P.Lyu, C.Shen, Q.Zhang, Y.Li, Study of high time resolution mrpc with the waveform digitizer system, Journal of Instrumentation). 15(01)(2020)C01049.); High time resolution multi-gap resistive plate MRPC here specifically refers to MRPC with time resolution of up to 30 ps, ​​without regard to position resolution; Time-of-flight spectrometer 2 can obtain the flight time t of muons; Assuming the muons arrive at the muon incident time detection module 21 and the muon exit time detection module 22 at times t1 and t2 respectively, then the flight time of the muons t = t1 – t2; The high time resolution MRPC is enclosed in a sealed cavity, but works in an airflow environment; The time resolution of the high time resolution MRPC can be better than 30 ps, ​​which is at the top level; Due to the extremely high time resolution of the high time resolution MRPC used, the detection device of this application can realize continuous spectrum measurement of the momentum of low-energy muons (low-energy muons and high-energy muons are divided by 1.5 GeV / c), which can significantly improve the material identification capability and imaging accuracy of the muon imaging device and greatly shorten the imaging time;

[0054] The detection device based on muon imaging technology of the present invention does not require triggering the detector. It uses the signal of the time-of-flight spectrometer as the external trigger signal or the signal of the muon track acquisition module as the self-trigger signal. When the trigger signal is effectively acquired, a muon event is recorded, which greatly reduces the cost.

[0055] In one embodiment of this application, the detection device based on muon imaging technology further includes an adjustable frame for fixing the muon incident time detection module 21, the muon incident track acquisition submodule 11, the muon exit track acquisition submodule 12, and the muon exit time detection module 22. In one embodiment of this application, the muon incident time detection module 21 and the muon incident track acquisition submodule 11 are located above the object to be detected, and the muon exit track acquisition submodule 12 and the muon exit time detection module 22 are located below the object to be detected. The adjustable frame can not only adapt to objects of different volumes, but also further change the interval between the muon incident time detection module 21 and the muon exit time detection module 22. In specific situations, the distance between the muon track acquisition modules can be shortened to achieve higher geometric detection efficiency for small-volume objects, or the distance between the muon track acquisition modules can be increased to accommodate larger materials. Furthermore, the flight distance of muons can be further extended by increasing the distance between the muon incident time detection module 21 and the muon exit time detection module 22, thereby improving the accuracy of momentum measurement.

[0056] Because muons have small rest mass and high energy, they can be approximated as meeting the conditions for the smallest ionized particle. According to statistical results, for muons with momentum less than 1000 GeV / c, the energy required to ionize a particle is approximately 1 g·cm⁻¹. -2 The energy loss of the substance is about 2.2 MeV, which is very small for muons (energy on the order of GeV);

[0057] For ease of system construction and to maximize the geometric reception of cosmic ray muons, the muon incident time detection module 21 and the muon incident track acquisition submodule 11 can be set in parallel. Of course, tilting them can also achieve the basic function, but it will reduce the number of muons that can be measured per unit time. Because the smallest ionized particle has strong penetrating power, muons will not be absorbed after passing through matter. At the same time, since the flight time needs to be measured and the momentum needs to be calculated using the average velocity, the small energy loss of muons means that their velocity changes little throughout the device, which means that it is reasonable to use the average velocity to calculate momentum. There is no direct relationship between the smallest ionized particle and the position of the muon incident time detection module 21 and the muon incident track acquisition submodule 11.

[0058] The incident time detection module 21 is located above the muon incident track acquisition submodule 11. Because the system adopts a self-triggered working mode, the incident time detection module 21 has both time measurement and triggering functions. Similarly, the muon exit time detection module 22 is located below the muon exit track acquisition submodule 12.

[0059] Muns are incident on the mun incident trajectory acquisition submodule 11, which captures the initial incident position, timestamp, and incident direction of the muns. The muns are emitted from the mun exit trajectory acquisition submodule 12, which captures the exit position, timestamp, and exit direction of the muns. The mun scattering angle can be calculated by calculating the angle between the incident direction and the exit direction.

[0060] In one embodiment of this application, the detection device based on muon imaging technology further includes an intelligent central control system and at least one of the following devices connected to the intelligent central control system: a temperature monitoring device, a pressure monitoring device, a dark current monitoring device, a detector gas supply device, and a high-pressure control device; the intelligent central control system is used to receive signals from the temperature monitoring device, the pressure monitoring device, and the dark current monitoring device, and to send instructions to the detector gas supply device and the high-pressure control device; the intelligent central control system can connect to the temperature monitoring system, the pressure monitoring system, the dark current monitoring device, the detector gas supply device, and the high-pressure control device through standard communication protocols (such as Modbus, CAN bus, etc.) or customized interfaces respectively, to realize data acquisition and instruction issuance, thereby dynamically acquiring parameters such as temperature, pressure, and dark current, triggering abnormal alarms (such as excessively high temperature, abnormal pressure fluctuations, and dark current changes) through threshold settings, and linking with the high-pressure control system to realize overload protection, or adjusting the gas flow through the detector gas supply system to maintain stable equipment operation; the time jitter of the electronic system is affected by temperature, and the time resolution capability of the electronics (the time of the entire system) needs to be evaluated by temperature. Inter-polar resolution is influenced by both the electronic system and the MRPC. The atmospheric pressure monitoring device ensures a normal and stable working gas environment for the MRPC, thereby adjusting the gas flow rate. The dark current monitoring device monitors the dark current, which reflects whether the MRPC is working properly. During normal operation, the dark current of the MRPC is generally around 70nA. If the dark current increases, it indicates an abnormality, requiring immediate voltage adjustment to protect the detection device. If the gas environment is abnormal, for example, all experimental conditions are normal but the dark current continues to increase, it likely indicates a problem with the gas environment. This could be due to excessive exhaust gas affecting the normal operation of the detection device, or an abnormal working gas ratio causing the detection device to deviate from its operating point. In these cases, the gas supply device needs to be reconfigured to change the gas environment. A high-voltage control device is also necessary; when the detection device operates at 6700V, a power supply is required. If the muon imaging system is installed at a relatively low altitude, an atmospheric pressure monitoring system is not required. However, if the installation altitude is high, an atmospheric pressure monitoring system should be fully considered to ensure the MRPC operates at its optimal state.

[0061] In one embodiment of this application, the electronic acquisition sub-device receives electrical signals from high temporal resolution MRPC and high positional resolution MRPC and determines whether it is an incident event; the data analysis system can record valid incident events and use the PoCA algorithm for imaging or machine learning methods for analysis, thereby determining whether there is heavy core material in the object under test; in one embodiment of this application, the machine learning method is an improvement of the convolutional neural network;

[0062] The specific implementation process is as follows: Construct an initial convolutional neural network; change the single input layer of the initial convolutional neural network to a dual-channel input layer containing spatial feature channels and physical feature channels, and inject the multiple Coulomb scattering calculation formula (Formula 1) to form a convolutional layer with injected physical formula; change the mean square error loss to physical regularization loss; add an uncertainty quantization module to the output layer to obtain an improved dual-channel physical information fusion network.

[0063] The intermediate feature channel is used to process the geometric information of muon tracks (incident / exit positions, trajectory vectors), and a 3D convolutional layer is used to extract spatial patterns. The spatial attention module focuses on high scattering regions. The physical feature channel is used to fuse the physical relationship between momentum and scattering angle to achieve physical constraints on the multiple Coulomb scattering calculation formula (Formula 1). The physical regularization loss is used to calculate the error between the theoretical value and the predicted value, and to constrain the network output to conform to the muon scattering theoretical model. The uncertainty quantification module is used to calculate the propagation of momentum uncertainty and outputs the results with an added confidence score.

[0064]

[0065] Where p, v, and z are the momentum, velocity, and charge number of the incident muon, respectively, and L is the length of the muon path. rad Let σ be the radial length of the object. θ For multiple Coulomb scattering angles;

[0066] Collect historical labeled muon data; the historical labeled muon data includes muon track information, multiple Coulomb scattering angle and momentum information, as well as manually labeled atomic number predictions and confidence scores;

[0067] An improved dual-channel physical information fusion network is trained using historical labeled muon data and an optimization algorithm to obtain the final dual-channel physical information fusion network.

[0068] The process of training an improved dual-channel physical information fusion network using historical labeled muon data and an optimization algorithm to obtain the final dual-channel physical information fusion network includes the following steps:

[0069] An improved dual-channel physical information fusion network is trained using historical labeled muon data. During the training process, an optimization algorithm is used to find the optimal weights of the improved dual-channel physical information fusion network to obtain the optimal solution. The optimal solution is then used as the weights of the improved dual-channel physical information fusion network to obtain the final dual-channel physical information fusion network.

[0070] The optimization algorithms include various types, such as ant colony optimization, genetic algorithm, and particle swarm optimization. Taking the genetic algorithm as an example, the weights of the improved dual-channel physical information fusion network are used as chromosomes in the genetic algorithm. Through continuous iteration, crossover, mutation, and selection, the optimal solution is finally obtained.

[0071] The track information, multiple Coulomb scattering angle, and momentum information of muons are input into a dual-channel physical information fusion network, which outputs real-time atomic number predictions and confidence scores. Based on the atomic number predictions and confidence scores, it is determined whether there are heavy nuclei in the object under test.

[0072] The detection device based on muon imaging technology in this application adopts a modular design, which not only improves the overall robustness of the detection device, making the whole device more suitable for industrial applications, but also reduces the gas consumption of the detection device, saves resources, and is more environmentally friendly.

[0073] The above describes the detection device based on muon imaging technology in this specification. Based on the same approach, this specification also provides corresponding detection methods based on muon imaging technology, such as... Figure 8 As shown; the detection method based on muon imaging technology in this embodiment of the invention may include:

[0074] S101: Obtain the track information of the muon passing through the object to be tested;

[0075] S102: Obtain the flight time of the muon as it passes through the object under test;

[0076] S103: Based on the track information and the flight time, obtain the multiple Coulomb scattering angle and momentum information of the muon; use the multiple Coulomb scattering angle and momentum information of the muon to determine whether there is heavy core material in the object under test;

[0077] Assuming the arrival times of the muon at the muon incident time detection module 21 and the exit time detection module 22 are t1 and t2 respectively, then the flight time of the muon measured by the time-of-flight spectrometer is t = t1 – t2, and the flight distance obtained by the muon track acquisition module is L. Considering relativistic effects:

[0078] First, obtain the average velocity of the muon:

[0079]

[0080] Given the speed of light as c, we can obtain the Lorentz factor:

[0081]

[0082] The momentum p of the muon under relativity can be expressed as:

[0083]

[0084] Where m μ The rest mass of the muon is 105.6 MeV / c. 2 ;

[0085] In one embodiment of this application, muon momentum information can be obtained using formulas 2 and 3;

[0086] Molliere's theory provides a good description of multiple Coulomb scattering processes: the projected components of multiple Coulomb scattering onto the two planes are independent and identically distributed, and approximately follow a Gaussian distribution. The standard deviation of this distribution is related to the radiative length of the matter, and can be fitted using the Molliere distribution.

[0087]

[0088] Where p, v, and z are the momentum, velocity, and charge number of the incident muon, respectively, and L is the muon path length. rad Let σ be the radial length of the object. θ For multiple Coulomb scattering angles;

[0089] In one embodiment of this application, the radiation length of an object at different positions is obtained using Formula 1; the atomic number at different positions is obtained based on the radiation length, thereby determining whether heavy nuclei are present in the object under test.

[0090] In imaging using the PoCA algorithm, the scattering intensity is calculated through the following steps:

[0091] It is known that the projected components of the spatial scattering angle of muon multiple Coulomb scattering in two planes are independent and identically distributed, and approximately follow a Gaussian distribution. The standard deviation of this distribution is related to the radiative length of the matter:

[0092]

[0093] Using the above formula, since β≈1 and z=1 for muons, it can be simplified to:

[0094]

[0095] Based on this, we can further define the unit path scattering intensity λ of muons in matter:

[0096]

[0097] Scattering intensity is essentially a quantity that reflects the distribution of scattering angles; since it is related to the radiation length of a substance, calculating the value of scattering intensity at different regions (different pixels) in a substance can be used for imaging.

[0098] Assuming the muon is monoenergetic, the incident path is the same, and the momentum is constant at p0, then the relationship between the unit path scattering intensity λ0 of a substance for a monoenergetic muon and the standard deviation σ0 of the scattering angle of the monoenergetic muon can be determined:

[0099]

[0100] However, the actual momentum of muons is continuous, and the incident paths also vary. Therefore, the calculated scattering intensity of muons with different momentum incident on the same matter will be different. To eliminate this difference, the scattering intensity needs to be weighted so that it can directly reflect the properties of the matter (L). rad It only depends on the atomic number of the substance, that is:

[0101]

[0102] Because the spatial scattering angle of the muon multiple Coulomb scattering is projected onto two planes, the XZ plane and the YZ plane (assuming the Z direction is vertical, and X and Y are two horizontal directions), the component θ x and θ y All are independent and identically distributed, and approximately follow a Gaussian distribution with a mean of 0 and standard deviations of [missing information]. and Assuming N muons are incident, the projection scattering angle of the i-th muon on the XZ plane is: Similarly, its projected scattering angle on the YZ plane is We can obtain:

[0103]

[0104] Therefore, when the scattering angle θ of the two projection planes is obtained... x With θ y At that time, since the two are independently and identically distributed, Right now:

[0105]

[0106] For each incident momentum p i Each incident path is L i For the muon, considering the effects of incident momentum and incident path on its scattering angle, the weighted scattering intensity can be obtained. satisfy:

[0107]

[0108] At this point, p0 represents a constant momentum, and L represents a constant path; both are used for reference and can actually be omitted when calculating the scattering intensity, i.e., p0 and L are considered as 1, resulting in:

[0109]

[0110] Furthermore, consider the momentum measurement uncertainty of the muon. hour:

[0111]

[0112] Detection devices based on muon imaging technology can measure momentum relatively accurately, and thus assess the uncertainty of momentum measurement.

[0113] The detection method based on muon imaging technology proposed in this application significantly improves the detection performance of different substances; Figure 2 shows the effect of different momentum accuracies on material recognition performance in Monte Carlo simulation; Figures 2A to 2C The statistical results of the scattering intensity of muons for different materials with a thickness of 10 cm and an exposure time of 1 min are obtained under the conditions of not using momentum information, using precise momentum information (the precise incident momentum value (true value) of muons obtained in Monte Carlo simulation), and using the momentum information in the specific embodiment of the present invention. It can be clearly seen that the detection device of the present invention can greatly improve the performance of muon imaging and detection device.

[0114] Figure 3 This application presents a specific embodiment of a machine learning method processing Monte Carlo simulation to obtain the muon scattering angle and momentum information results, which yields the ROC (Receiver Operating Characteristic) curve. The ROC curve is a tool for evaluating the performance of binary classification models, particularly suitable for measuring model performance at different classification thresholds. The horizontal axis represents the true positive rate, abbreviated as TPR, which satisfies the following:

[0115]

[0116] Where FN (False Negative) is the number of false negatives in the classification results obtained by the classifier, and TP (True Positive) is the number of true positives; FN+TP is actually the sum of all positive cases; TPR represents the sensitivity of the classifier to positive cases; the vertical axis is the true negative rate, abbreviated as TNR, which satisfies:

[0117]

[0118] In this context, FP (False Positive) represents the number of false positives in the classification results obtained by the classifier, and TN (True Negative) represents the number of true negatives. FP+TN is actually the sum of all negative cases. TNR represents the classifier's sensitivity to negative cases. Therefore, the closer the ROC curve is to the upper right corner, the better the classifier's ability to distinguish between positive and negative cases. Some ROC curves show better performance the closer they are to the upper left corner, which is actually due to some symmetrical operations, but the principle is the same.

[0119] like Figure 3 As shown, three different other machine learning methods were used: augmented decision tree (BDT), convolutional neural network (CNN), and deep neural network (DNN), as well as the final dual-channel physical information fusion network (DCNN) in this application. These methods were used to determine whether heavy core materials were present in a cargo box filled with contents (maximum density) using incident data of 1 minute muons. The ROC curves obtained by the three machine learning methods were compared with those obtained by DCNN, and the performance advantage of DCNN can be seen objectively.

[0120] Figure 4 This is a graph showing the relationship between the accuracy of machine learning methods for identifying heavy core materials and muon exposure time in a specific embodiment of this application. It can be seen that with only 30s to 45s of muon exposure time, the detection device based on muon imaging technology of this application can identify the presence of heavy core materials in a fully loaded (maximum density) cargo box. The accuracy of various machine learning methods all exceed 90%, and the accuracy of DCNN significantly surpasses the other three machine learning methods. This demonstrates the reliability of the detection device based on muon imaging technology of this application and the advantages of the final dual-channel physical information fusion network.

[0121] A specific embodiment of the detection device based on muon imaging technology in this application also exhibits excellent performance in monitoring the internal structure of nuclear materials; such as... Figure 5 As shown, for two nuclear material models, namely, a material with internal vacancies (red) and a homogeneous material (blue), after several hours of muon exposure, by screening muon incident events with momentum below a certain threshold, it can be found that the distribution of muon scattering angles is significantly different, which confirms that the detection device based on muon imaging technology in a specific embodiment of this application has the function of monitoring and imaging the internal structure of nuclear materials.

[0122] Considering the sensitivity and confidentiality of the geometry of dedicated nuclear devices, in one specific embodiment of this application, an imaging simulation was performed on the Fetter model—a hypothetical nuclear device publicly published by Professor Steve Fetter, an international expert in nuclear arms control and nonproliferation. The simulation was conducted using a detection device based on muon imaging technology, as described in one specific embodiment of this application. Figure 6 (left image) and traditional momentum-free imaging methods ( Figure 6 The right figure shows a comparison of the imaging results obtained under these two conditions; it can be seen that the detection device based on muon imaging technology in a specific embodiment of this application can greatly improve the performance of muon imaging, and the beryllium reflection layer inside the Fetter model can be clearly seen; at the same time, Monte Carlo simulations were also performed on the displacement scenario of the Fetter model during transportation or long-term storage, and the scanning imaging results of muon quantity for 5 minutes are as follows. Figure 7 As shown ( Figure 7 The left image shows the initial position image. Figure 7 The right image shows the image after movement. The detection device based on muon imaging technology according to a specific embodiment of this application shows that even a movement of only 1 cm is clearly displayed in the image.

[0123] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple. The relevant parts can be referred to the description of the method embodiments.

[0124] The above description is merely an embodiment of this specification and is not intended to limit this specification. For those skilled in the art, this specification can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification should be included within the scope of the claims of this specification.

Claims

1. A detection device based on muon imaging technology, characterized in that, include: The system comprises a muon track acquisition module, a time-of-flight spectrometer, an electronic acquisition sub-device, and a data analysis system; the muon track acquisition module and the time-of-flight spectrometer are connected to the electronic acquisition sub-device, and the electronic acquisition sub-device is connected to the data analysis system. The muon track acquisition module includes a muon incident track acquisition submodule and a muon exit track acquisition submodule; the muon incident track acquisition submodule and the muon exit track acquisition submodule are respectively at least three high position resolution multi-gap resistive plate chambers placed in parallel in a sealed cavity; The time-of-flight spectrometer includes a muon incident time detection module and a muon exit time detection module, wherein the muon incident time detection module and the muon exit time detection module are respectively at least one high time resolution multi-gap resistive plate chamber in a sealed cavity; The data analysis system includes a final dual-channel physical information fusion network. The first channel processes the spatial coordinates of muon tracks, and the second channel fuses the physical constraint relationship between momentum and scattering angle. The output layer contains a joint prediction of radiation length and atomic number.

2. The detection device according to claim 1, characterized in that, The detection device also includes an adjustable frame for fixing the muon incident time detection module, the muon incident track acquisition submodule, the muon exit track acquisition submodule, and the muon exit time detection module.

3. The detection device according to claim 2, characterized in that, The muon incident time detection module and the muon incident trajectory acquisition submodule are located above the object to be detected, while the muon exit trajectory acquisition submodule and the muon exit time detection module are located below the object to be detected.

4. The detection device according to claim 3, characterized in that, The muon incident time detection module, the muon incident track acquisition submodule, the muon exit time detection module, and the muon exit track acquisition submodule are arranged in parallel.

5. The detection device according to claim 3, characterized in that, The muon incident time detection module is located above the muon incident track acquisition submodule, and the muon exit time detection module is located below the muon exit track acquisition submodule.

6. The detection device according to claim 1, characterized in that, The detection device also includes an intelligent central control system and at least one of the following devices connected to the intelligent central control system: a temperature monitoring device, an air pressure monitoring device, a dark current monitoring device, a detector air supply device, and a high-pressure control device. The intelligent central control system is used to receive signals from the temperature monitoring device, the air pressure monitoring device, and the dark current monitoring device, and to send instructions to the detector air supply device and the high-pressure control device.

7. The detection device according to claim 1, characterized in that, The position resolution of the high position resolution multi-gap resistive plate chamber is ≤0.5mm; the time resolution of the high time resolution multi-gap resistive plate chamber is ≤30ps.

8. A method for detecting heavy nuclei based on muon imaging technology, characterized in that, include: Obtain the track information of the muon passing through the object to be measured; Obtain the flight time of the muon as it passes through the object under test; Based on the track information and the flight time, the multiple Coulomb scattering angle and momentum information of the muon are obtained; the track information, multiple Coulomb scattering angle and momentum information of the muon are input into the final dual-channel physical information fusion network, and the predicted atomic number and confidence score are output. When the predicted atomic number is greater than the predicted atomic number threshold and the confidence level is greater than the confidence level threshold, it is determined that the object under test contains heavy nucleus material.

9. The method according to claim 8, characterized in that, The momentum information of the muon is obtained using track information and flight time.

10. The method according to claim 8, characterized in that, The track information, multiple Coulomb scattering angle, and momentum information of the muon are input into the final dual-channel physical information fusion network. The output atomic number prediction and confidence score include: obtaining the radiation length of the object at different locations using the multiple Coulomb scattering calculation formula combined with the dual-channel physical information fusion network; and obtaining the atomic number and confidence score at different locations based on the radiation length, thereby determining whether there is heavy nucleus material in the object under test. Construct an initial convolutional neural network; change the single input layer of the initial convolutional neural network to a dual-channel input layer containing spatial feature channels and physical feature channels, and inject the multiple Coulomb scattering calculation formula to form a convolutional layer with injected physical formula; change the mean square error loss to physical regularization loss; add an uncertainty quantization module to the output layer to obtain an improved dual-channel physical information fusion network. The spatial feature channel is used to process the geometric information of muon tracks, using 3D convolutional layers to extract spatial patterns, and the spatial attention module focuses on high scattering regions; the physical feature channel is used to fuse the physical relationship between momentum and scattering angle, realizing the physical constraint of the multiple Coulomb scattering calculation formula; the physical regularization loss is used to calculate the error between theoretical and predicted values, constraining the network output to conform to the muon scattering theoretical model; the uncertainty quantification module is used to calculate the propagation of momentum uncertainty and outputs the results with an added confidence score; Collect historical labeled muon data; the historical labeled muon data includes track information, multiple Coulomb scattering angle and momentum information of historical muons, as well as manually labeled atomic number predictions and confidence scores; An improved dual-channel physical information fusion network is trained using historical labeled muon data and an optimization algorithm to obtain the final dual-channel physical information fusion network. Real-time muon track information, multiple Coulomb scattering angles, and momentum information are input into a dual-channel physical information fusion network, which outputs real-time atomic number predictions and confidence scores. Based on the atomic number predictions and confidence scores, it is determined whether heavy nuclei exist in the object under test.

Images