Method and device for rapidly detecting heavy nuclear materials by using cosmic ray muons

Abstract

The present invention relates to a method and device for rapidly detecting heavy nuclear materials using cosmic ray muons, belonging to the field of nuclear detection technology, and solves the problems of long time consumption and low efficiency in detecting and distinguishing heavy nuclear materials in the prior art. The method includes the following steps: based on the muon Coulomb scattering angle data sets of multiple target materials, respectively analyze and obtain the muon detection data feature sets of multiple target materials; classify and train the muon detection data feature sets of multiple target materials to obtain a classification model; use the classification model to detect the material to be detected to determine whether there is heavy nuclear material in the material to be detected. This method can rapidly detect whether there is heavy nuclear material in space in different heavy nuclear material detection scenarios.

Description

Technical Field

[0001] The present invention relates to the field of nuclear detection technology, and in particular to a method and device for rapidly detecting heavy nuclear materials using cosmic ray muons. Background Art

[0002] Cosmic ray Muons have the characteristics of strong penetration and no additional radiation, and have significant Coulomb scattering characteristics for heavy nuclear materials. The principle is that when cosmic ray Muons pass through the material, the distribution of their multiple Coulomb scattering angles obeys a Gaussian distribution with a mean of zero, and its distribution variance is related to the atomic number of the material. Compared with existing heavy nuclear material detection technology, this method can easily penetrate dense materials, has no radiation hazard to detection personnel and materials to be tested, and is an ideal heavy nuclear material detection technology. In recent years, a method of using Muon Coulomb scattering angle root mean square grayscale correlation analysis to detect heavy nuclear materials has been proposed. This method does not image the object to be tested to reduce the amount of detection data required to identify heavy nuclear materials, but it still takes a long time to accumulate Muon events to accurately distinguish heavy nuclear materials from detection interferences, resulting in a long time, slow detection speed, and low efficiency. Summary of the invention

[0003] In view of the above analysis, the embodiments of the present invention aim to provide a method and device for quickly detecting heavy nuclear materials using cosmic ray muons, so as to solve the problem that the methods in the prior art for detecting and distinguishing whether heavy nuclear materials exist in the material to be detected are time-consuming, slow in detection speed and low in efficiency.

[0004] In one aspect, the present invention provides a method for rapidly detecting heavy nuclear materials using cosmic ray muons, comprising the following steps:

[0005] Based on the Muon sub-Coulomb scattering angle data sets of various target materials, the Muon sub-detection data feature sets of various target materials are analyzed and obtained respectively;

[0006] Performing classification training on the Muon sub-detection data feature sets of the target materials to obtain a classification model;

[0007] The classification model is used to detect the material to be detected to determine whether heavy nuclear substances exist in the material to be detected.

[0008] Furthermore, it also includes building a detection system, including:

[0009] Construct a detection space of preset volume;

[0010] At preset distances on the upper and lower sides inside the detection space, a first set of detection devices and a second set of detection devices are horizontally placed respectively; both the first set of detection devices and the second set of detection devices include 3 Muon sub-detection sensors, and the 3 Muon sub-detection sensors are longitudinally and horizontally placed at preset intervals.

[0011] Further, the steps of respectively analyzing and obtaining the Muon sub-detection data feature sets of various target materials based on the Muon sub-Coulomb scattering angle data sets of various target materials include:

[0012] When no target material is placed at the central position in the detection space, detection is carried out, and the Muon sub-Coulomb scattering angle is calculated using the Monte Carlo method, and then the Muon sub-Coulomb scattering angle data set of air is obtained;

[0013] Each target material is sequentially placed at the central position, detection is carried out respectively, and the corresponding Muon sub-Coulomb scattering angle is calculated using the Monte Carlo method respectively, and then the Muon sub-Coulomb scattering angle data set of each target material is obtained;

[0014] Based on the Muon sub-Coulomb scattering angle data set of air and the Muon sub-Coulomb scattering angle data sets of each target material, the Muon sub-detection data set of each target material is obtained;

[0015] Based on the Muon sub-detection data set of each target material, the Muon sub-detection data feature set of each target material is calculated. The Muon sub-detection data feature set of each target material includes the root mean square of the Coulomb scattering angle distribution corresponding to the target material, the proportion of Coulomb scattering angle data, and the kurtosis.

[0016] Further, the obtaining of the Muon sub-detection data set of each target material includes:

[0017] Random sampling step: Randomly sample the first preset number of Muon sub-Coulomb scattering angles from the Muon sub-Coulomb scattering angle data set of each target material; randomly sample the second preset number of Muon sub-Coulomb scattering angles from the Muon sub-Coulomb scattering angle data set of air; under the condition that the first preset number and the second preset number remain unchanged, the random sampling process is executed multiple times to obtain the first detection data group;

[0018] Adjust the ratio of the first preset number to the second preset number multiple times within a preset range according to a preset step length. According to the new ratio obtained each time, repeat the random sampling step to obtain multiple detection data groups corresponding to different ratios;

[0019] The first detection data set and the multiple detection data sets corresponding to different ratios constitute the Muon detection data set of each target material.

[0020] Further, the step of calculating the Muon detection data feature set of each target material based on the Muon detection data set of each target material includes calculating the detection data features corresponding to each detection data set in the Muon detection data set of each target material, specifically including:

[0021] Using the expectation-maximization algorithm to estimate the root mean square of the Coulomb scattering angle distribution corresponding to each detection data set:

[0022]

[0023] where n represents the number of iterations. When the update amplitude is less than the preset value, the corresponding is the root mean square of the Coulomb scattering angle distribution corresponding to the detection data set, and x i represents the i-th Coulomb scattering angle in the detection data set, i = 1, 2, 3... N, and σ air is the root mean square of the air Coulomb scattering angle distribution corresponding to the detection data set, represents the root mean square of the Coulomb scattering angle distribution corresponding to the detection data set obtained in the n-th iteration;

[0024] Using the expectation-maximization algorithm to estimate the proportion of Coulomb scattering angle data corresponding to each detection data set:

[0025]

[0026] where ρ EM represents the proportion of Coulomb scattering angle data corresponding to the detection data set;

[0027] Calculating the kurtosis corresponding to each detection data set:

[0028]

[0029] where K represents the kurtosis corresponding to the detection data set.

[0030] Further, the step of training the Muon detection data feature sets of multiple target materials to obtain a classification model includes:

[0031] Based on the training set composed of the Muon detection data feature sets of each target material, using the support vector machine method to perform binary classification training on each target material to obtain the classification model function when each target material is the positive class;

[0032] The classification model function when each target material is the positive class is:

[0033] f(d) = ∑ j a j y j k(d, d j ) + b,

[0034]

[0035] d = (σ EM , ρ EM , K),

[0036] where d j is the j-th support vector obtained by training the support vector machine, j = 1, 2, 3...M, a j represents the parameter corresponding to the j-th support vector, b represents the parameter obtained by training the support vector machine, y j is the classification label value corresponding to d j , d is the Muon detection data feature of the target material; σ EM is the root mean square of the Coulomb scattering angle distribution of the target material, ρ EM represents the proportion of the Coulomb scattering angle data of the target material, and K represents the kurtosis of the Muon detection data set of the target material.

[0037] Furthermore, the step of using the classification model to detect the material to be detected to determine whether there is heavy nuclear material in the material to be detected includes:

[0038] Detecting to obtain the Muon detection data set of the material to be detected, and analyzing to obtain the Muon detection data set features of the material to be detected;

[0039] Substituting the Muon detection data set features of the material to be detected into the classification model function corresponding to each target material to obtain the corresponding similarity value, and taking the target material corresponding to the maximum similarity value as the classification result of the material to be detected, and determining whether there is heavy nuclear material in the material to be detected according to the classification result.

[0040] Furthermore, the Muon detection sensor includes a first detection plate and a second detection plate that are coupled to each other. Both the first detection plate and the second detection plate include a plurality of strip-shaped scintillators arranged in parallel, and the arrangement direction of the strip-shaped scintillators in the first detection plate is perpendicular to the arrangement direction of the strip-shaped scintillators in the second detection plate.

[0041] On the other hand, the present invention provides a device for quickly detecting heavy nuclear materials using cosmic ray Muons, including:

[0042] A data processing module analyzes and obtains the characteristic sets of Muon sub-detection data for various target materials based on the Muon sub-Coulomb scattering angle data sets of various target materials.

[0043] A model establishment module is used to perform classification training on the characteristic sets of Muon sub-detection data for various target materials to obtain a classification model.

[0044] A detection module is used to detect the material to be detected by using the classification model to determine whether there is heavy nuclear material in the material to be detected.

[0045] Furthermore, it further includes a detection system.

[0046] The detection system is used to construct a detection space with a preset volume.

[0047] A first group of detection devices and a second group of detection devices are horizontally placed at preset distances on the upper side and the lower side inside the detection space respectively; both the first group of detection devices and the second group of detection devices include 3 Muon sub-detection sensors, and the 3 Muon sub-detection sensors are longitudinally and horizontally placed at preset intervals.

[0048] Compared with the prior art, the present invention can at least achieve one of the following beneficial effects:

[0049] 1. The method and device for quickly detecting heavy nuclear materials by using cosmic ray Muons proposed by the present invention obtain a classification model through model training based on the detection data characteristics of target materials, and discriminate the material to be detected according to the classification model, greatly reducing the amount of Muon sub-detection data required for detection. It only needs to detect Muon sub-events (Muon sub-Coulomb scattering angle) for 1 minute at least, and can accurately identify whether there is heavy nuclear material in the material to be detected. It is applicable to various security inspection scenarios with large material flow, facilitating the quick detection of whether there is heavy nuclear material, thereby being able to greatly improve the detection rate.

[0050] 2. The method and device for quickly detecting heavy nuclear materials by using cosmic ray Muons proposed by the present invention use the support vector machine method to perform binary classification training on the characteristic sets of detection data of target materials to obtain a classification model, which can effectively utilize the Muon sub-Coulomb scattering angle data and eliminate the influence of air noise, thereby being able to improve the classification accuracy of the obtained classification model.

[0051] 3. The present invention constructs a dataset of Coulomb scattering angles for different materials through the Monte Carlo method, and quickly obtains a Muon detection dataset for different materials by means of random sampling. The detection dataset contains the characteristics corresponding to various sizes of the same material. According to the Muon detection datasets of different materials, classification models of different materials can be quickly obtained in different detection environments without the need to detect and collect Muon data for a long time, which is convenient for quickly establishing a heavy nuclear material detection system.

[0052] In the present invention, the above technical solutions can also be combined with each other to achieve more preferred combination schemes. Other features and advantages of the present invention will be described in the subsequent specification, and some advantages can be made obvious from the specification, or understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the content specifically pointed out in the specification and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings are only for the purpose of showing specific embodiments and are not considered to be a limitation of the present invention. Throughout the drawings, the same reference signs denote the same components.

[0054] Figure 1 It is a flowchart of the method for quickly detecting heavy nuclear materials by using cosmic ray Muons in an embodiment of the present invention;

[0055] Figure 2 It is a schematic diagram of the detection system in an embodiment of the present invention;

[0056] Figure 3 It is a schematic diagram of the size of the detection device inside the detection system and the target material when the ratio of the first preset number to the second preset number is 0.01 in an embodiment of the present invention;

[0057] Figure 4 It is a schematic diagram of the size of the detection device inside the detection system and the target material when the ratio of the first preset number to the second preset number is 0.25 in an embodiment of the present invention;

[0058] Figure 5 It is a schematic diagram of the characteristic distributions of three target materials (uranium, lead, iron) in an embodiment of the present invention;

[0059] Figure 6 It is a schematic diagram of the device for quickly detecting heavy nuclear materials by using cosmic ray Muons in an embodiment of the present invention.

[0060] Reference Signs:

[0061] 110 - Detection System; 120 - Data Processing Module; 130 - Model Establishment Module; 140 - Detection Module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] The preferred embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings. The accompanying drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, rather than to limit the scope of the present invention.

[0063] A specific embodiment of the present invention discloses a method for rapidly detecting heavy nuclear materials using cosmic ray Muons. As Figure 1 shown, the method includes the following steps:

[0064] S100. Based on the Muon Coulomb scattering angle data sets of multiple target materials, respectively analyze and obtain the Muon detection data feature sets of multiple target materials.

[0065] S200. Classify and train the Muon detection data feature sets of multiple target materials to obtain a classification model.

[0066] S300. Use the classification model to detect the material to be detected to determine whether there is heavy nuclear material in the material to be detected.

[0067] Specifically, the target materials include uranium (U), lead (Pb), and iron (Fe). Among them, uranium is a heavy nuclear material, lead is a material similar to heavy nuclear materials, and iron is a common material. By calculating the detection data characteristics of these three target materials and establishing a classification model, the material to be detected can be accurately identified and determined.

[0068] Preferably, it further includes constructing a detection system. As Figure 2 shown, it specifically includes:

[0069] Construct a detection space with a preset volume. Specifically, a detection space with a corresponding volume size can be adaptively established according to the actual application scenario. Exemplarily, in the subway or airport security inspection environment, a detection space of 1m×1m×1m can be constructed.

[0070] Exemplarily, the target material or the material to be detected can be placed at the central position of the detection space. Preferably, during the process of training the classification model of the target material, the thickness of the target material is set to 10 cm.

[0071] At preset distances on the upper and lower sides inside the detection space, a first group of detection devices and a second group of detection devices are horizontally placed respectively. Among them, the preset distance can be set according to the actual situation. Specifically, both the first group of detection devices and the second group of detection devices include 3 Muon detection sensors, and the 3 Muon detection sensors are all horizontally placed longitudinally at a preset interval. Exemplarily, when detecting the target material (uranium (U), lead (Pb), iron (Fe)), the preset interval is set to 10 cm.

[0072] Preferably, in order to improve the efficiency of establishing a classification model, during the process of establishing the classification model, a Muon sub-emission source with a vertically downward direction can be set at the central position above the first detection device. Exemplarily, the energy of the set Muon sub-emission source is 3 GeV. Preferably, during the actual application process, natural cosmic ray Muons are used to detect the material to be detected, and based on the detection data, the classification model is used to determine whether there is heavy nuclear material in the material to be detected.

[0073] Preferably, the Muon detection sensor includes a first detection plate and a second detection plate that are coupled to each other. Preferably, its size is 100 cm × 100 cm × 5 cm. Among them, both the first detection plate and the second detection plate include a plurality of strip-shaped scintillators arranged in parallel, and the arrangement direction of the strip-shaped scintillators in the first detection plate and the arrangement direction of the strip-shaped scintillators in the second detection plate are perpendicular to each other. Therefore, in the space rectangular coordinate system, the Muon detection sensor can simultaneously detect the X coordinate and Y coordinate of the Muon track, and its Z coordinate can be determined by the longitudinal position of the Muon detection sensor.

[0074] Muons are incident on each Muon detection sensor of the two detection devices. The Muon detection sensor can detect and obtain the position coordinates of the Muons. Only when each Muon detection sensor detects the position coordinates of the Muon, the position coordinate data corresponding to the Muon is valid data. Specifically, the least squares method is used to fit the 3 position coordinates detected by the first detection device to obtain a straight line, and the least squares method is used to fit the 3 position coordinates detected by the second detection device to obtain a straight line, and the included angle between the two straight lines is used as the Coulomb scattering angle of the Muons.

[0075] Preferably, the steps of using the Monte Carlo method to analyze and obtain the Muon Coulomb scattering angle data of multiple target materials, and then obtaining the Muon detection data feature set, include:

[0076] S110: When no target material is placed at the central position in the detection space, detection is carried out, and the Monte Carlo method is used to calculate and obtain the Muon Coulomb scattering angle, and then the Muon Coulomb scattering angle data set of air is obtained.

[0077] S120: Each target material, namely uranium, lead, and iron, is respectively placed at the central position in turn, detection is carried out respectively, and the Monte Carlo method is used to calculate and obtain the corresponding Muon Coulomb scattering angle respectively, and then the Muon Coulomb scattering angle data set of each target material is obtained.

[0078] S130. Obtain the muon detection data set of each target material based on the muon Coulomb scattering angle data set of air and the muon Coulomb scattering angle data set of each target material.

[0079] S140. Calculate and obtain the muon detection data feature set of each target material based on the muon detection data set of each target material. The muon detection data feature set of each target material includes the root mean square of the Coulomb scattering angle distribution corresponding to the target material, the proportion of Coulomb scattering angle data, and the kurtosis.

[0080] Preferably, obtaining the muon detection data set of each target material includes:

[0081] Random extraction step: Randomly extract the first preset number of muon Coulomb scattering angles from the muon Coulomb scattering angle data set of each target material; randomly extract the second preset number of muon Coulomb scattering angles from the muon Coulomb scattering angle data set of air; under the condition that the first preset number and the second preset number remain unchanged, execute the random extraction process multiple times to obtain the first detection data set. Preferably, the sum of the first preset number and the second preset number is 10 4 pieces, that is, the flux of muons within 1 minute on a 1m 2 muon detection sensor. Considering that in the actual detection process, the detection device can not only detect the muons passing through the target material, but also detect the muons that do not pass through the target material. Therefore, the detection data set of the target material randomly extracted from the muon Coulomb scattering angle data set of the target material and the muon Coulomb scattering angle data set of air can better simulate the actual detection situation. In addition, multiple random extractions are to improve the universality of the data and reduce its contingency, thereby improving the accuracy of the trained classification model.

[0082] Considering that different sizes of materials will affect the discrimination results of the materials, therefore, the ratio ρ of the first preset number to the second preset number is adjusted multiple times within a preset range according to a preset step size. According to the new ratio obtained by each adjustment, the above random extraction steps are repeated to obtain multiple detection data sets corresponding to different ratios. Specifically, the second preset number actually represents the size of the detection device. When the size of the given detection device is determined, the first preset number actually represents the size of the target material. Preferably, the preset range can be set to 0.1 - 0.4, and the corresponding step size is set to 0.01, or the preset range can be set to 0.01 - 0.1, and the corresponding step size is set to 0.01. Exemplarily, when the ratio of the first preset number to the second preset number is 0.01, the size relationship between the target material and the detection device in the corresponding detection system is as Figure 3As shown, when the ratio of the first preset number to the second preset number is 0.25, the size relationship between the target material and the detection device in the corresponding detection system is as follows Figure 4 As shown

[0083] Preferably, the first detection data set and multiple detection data sets corresponding to different ratios constitute the Muon sub-detection data set of each target material, that is, the Muon sub-detection data set of each target material includes multiple detection data sets corresponding to multiple ratios

[0084] Preferably, based on the Muon sub-detection data set of each target material, the steps of calculating the Muon sub-detection data feature set of each target material include calculating the detection data features corresponding to each detection data set in the Muon sub-detection data set of each target material, specifically including

[0085] Using the Expectation-Maximization algorithm to estimate the root mean square of the Coulomb scattering angle distribution corresponding to each detection data set

[0086]

[0087] Where n represents the number of iterations, when the update amplitude is less than the preset value (i.e ), the corresponding is the root mean square of the Coulomb scattering angle distribution corresponding to the detection data set, x i represents the i-th Coulomb scattering angle in the detection data set, i = 1, 2, 3... N, σ air is the root mean square of the air Coulomb scattering angle distribution corresponding to the detection data set represents the root mean square of the Coulomb scattering angle distribution corresponding to the detection data set obtained in the n-th iteration

[0088] Using the Expectation-Maximization algorithm to estimate the proportion of Coulomb scattering angle data corresponding to each detection data set

[0089]

[0090] Where ρ EM represents the proportion of Coulomb scattering angle data corresponding to the detection data set

[0091] Calculating the kurtosis corresponding to each detection data set

[0092]

[0093] Where K represents the kurtosis corresponding to the detection data set

[0094] Through the above calculation method, multiple sets of features (σ EM , ρ EM , K) corresponding to multiple sets of detection data of each target material can be obtained

[0095] The characteristic distributions of the three target materials (uranium, lead, iron) obtained are as Figure 5 shown.

[0096] Preferably, the steps of training the Muon detection data feature set of multiple target materials to obtain a classification model include:

[0097] Preferably, a part of the Muon detection data feature set of each target material is used as the training set, and the other part is used as the verification set. Exemplarily, for each different ρ value, 2×10 3 groups of data features are taken as the training set, and 1×10 2 groups of data features are taken as the verification set. Then the total training set finally obtained includes 6×10 3 groups of data features, and the total verification set includes 3×10 2 groups of data features.

[0098] Based on the total training set, the support vector machine method is used to perform binary classification training on each target material to obtain the classification model function when each target material is the positive class. Exemplarily, for binary classification training of the target material uranium, the data features corresponding to uranium are marked as the positive class, and the corresponding classification label value is 1. The data features corresponding to lead and iron are marked as the negative class, and the corresponding classification label value is -1.

[0099] The classification model function when each target material is the positive class is:

[0100] f(d) = ∑ j a j y j k(d, d j ) + b,

[0101]

[0102] d = (σ EM , ρ EM , K),

[0103]

[0104] where d j is the jth support vector obtained by support vector machine training, j = 1, 2, 3...M, a j represents the parameter corresponding to the jth support vector, b represents the parameter obtained by support vector machine training, y j is the classification label value corresponding to d j , d is the Muon detection data feature of the target material; σ EM is the root mean square of the Coulomb scattering angle distribution of the target material, ρ EMrepresents the proportion of the Coulomb scattering angle data of the target material, K represents the kurtosis of the Muon detection data set of the target material, k(d, d j ) represents the kernel function, and T represents the transpose of a matrix or vector.

[0105] Preferably, it further includes verifying each classification model function corresponding to the obtained target material by using the total test set to determine the accuracy of the classification model function. Based on the classification model established in this embodiment, when ρ = 0.01, when using a 1m×1m Muon detection sensor to detect a cube-shaped material to be detected with a side length of 10 cm, the classification accuracy of U is 98.3%, the accuracy of Pb is 99.7%, and the classification accuracy of Fe is 100%.

[0106] Preferably, the step of using the classification model to detect the material to be detected to determine whether the material to be detected is a heavy nuclear material includes:

[0107] Detect the Muon detection data set of the material to be detected and analyze the characteristics of the Muon detection data set of the material to be detected.

[0108] Substitute the characteristics of the Muon detection data set of the material to be detected into the classification model function corresponding to each target material to obtain the corresponding similarity value, and use the target material corresponding to the largest similarity value as the classification result of the material to be detected, and determine whether there is a heavy nuclear substance in the material to be detected according to the classification result. Exemplarily, if the target material corresponding to the largest similarity value obtained is U, then there is a heavy nuclear substance in the material to be detected. If the target material corresponding to the largest similarity value obtained is Pb or Fe, then there is no heavy nuclear substance in the material to be detected.

[0109] Exemplarily, the present invention can also be used for the detection of heavy nuclear substances in 20GP standard containers. After obtaining the classification model using the above method, using a 1m×1m plastic scintillator detector as a unit array to arrange a large-area Muon detection sensor to realize the detection of heavy nuclear substances in a 20GP standard container.

[0110] Using cosmic ray Muon detection to obtain the scattering angle distribution characteristics of the object to be measured (i.e., (σ EM , ρ EM, K)), and classify using a classification model based on this distribution characteristic, enabling rapid detection of heavy nuclear materials. The present invention utilizes naturally occurring cosmic ray Muons, which do not generate additional radiation and do not require an additional radiation source, and has the advantages of being simple, convenient, reliable, and stable; cosmic ray Muons have high energy, with an average energy of 3 - 4 GeV, and the main energy loss mode is ionization loss, having good penetration ability and being able to easily penetrate the object to be measured, without problems such as being shielded and unable to be detected, and is particularly suitable for situations where there are shielding objects and effective detection is impossible in cracking down on nuclear material smuggling; the scattering angle distribution characteristic obtained by detecting cosmic ray Muons is related to the material composition of the object to be measured. The higher its atomic number, the more obvious the scattering angle distribution characteristic. Therefore, it has unique advantages in the detection of heavy nuclear material.

[0111] Another embodiment of the present invention discloses a device for rapidly detecting heavy nuclear materials using cosmic ray Muons.

[0112] Since the principle of this device is the same as that of the above method embodiment, the repeated parts can refer to the method embodiment and will not be elaborated here.

[0113] As Figure 6 shown, the device includes:

[0114] A detection system 110 for detecting Muons to obtain corresponding detection data.

[0115] A data processing module 120, based on the Muon Coulomb scattering angle data sets of various target materials, respectively analyzes and obtains the Muon detection data characteristic sets of various target materials.

[0116] A model establishment module 130 for classifying and training the Muon detection data characteristic sets of various said target materials to obtain a classification model;

[0117] A detection module 140 for using the classification model to detect the material to be detected to determine whether there is heavy nuclear material in the material to be detected.

[0118] Preferably, the detection system 110 is specifically used to construct a detection space with a preset volume.

[0119] Inside the detection space, a first group of detection devices and a second group of detection devices are horizontally placed at preset distances above and below the central position respectively. Both the first group of detection devices and the second group of detection devices include 3 Muon detection sensors, and the 3 Muon detection sensors are longitudinally horizontally placed at a preset interval.

[0120] Compared with the prior art, the method and device for rapidly detecting heavy nuclear materials using cosmic ray Muons proposed in the embodiments of the present invention. First, the method and device for rapidly detecting heavy nuclear materials using cosmic ray Muons proposed by the present invention obtain a classification model through model training based on the detection data characteristics of the target material, and discriminate the material to be detected according to the classification model, greatly reducing the amount of Muon detection data required for detection. It only needs to detect Muon events (Muon Coulomb scattering angle) for 1 minute at least, and it can accurately identify whether the material to be detected is a heavy nuclear material, which is applicable to various security inspection scenarios with large material flow, facilitating the rapid detection of whether there are heavy nuclear substances, thus being able to greatly improve the detection rate. Secondly, using the support vector machine method to perform binary classification training based on the detection data feature set of the target material to obtain a classification model can effectively utilize the Muon Coulomb scattering angle data and eliminate the influence of air noise, thereby improving the classification accuracy of the obtained classification model. Importantly, the present invention constructs a Coulomb scattering angle data set of different materials through the Monte Carlo method, and quickly obtains a Muon detection data set of different materials by means of random sampling. The detection data set contains the characteristics corresponding to various sizes of the same material. According to the Muon detection data sets of different materials, classification models of different materials can be quickly obtained in different detection environments without long-term detection and collection of Muon data, facilitating the rapid establishment of a heavy nuclear material detection system.

[0121] Those skilled in the art can understand that all or part of the processes of implementing the above embodiment methods can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a disk, an optical disc, a read-only memory, or a random access memory, etc.

[0122] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for rapidly detecting heavy nuclear materials using cosmic ray muons, characterized in that, It includes the following steps: Based on the muon Coulomb scattering angle datasets of multiple target materials, respectively analyze and obtain the muon detection data feature sets of multiple target materials; Perform classification training on the muon detection data feature sets of multiple said target materials to obtain a classification model; Use the classification model to detect the material to be detected to determine whether there is heavy nuclear matter in the material to be detected; Respectively analyzing and obtaining the muon detection data feature sets of multiple target materials includes obtaining the muon Coulomb scattering angle dataset of air; based on the muon Coulomb scattering angle dataset of air and the muon Coulomb scattering angle dataset of each target material, obtain the muon detection dataset of each target material; based on the muon detection dataset of each said target material, calculate and obtain the muon detection data feature set of each target material; The obtaining of the muon detection dataset of each target material includes: Random sampling step: Randomly sample the first preset number of muon Coulomb scattering angles from the muon Coulomb scattering angle dataset of each target material; randomly sample the second preset number of muon Coulomb scattering angles from the muon Coulomb scattering angle dataset of air; Under the condition that the first preset number and the second preset number remain unchanged, perform the random sampling process multiple times to obtain the first detection data group; Adjust the ratio of the first preset number to the second preset number multiple times within a preset range according to a preset step size, and according to the new ratio obtained each time, repeat the random sampling step to obtain multiple detection data groups corresponding to different said ratios; The first detection data group and the multiple detection data groups corresponding to different said ratios constitute the muon detection dataset of each target material.

2. The method for rapidly detecting heavy nuclear materials according to claim 1, wherein It also includes constructing a detection system, specifically including: Construct a detection space with a preset volume; A first group of detection devices and a second group of detection devices are respectively horizontally placed at a preset distance on the upper side and the lower side inside the detection space; both the first group of detection devices and the second group of detection devices include 3 muon detection sensors, and the 3 muon detection sensors are horizontally placed longitudinally at a preset interval.

3. The method for rapidly detecting heavy nuclear materials according to claim 2, wherein The step of respectively analyzing and obtaining the muon detection data feature sets of multiple target materials based on the muon Coulomb scattering angle datasets of multiple target materials further includes: When no target material is placed at the central position in the detection space, perform detection, and use the Monte Carlo method to calculate and obtain the muon Coulomb scattering angle, and then obtain the muon Coulomb scattering angle dataset of air; Place each target material at the central position in turn, respectively perform detection, and use the Monte Carlo method to calculate and obtain the corresponding muon Coulomb scattering angle respectively, and then obtain the muon Coulomb scattering angle dataset of each target material; The muon detection data feature set of each target material includes the root mean square of the Coulomb scattering angle distribution corresponding to the target material, the proportion of the Coulomb scattering angle data, and the kurtosis.

4. The method for rapidly detecting heavy nuclear materials according to claim 1, characterized in that The step of calculating and obtaining the muon detection data feature set of each target material based on the muon detection data set of each target material includes calculating the detection data features corresponding to each detection data group in the muon detection data set of each target material, specifically including: Estimating the root mean square of the Coulomb scattering angle distribution corresponding to each detection data group by using the expectation-maximization algorithm; where n represents the number of iterations. When the update amplitude is less than a preset value, the corresponding is the root mean square of the Coulomb scattering angle distribution corresponding to the detection data group, and x i represents the i-th Coulomb scattering angle in the detection data group, where i = 1, 2, 3... N, and σ air is the root mean square of the air Coulomb scattering angle distribution corresponding to the detection data group, represents the root mean square of the Coulomb scattering angle distribution corresponding to the detection data group obtained in the n-th iteration; Estimating the proportion of the Coulomb scattering angle data corresponding to each detection data group by using the expectation-maximization algorithm; Among them, ρ EM represents the proportion of the Coulomb scattering angle data corresponding to the detection data group; Calculating the kurtosis corresponding to each detection data group; where K represents the kurtosis corresponding to the detection data group.

5. The method for rapidly detecting heavy nuclear materials according to any one of claims 2-4, characterized in that, The step of training the muon detection data feature sets of multiple target materials to obtain a classification model includes: Based on the training set composed of the muon detection data feature sets of each target material, performing binary classification training on each target material by using the support vector machine method to obtain the classification model function when each target material is used as the positive class; The classification model function when each target material is used as the positive class is: f(d) = ∑ j a j y j k(d, d j ) + b, d = (σ EM , ρ EM , K), where d j is the j-th support vector obtained by support vector machine training, j = 1, 2, 3... M, a j represents the parameter corresponding to the j-th support vector, b represents the parameter obtained by support vector machine training, y j is the classification label value corresponding to d j , d is the Muon sub-detection data feature of the target material; σ EM is the root mean square of the Coulomb scattering angle distribution of the target material, ρ EM represents the proportion of the Coulomb scattering angle data of the target material, and K represents the kurtosis of the Muon sub-detection data set of the target material.

6. The method for rapidly detecting heavy nuclear materials according to claim 5, wherein, The step of using the classification model to detect the material to be detected to determine whether there is heavy nuclear material in the material to be detected includes: Detecting the muon detection data set of the material to be detected and analyzing to obtain the characteristics of the muon detection data set of the material to be detected; Substituting the characteristics of the muon detection data set of the material to be detected into the classification model function corresponding to each target material to obtain the corresponding similarity value, and taking the target material corresponding to the maximum similarity value as the classification result of the material to be detected, and determining whether there is heavy nuclear material in the material to be detected according to the classification result.

7. The method for rapidly detecting heavy nuclear materials according to any one of claims 2-4 and 6, characterized in that, The muon detection sensor includes a first detection plate and a second detection plate which are mutually coupled. Both the first detection plate and the second detection plate include a plurality of strip-shaped scintillators arranged in parallel, and the arrangement direction of the strip-shaped scintillators in the first detection plate is perpendicular to the arrangement direction of the strip-shaped scintillators in the second detection plate.

8. An apparatus for implementing the method of rapidly detecting heavy nuclear materials by using cosmic ray Muons as described in claim 1, characterized in that, Including: A data processing module, which respectively analyzes and obtains the muon detection data feature sets of multiple target materials based on the muon Coulomb scattering angle data sets of multiple target materials; A model establishment module, which is used to perform classification training on the muon detection data feature sets of multiple target materials to obtain a classification model; A detection module, which is used to use the classification model to detect the material to be detected to determine whether there is heavy nuclear material in the material to be detected.

9. The device according to claim 8, characterized in that It also includes a detection system; The detection system is used to construct a detection space with a preset volume; A first group of detection devices and a second group of detection devices are respectively horizontally placed at a preset distance on the upper side and the lower side inside the detection space; both the first group of detection devices and the second group of detection devices include 3 muon detection sensors, and the 3 muon detection sensors are longitudinally and horizontally placed at a preset interval.

Images