A multi-angle PGNAA ore grade detection method based on compressed sensing

Abstract

The present invention provides a multi-angle PGNAA ore grade detection method based on compressed sensing, which relates to the technical field of ore grade detection, including: obtaining multiple groups of ore grade detection information of the test ore to construct a multidimensional matrix of detection information, each group of ore grade detection information includes multi-angle detection results; obtaining multiple groups of ore grade actual information of the test ore to construct a multidimensional matrix of actual information, each group of ore grade actual information includes multi-angle actual results; obtaining a primary weight matrix and combining each group of ore grade actual information to construct a primary detection equation to calculate the corresponding ore grade detection information; optimizing the primary weight matrix through compressed sensing technology to obtain a target weight matrix, and obtaining a target detection equation according to the target weight matrix; obtaining the actual ore grade information of the ore to be detected, and calculating the ore grade detection information in combination with the target detection equation. The present application can calculate the grade information of each piece of ore and realize accurate detection of ore grade.

Description

Technical Field

[0001] The present invention relates to the technical field of ore grade detection, and in particular to a multi-angle PGNAA ore grade detection method based on compressed sensing. Background Art

[0002] As modern mining is developing towards refinement and efficiency, traditional chemical detection methods can no longer meet the needs of real-time monitoring, making methods with faster detection speed and more accurate test results more popular.

[0003] Ore grade refers to the mass content of useful minerals in ore, usually expressed as a percentage. Grade testing, as the process of measuring and analyzing the content of valuable minerals in ore, is an important part of the mineral processing process. It can help determine the value and quality of the ore, which is of great significance for subsequent ore mining and processing.

[0004] In the prior art, the common grade detection methods are divided into two types according to the physical and chemical properties of the ore. Among them, the physical detection sorting methods include: PGNAA, XRT, NIR and other sensor-based grade detection methods; chemical detection sorting methods include: flotation method, etc. Taking the PGNAA method (i.e., prompt gamma neutron activation analysis) as an example, this method can perform high-sensitivity and accurate determination of multiple mineral components without destroying the sample, which can improve resource utilization efficiency and simplify sample pretreatment steps. However, it can only measure the overall grade of the sample, and cannot obtain the specific grade information of each stone in a group of samples. Summary of the invention

[0005] Based on this, the purpose of the present invention is to provide a multi-angle PGNAA ore grade detection method based on compressed sensing, which is used to solve the technical problem that the detection method in the prior art can only measure the overall grade of the sample, but cannot obtain the specific grade information of each stone in a group of samples.

[0006] On one hand, the present invention provides a multi-angle PGNAA ore grade detection method based on compressed sensing, comprising: obtaining multiple groups of ore grade detection information of the test ore to construct a detection information multi-dimensional matrix S, each group of ore grade detection information is a detection information one-dimensional matrix s, each group of ore grade detection information includes multi-angle detection results, wherein s=m*1, S= m*I, I is the number of groups, m is the number of PGNAA detection devices, and each PGNAA detection device corresponds to a projection angle;

[0007] Acquire multiple groups of actual ore grade information of the test ore to construct an actual information multi-dimensional matrix X, each group of actual ore grade information is an actual information one-dimensional matrix x, and each group of actual ore grade information includes multi-angle actual results, wherein x=n*1, X= n*I, I is the number of groups, and n is the number of ores contained in a group of ores to be tested;

[0008] Obtain the primary weight matrix K and construct the primary detection equation in combination with each group of actual ore grade information to calculate the corresponding ore grade detection information, wherein the primary detection equation is K*x=s; optimize the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1, and obtain the target detection equation according to the target weight matrix K1, wherein the target detection equation is K1*x =s;

[0009] The multi-angle PGNAA detection equipment analysis images of the ore to be detected are obtained to obtain multiple groups of actual ore grade information, and the ore grade detection information of the ore to be detected is calculated based on the actual ore grade information and combined with the target detection equation.

[0010] In addition, the multi-angle PGNAA ore grade detection method based on compressed sensing according to the present invention may also have the following additional technical features:

[0011] Furthermore, the step of optimizing the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1 includes:

[0012] Multiple sets of ore grade detection information and multiple sets of ore grade actual information are constructed to obtain multiple sets of training sets, and the primary weight matrix K is optimized by combining the multiple sets of training sets through compressed sensing technology to obtain the target weight matrix K1.

[0013] Furthermore, the method of obtaining the primary weight matrix K includes:

[0014] The primary weight matrix K is constructed according to the number m of PGNAA detection equipment and the number n of ores contained in a group of ores, K=m*n, where m is the number of PGNAA detection equipment and n is the number of ores contained in a group of ores to be detected.

[0015] Furthermore, the step of optimizing the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1 includes:

[0016] Initialize the residual R = detection information multidimensional matrix S, initialize K0 to an empty matrix, and preset the maximum number of iterations T, where t1 = 1, t max =T;

[0017] Select the column with the greatest correlation with the residual in the actual information multidimensional matrix X x imax , and the column with the largest residual correlationx imax Add to matrix K t Among them, K t =[K t-1 , x imax ], is the weight matrix after the tth iteration;

[0018] Solve the matrix K by the least squares method t So that S≈K t *X;

[0019] Subtract the detection information multidimensional matrix S and K t *The matching part between X is used to update the residual R;

[0020] Update the number of iterations t=t+1 until the number of iterations t reaches T, and the iteration ends and the output matrix K t , the matrix K t is the target weight matrix K1.

[0021] Furthermore, the column with the greatest correlation with the residual is selected in the actual information multidimensional matrix X x imax In the step of x imax The methods for obtaining include:

[0022] ;

[0023] In the formula, argmax represents the variable value at which the function takes the maximum value; r t-1 is the residual of the t-1th iteration; x i is the first i List, i The value of ranges from 1 to N; N Indicates that there are N columns in the actual information multidimensional matrix X; imax is the column with the largest residual correlation.

[0024] Furthermore, by using the least squares method to solve the matrix K t In the step, K t The calculation formula is:

[0025] K t =argmin||SK t *X||2;

[0026] In the formula, argmin represents the variable value of the function taking the minimum value; K t It is represented as the weight matrix after the tth iteration; S is the multidimensional matrix of detection information; X is the multidimensional matrix of actual information.

[0027] Further, after subtracting the detection information multidimensional matrix S and K t * In the step of updating the residual R, the calculation formula for updating the residual R is:

[0028] R=S-K t *X;

[0029] Where R is the residual, S is the multidimensional matrix of detection information; K t It is represented as the weight matrix after the tth iteration; X is the actual information multidimensional matrix.

[0030] On the other hand, the present invention provides a multi-angle PGNAA ore grade detection system based on compressed sensing, which is used to implement the above-mentioned multi-angle PGNAA ore grade detection method based on compressed sensing. The system includes an ore conveyor belt and multiple PGNAA detection devices. A detection area is provided on the ore conveyor belt, and multiple PGNAA detection devices are arranged around the periphery of the detection area; the PGNAA detection device includes a neutron generator and a detector, and the neutron generator and the detector are arranged in a one-to-one correspondence, and the neutron generator and the detector are arranged diagonally relative to the detection area.

[0031] In addition, the multi-angle PGNAA ore grade detection system based on compressed sensing according to the present invention may also have the following additional technical features:

[0032] Further, the plurality of PGNAA detection devices include a first PGNAA detection device, and a second PGNAA detection device and a third PGNAA detection device respectively arranged on both sides of the first PGNAA detection device, the first PGNAA detection device includes a first neutron generator and a first detector; the second PGNAA detection device includes a second neutron generator and a second detector; the third PGNAA detection device includes a third neutron generator and a third detector;

[0033] The first detector, the second detector, the third detector and the second neutron generator are arranged on the same side of the ore conveying belt; the first neutron generator and the third neutron generator are arranged on the other side of the ore conveying belt.

[0034] Furthermore, the second neutron generator is arranged in parallel with the detection area.

[0035] The above-mentioned multi-angle PGNAA ore grade detection method based on compressed sensing, by combining the PGNAA method with compressed sensing technology, uses the projection information of the data in multiple directions to reversely obtain the original data, and can calculate the grade information of each piece of ore, so as to realize the accurate detection of the ore grade; specifically, first obtain multiple groups of ore grade detection information of the test ore, multiple groups of ore grade actual information and primary weight matrix K to construct the primary detection equation, and then optimize the primary weight matrix K to obtain the optimal solution of the primary weight matrix K, so as to avoid the primary weight matrix K being a non-full rank matrix, and the optimal solution is used as the target weight matrix K1, and the target weight matrix K1 replaces the primary weight matrix K in the primary detection equation to obtain the target detection equation, and then substitutes the actual ore grade information of the ore to be detected into the target detection equation to obtain the ore grade detection information of the ore to be detected. This application can obtain the grade information of multiple pieces of ore and their grade spatial distribution position information at one time in the face of a large number of ores, so as to improve the efficiency of the PGNAA detection equipment in detecting the target ore grade during the mineral processing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Schematic diagram of the reaction between thermal neutrons and atomic nuclei in the PGNAA method;

[0037] Figure 2 This is a schematic diagram of the working mode of traditional PGNAA detection equipment;

[0038] Figure 3 This is a flow chart of the multi-angle PGNAA ore grade detection method based on compressed sensing of the present invention;

[0039] Figure 4 It is a schematic diagram of a multi-angle PGNAA ore grade detection system based on compressed sensing of the present invention;

[0040] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0041] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0042] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0044] Regarding the PGNAA method, the PGNAA detection equipment uses an isotope neutron excitation source. The thermal neutrons are slowed down by the measuring device. The thermal neutrons and the element nuclei in the material undergo a thermal neutron capture reaction, causing the nuclei to enter an excited state. Since the excited state is not stable, in the process of de-excitation toward a stable state, the nuclei of different elements will emit gamma rays of different energies. The principle is shown in Figure 1 The detector receives the characteristic gamma rays emitted by the material and forms a corresponding energy spectrum after signal processing. The energy spectrum contains the type and grade information of the target element.

[0045] The same group of ore samples were analyzed by PGNAA method from different angles. Due to the uneven distribution inside the ore samples to be tested, the neutrons reacted with the atomic nuclei of the ore to be tested in different ways, the nuclei entered the excited state to different degrees, and the intensities of gamma rays released after deexcitation were different, resulting in different PGNAA results.

[0046] In actual use, Figure 2 As shown, the ore is transported to the machine by a conveyor belt. Above or in a certain direction of the conveyor belt is a neutron emitter. The neutron emitter serves as a neutron emission source, and a corresponding detector is provided on the opposite side or nearby to receive the gamma rays released by the material. The detector is surrounded by a neutron shield and a gamma shield to prevent the environment from interfering with the test results. Due to the characteristics of the PGNAA method itself, the traditional method can only obtain one result for a group of samples to be tested, that is, the overall grade of this group of samples. However, with the increasing demand for accuracy and efficiency of ore grade detection, the original method has gradually been unable to meet the requirements. In order to accurately detect the ore grade of each piece of ore in a group of samples and its spatial distribution position information, the present application provides a multi-angle PGNAA ore grade detection method based on compressed sensing.

[0047] Specifically, this application combines PGNAA technology with compressed sensing technology, and uses the projection information of data in multiple directions to invert the original data, so as to calculate the grade information of each piece of ore and realize accurate detection of ore grade; specifically, first obtain multiple groups of ore grade detection information of the test ore, multiple groups of actual ore grade information and primary weight matrix K to construct a primary detection equation, and then optimize the primary weight matrix K to obtain the optimal solution of the primary weight matrix K, so as to avoid the primary weight matrix K being a non-full rank matrix. The optimal solution is used as the target weight matrix K1, and the target weight matrix K1 replaces the primary weight matrix K in the primary detection equation to obtain the target detection equation, and then the actual ore grade information of the ore to be detected is substituted into the target detection equation to obtain the ore grade detection information of the ore to be detected. This application can obtain the grade information of multiple pieces of ore at a time when facing a large amount of ore, and improve the efficiency of PGNAA detection equipment in detecting the target ore grade during the mineral processing process.

[0048] In order to facilitate understanding of the present invention, several embodiments of the present invention are provided below. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0049] Embodiment 1

[0050] See also Figure 3 , which shows a multi-angle PGNAA ore grade detection method based on compressed sensing in the first embodiment of the present invention, including steps S101 to S104:

[0051] S101. Obtain multiple groups of ore grade detection information of the test ore to construct a multi-dimensional matrix S of detection information, each group of ore grade detection information is a one-dimensional matrix s of detection information, and each group of ore grade detection information includes multi-angle detection results, wherein s=m*1, S= m*I, I is the number of groups, m is the number of PGNAA detection equipment, and each PGNAA detection equipment corresponds to a projection angle.

[0052] In this embodiment, multiple PGNAA detection devices detect the ore on the conveyor belt from different angles to obtain the detection results, and then analyze the detection results through compressed sensing technology to calculate the ore grade information and its grade spatial distribution position information through the target detection equation, that is, the grade of a certain ore and its specific position in this group of ores can be obtained. Figure 4 As shown, Figure 4 There are three pieces of ores in total, assuming that from left to right they are ore a, ore b and ore c. In this embodiment, not only can the ore grades of ore a, ore b and ore c be calculated, but it can also be obtained that ore a is located on the far left, ore b is located in the middle, and ore c is located on the far right.

[0053] S102. Acquire multiple groups of actual ore grade information of the test ore to construct an actual information multidimensional matrix X, where each group of actual ore grade information is an actual information one-dimensional matrix x, and each group of actual ore grade information includes multi-angle actual results, wherein x=n*1, X=n*I, I is the number of groups, and n is the number of ores contained in a group of ores to be tested.

[0054] After I groups of data are collected, X is an n*I matrix, which contains the actual information of the ore grade of I groups; S is an m*I matrix, which represents the detection information of the ore grade of I groups.

[0055] S103. Obtain the primary weight matrix K and construct a primary detection equation in combination with the actual information of each group of ore grade to calculate the corresponding ore grade detection information, wherein the primary detection equation is K*x=s; optimize the primary weight matrix K through compressed sensing technology to obtain the target weight matrix K1, and obtain the target detection equation according to the target weight matrix K1, wherein the target detection equation is K1*x=s.

[0056] As a specific example, the method for obtaining the primary weight matrix K includes: constructing the primary weight matrix K according to the number m of PGNAA detection equipment and the number n of ores contained in a group of ores, K=m*n, wherein m is the number of PGNAA detection equipment, and n is the number of ores contained in a group of ores to be detected.

[0057] In this embodiment, the step of optimizing the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1 includes: constructing multiple sets of training sets from multiple sets of ore grade detection information and multiple sets of actual ore grade information, combining the multiple sets of training sets and optimizing the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1.

[0058] In this embodiment, the OMP algorithm is used to estimate the primary weight matrix K matrix in the case of insufficient rank. Since the primary weight matrix K is not necessarily a full rank matrix, the OMP algorithm in the compressed sensing technology is used to calculate the optimal solution of the primary weight matrix K using multiple sets of known ore grade actual information x and ore grade detection information s. The OMP algorithm is sparse and has certain robustness to noise and measurement errors, which is simpler than other algorithms. Through multiple iterations, the residual is continuously reduced, so that the target weight matrix K1 is restored as much as possible.

[0059] Specifically, the method for obtaining the target weight matrix K1 includes steps S1031 to S1035:

[0060] S1031, initialize the residual R = detection information multidimensional matrix S, initialize K0 to an empty matrix, and preset the maximum number of iterations T, where t1 = 1, tmax =T.

[0061] S1032. Select the column with the greatest correlation with the residual in the actual information multidimensional matrix X. x imax , and the column with the largest residual correlation x imax Add to matrix K t Among them, K t =[K t-1 , x imax ] is the weight matrix after the tth iteration.

[0062] In this example, the column with the largest residual correlation x imax The methods for obtaining include:

[0063] ;

[0064] In the formula, argmax represents the variable value at which the function takes the maximum value; r t-1 is the residual of the t-1th iteration; x i is the first i List, i The value of ranges from 1 to N; N Indicates that there are N columns in the actual information multidimensional matrix X; imax is the column with the largest residual correlation.

[0065] S1033. Solve the matrix K by the least squares method t So that S≈K t *X.

[0066] In this embodiment, K t The calculation formula is: K t =argmin||SK t *X||2;

[0067] In the formula, argmin represents the variable value of the function taking the minimum value; K t It is represented as the weight matrix after the tth iteration; S is the multidimensional matrix of detection information; X is the multidimensional matrix of actual information.

[0068] S1034, subtract the detection information multidimensional matrix S and K t *X to update the residual R.

[0069] In this embodiment, the calculation formula of the updated residual R is: R = S - K t *X;

[0070] Where R is the residual, S is the multidimensional matrix of detection information; K t It is represented as the weight matrix after the tth iteration; X is the actual information multidimensional matrix.

[0071] S1035, update the number of iterations t=t+1, until the number of iterations t reaches T, the iteration ends and the output matrix K t , the matrix K t is the target weight matrix K1.

[0072] After obtaining the target weight matrix K1, any set of actual ore grade information x obtained through detection can be calculated by s=K1*x to get the value of the corresponding set of ore grade detection information s, and each set of samples to be detected includes multiple ores, that is, the specific grade of each ore in a set of samples to be detected can be calculated by s=K1*x.

[0073] S104, obtaining multi-angle PGNAA detection equipment analysis images of the ore to be detected to obtain multiple groups of actual ore grade information, and calculating the ore grade detection information of the ore to be detected based on the actual ore grade information and in combination with the target detection equation.

[0074] The present invention is based on compressed sensing technology. It uses multiple PGNAA detection devices to perform multi-angle measurements on the ore, and substitutes the results into the compressed sensing module, so that it can detect multiple ores at the same time, and achieve accurate measurement of the grade of each piece of ore, solving the problem of low detection efficiency of traditional methods in ore detection. At the same time, the grade of larger pieces of ore can be detected, and even ores with complex matrices can obtain better detection results. The operating steps in the detection process are convenient, and the material requirements are simpler than chemical methods. By using the compressed sensing method, the shortcomings of the original PGNAA method in terms of efficiency are compensated. The algorithm proposed in the present invention can be beneficially applied to the ore detection process to perform efficient real-time detection of mined ore.

[0075] In summary, the multi-angle PGNAA ore grade detection method based on compressed sensing in the above-mentioned embodiment of the present invention, by combining the PGNAA method with compressed sensing technology, uses the projection information of the data in multiple directions to reversely obtain the original data, and can calculate the grade information of each piece of ore, so as to realize the accurate detection of the ore grade; specifically, first obtain multiple groups of ore grade detection information of the test ore, multiple groups of ore grade actual information and primary weight matrix K to construct the primary detection equation, and then optimize the primary weight matrix K to obtain the optimal solution of the primary weight matrix K, so as to avoid the primary weight matrix K being a non-full rank matrix, and the optimal solution is used as the target weight matrix K1, and the target weight matrix K1 replaces the primary weight matrix K in the primary detection equation to obtain the target detection equation, and then substitutes the actual ore grade information of the ore to be detected into the target detection equation to obtain the ore grade detection information of the ore to be detected. This application can obtain the grade information of multiple pieces of ore and their grade spatial distribution position information at one time in the face of a large number of ores, and improve the efficiency of the PGNAA detection equipment in detecting the target ore grade during the mineral processing process.

[0076] Embodiment 2

[0077] Please refer to Figure 4 , shown is a multi-angle PGNAA ore grade detection system based on compressed sensing in the second embodiment of the present invention, which is used to implement the multi-angle PGNAA ore grade detection method based on compressed sensing in the above embodiment. The system includes an ore conveyor belt and a plurality of PGNAA detection devices. A detection area is provided on the ore conveyor belt, and a plurality of PGNAA detection devices are arranged at the periphery of the detection area; the PGNAA detection device includes a neutron generator and a detector, the neutron generator and the detector are arranged in a one-to-one correspondence, and the neutron generator and the detector are arranged diagonally relative to the detection area.

[0078] As a specific example, the plurality of PGNAA detection devices include a first PGNAA detection device, and a second PGNAA detection device and a third PGNAA detection device respectively arranged on both sides of the first PGNAA detection device, the first PGNAA detection device includes a first neutron generator and a first detector; the second PGNAA detection device includes a second neutron generator and a second detector; the third PGNAA detection device includes a third neutron generator and a third detector;

[0079] Specifically, Figure 4There are three sets of PGNAA detection equipment in total. The first detector, the second detector, the third detector and the second neutron generator are arranged on the same side of the ore conveyor belt; the first neutron generator and the third neutron generator are arranged on the other side of the ore conveyor belt. Furthermore, the second neutron generator is arranged in parallel with the detection area. Neutron generators emit neutrons from three different angles to the test samples transported by the conveyor belt to the inside of the device for grade detection. The test results are received by the corresponding detectors respectively, and the result is the sum of the grades detected on the samples at a certain angle.

[0080] In summary, the multi-angle PGNAA ore grade detection system based on compressed sensing in the above-mentioned embodiment of the present invention, by combining the PGNAA method with compressed sensing technology, uses the projection information of the data in multiple directions to reversely obtain the original data, and can calculate the grade information of each piece of ore, so as to realize the accurate detection of the ore grade; specifically, first obtain multiple groups of ore grade detection information of the test ore, multiple groups of ore grade actual information and primary weight matrix K to construct the primary detection equation, and then optimize the primary weight matrix K to obtain the optimal solution of the primary weight matrix K, so as to avoid the primary weight matrix K being a non-full rank matrix, and the optimal solution is used as the target weight matrix K1, and the target weight matrix K1 replaces the primary weight matrix K in the primary detection equation to obtain the target detection equation, and then substitutes the actual ore grade information of the ore to be detected into the target detection equation to obtain the ore grade detection information of the ore to be detected. This application can obtain the grade information of multiple pieces of ore and their grade spatial distribution position information at one time in the face of a large number of ores, and improve the efficiency of the PGNAA detection equipment in detecting the target ore grade during the mineral processing process.

[0081] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0082] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A multi-angle PGNAA ore grade detection method based on compressed sensing, characterized in that: include: Acquire multiple groups of ore grade detection information of the test ore to construct a detection information multidimensional matrix S, each group of ore grade detection information is a detection information one-dimensional matrix s, and each group of ore grade detection information includes multi-angle detection results, where s=m*1, S= m*I, I is the number of groups, m is the number of PGNAA detection equipment, and each PGNAA detection equipment corresponds to a projection angle; Acquire multiple groups of actual ore grade information of the test ore to construct an actual information multi-dimensional matrix X, each group of actual ore grade information is an actual information one-dimensional matrix x, and each group of actual ore grade information includes multi-angle actual results, wherein x=n*1, X= n*I, I is the number of groups, and n is the number of ores contained in a group of ores to be tested; Obtain the primary weight matrix K and construct the primary detection equation in combination with each group of actual ore grade information to calculate the corresponding ore grade detection information, wherein the primary detection equation is K*x=s; optimize the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1, and obtain the target detection equation according to the target weight matrix K1, wherein the target detection equation is K1*x =s; Acquire multi-angle PGNAA detection equipment analysis images of the ore to be detected to obtain multiple groups of actual ore grade information, and calculate the ore grade detection information of the ore to be detected based on the actual ore grade information and in combination with the target detection equation; The method for obtaining the primary weight matrix K includes: The primary weight matrix K is constructed according to the number m of PGNAA detection equipment and the number n of ores contained in a group of ores, K=m*n, where m is the number of PGNAA detection equipment and n is the number of ores contained in a group of ores to be detected; The steps of optimizing the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1 include: Multiple sets of ore grade detection information and multiple sets of ore grade actual information are constructed to obtain multiple sets of training sets, and the primary weight matrix K is optimized by using the compressed sensing technology to obtain the target weight matrix K1; The step of optimizing the primary weight matrix K by compressed sensing technology to obtain the target weight matrix K1 also includes: Initialize the residual R = detection information multidimensional matrix S, initialize K0 to an empty matrix, and preset the maximum number of iterations T, where t1 = 1, t max =T; Select the column with the greatest correlation with the residual in the actual information multidimensional matrix X x imax , and the column with the largest residual correlation x imax Add to matrix K t Among them, K t =[K t-1 , x imax ], is the weight matrix after the tth iteration; Solve the matrix K by the least squares method t So that S≈K t *X; Subtract the detection information multidimensional matrix S and K t *The matching part between X to update the residual R; Update the number of iterations t=t+1 until the number of iterations t reaches T, and the iteration ends and the output matrix K t , the matrix K t is the target weight matrix K1.

2. The multi-angle PGNAA ore grade detection method based on compressed sensing according to claim 1 is characterized in that: Select the column with the greatest correlation with the residual in the actual information multidimensional matrix X x imax In the step of x imax The methods for obtaining include: ； In the formula, argmax represents the variable value at which the function takes the maximum value; r t-1 is the residual of the t-1th iteration; x i is the first i List, i The value of ranges from 1 to N; N Indicates that there are N columns in the actual information multidimensional matrix X; imax is the column with the largest residual correlation.

3. The multi-angle PGNAA ore grade detection method based on compressed sensing according to claim 1 is characterized in that: Solving the matrix K by the least squares method t In the step, K t The calculation formula is: K t =argmin||S-K t *X||2; In the formula, argmin represents the variable value of the function taking the minimum value; K t It is represented as the weight matrix after the tth iteration; S is the multidimensional matrix of detection information; X is the multidimensional matrix of actual information.

4. The multi-angle PGNAA ore grade detection method based on compressed sensing according to claim 1 is characterized in that: After subtracting the detection information multidimensional matrix S from K t * In the step of updating the residual R in the matching part between X, the calculation formula for updating the residual R is: R=S- K t *X; Where R is the residual, S is the multidimensional matrix of detection information; K t It is represented as the weight matrix after the tth iteration; X is the actual information multidimensional matrix.

Images