Neutron spectrum unfolding method capable of using multiple prior information
By constructing the training matrix and performing sparse training, a sparse expression matrix of the prior information of neutron energy spectrum is generated, and combining sparse reconstruction and sparse expression matrix to solve the neutron energy spectrum, the problem of difficult to effectively utilize multiple prior information in the existing technology is solved, and the high accuracy of neutron energy spectrum solution is achieved.
Patent Information
- Application Number
- CN202411626407.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-14
- Publication Date
- 2025-03-18
AI Technical Summary
The existing neutron energy spectrum despectral technology is difficult to effectively utilize multiple prior information, resulting in limited accuracy of the solution spectrum.
By constructing the training matrix and performing sparse training, a sparse expression matrix of the prior information of the neutron energy spectrum is generated, and a method of solving the neutron energy spectrum is combined with sparse reconstruction and sparse expression matrix to solve the neutron energy spectrum, the use of multiple prior information of the neutron energy spectrum is realized.
The accuracy of neutron energy spectrum solution is improved, and a higher precision solution can be achieved in the typical shape of the shielded structure of the nuclear reactor.
Smart Images
Figure CN119644396A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of radiation detection, and in particular relates to a neutron energy spectrum decomposition method that can use multiple prior information. Background Art
[0002] Accurate neutron spectrum measurement has always been one of the key issues in the fields of neutron science, reactor physics and shielding. Neutron spectrum measurement usually faces difficulties such as a large energy span (from hot zone to tens of MeV), severe γ background interference and violent statistical fluctuations of the radiation field. The neutron spectrum cannot usually be measured directly, but needs to be obtained by decomposing the spectrum in combination with the detector response function.
[0003] Existing spectrum decomposition technologies usually treat the problem of solving the neutron energy spectrum as an indeterminate problem. The accuracy of the prior information has a great influence on the accuracy of the spectrum decomposition. Existing spectrum decomposition methods usually use the prior information as the initial value for iteration, so only a single prior information can be used for each spectrum decomposition, which is not conducive to the accurate decomposition of the neutron energy spectrum.
[0004] In this context, the present invention proposes a novel neutron energy spectrum interpretation method, which can use multiple prior information in the interpretation, thereby improving the accuracy of the interpretation. Summary of the invention
[0005] The purpose of the present invention is to provide a neutron energy spectrum decomposition method that can use multiple prior information, process the neutron energy spectrum prior information into a sample matrix, use a training dictionary method to train the neutron energy spectrum sample matrix, and generate a sparse expression matrix of the neutron energy spectrum prior information; secondly, combine the results of sparse reconstruction of the neutron energy spectrum measurement linear equation group to achieve neutron energy spectrum decomposition, and realize the use of multiple prior information of the neutron energy spectrum, thereby improving the accuracy of neutron energy spectrum decomposition.
[0006] The technical solution of the present invention is as follows: A neutron spectrum analysis method that can use multiple prior information includes the following steps:
[0007] Step 1: Construct a training matrix;
[0008] Step 2: Perform sparse training;
[0009] Step 3: Sparsely reconstruct the neutron energy spectrum;
[0010] Step 4: Solve the neutron spectrum;
[0011] Step 5: Output the decomposition results.
[0012] In step 1, the neutron energy spectrum x is obtained by Monte Carlo simulation. i As a priori information for dictionary training, it forms the training matrix
[0013] In step 1, n is the number of neutron energy spectra, ranging from hundreds to hundreds of thousands.
[0014] The step 2 comprises:
[0015]
[0016] Among them, l(x,D) is the regularized objective function, which is obtained by seeking Minimize D and α i Value, determine the sparse expression matrix D and sparse coefficient vector α i ; λ is the regularization parameter; x i is the i-th neutron energy spectrum in the training matrix X; α i For the corresponding x i , which can minimize the regularization objective function; ||·||1 is the sum of the absolute values of the elements; ||·||2 is the sum of the square values of the elements.
[0017] In step 3, the response matrix M is obtained by detector calibration and Monte Carlo simulation, and the detector count A is obtained by experimental measurement. Based on the response matrix M and the detector count A, the neutron energy spectrum is sparsely reconstructed:
[0018]
[0019] In step 3, the annealing algorithm and the iterative regularization algorithm are used to obtain The minimum value of value.
[0020] In step 4, the sparse expression matrix D is used with Solve the neutron spectrum by multiplication;
[0021]
[0022] In step 4 The meaning of the value is the representation of the neutron energy spectrum φ in the sparse expression matrix D.
[0023] The spectrum decomposition result in step 5 is a set of vectors, the number of vector elements is the same as the number of energy group divisions, the value of each element represents the neutron flux measurement result of the energy group, and the sum of each element represents the total neutron flux measurement result.
[0024] The beneficial effect of the present invention is that the present invention can realize the use of multiple prior information of neutron energy spectrum, thereby improving the accuracy of neutron energy spectrum interpretation. For neutron energy spectra of several typical shapes of nuclear reactor shielding structures, the present invention can achieve high-precision spectrum interpretation, see Figure 1-4 . BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 Neutron energy spectrum for the first typical shape of nuclear reactor shielding structure;
[0026] Figure 2 Neutron energy spectrum for the second typical shape of nuclear reactor shielding structure;
[0027] Figure 3 Neutron energy spectrum for the third typical shape of nuclear reactor shielding structure;
[0028] Figure 4 This is the fourth typical shape of neutron energy spectrum for nuclear reactor shielding structure. DETAILED DESCRIPTION
[0029] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0030] A neutron spectrum analysis method that can use multiple prior information includes the following steps:
[0031] Step 1: Use Monte Carlo simulation to obtain the neutron energy spectrum x i As a priori information for dictionary training, it forms the training matrix n is the number of neutron energy spectra, ranging from hundreds to hundreds of thousands;
[0032] Step 2: Perform sparse training:
[0033]
[0034] Among them, l(x,D) is the regularized objective function, which is obtained by seeking Minimize D and α i Value, determine the sparse expression matrix D and sparse coefficient vector α i ; λ is the regularization parameter; x i is the i-th neutron energy spectrum in the training matrix X; α i For the corresponding x i , which can minimize the regularization objective function; ||·||1 is the sum of the absolute values of the elements; ||·||2 is the sum of the square values of the elements;
[0035] Step 3: Obtain the response matrix M through detector calibration and Monte Carlo simulation, obtain the detector count A through experimental measurement, and perform sparse reconstruction of the neutron energy spectrum based on the response matrix M and the detector count A:
[0036]
[0037] It is solved by annealing algorithm and iterative regularization algorithm. The minimum value of value;
[0038] Step 4: Through the sparse expression matrix D and Solve the neutron spectrum by multiplication;
[0039]
[0040] The meaning of the value is the representation of the neutron energy spectrum φ in the sparse expression matrix D;
[0041] Step 5: Output the spectrum analysis results
[0042] The result of spectrum decomposition is a set of vectors. The number of vector elements is the same as the number of energy group divisions. The value of each element represents the neutron flux measurement result of the energy group, and the sum of each element represents the total neutron flux measurement result.
Claims
1. A neutron spectrum analysis method that can use multiple prior information, characterized in that: The steps include: Step 1: Construct a training matrix; Step 2: Perform sparse training; Step 3: Sparsely reconstruct the neutron energy spectrum; Step 4: Solve the neutron spectrum; Step 5: Output the decomposition results.
2. A neutron spectrum analysis method that can use multiple prior information as claimed in claim 1, characterized in that: In step 1, the neutron energy spectrum x is obtained by Monte Carlo simulation. i As a priori information for dictionary training, it forms the training matrix 3. A neutron spectrum analysis method that can use multiple prior information as claimed in claim 2, characterized in that: In step 1, n is the number of neutron energy spectra, ranging from hundreds to hundreds of thousands.
4. A neutron spectrum analysis method that can use multiple prior information as claimed in claim 1, characterized in that: The step 2 comprises: Among them, l(x,D) is the regularized objective function, which is obtained by seeking Minimize D and α i Value, determine the sparse expression matrix D and sparse coefficient vector α i ; λ is the regularization parameter; x i is the i-th neutron energy spectrum in the training matrix X; α i For the corresponding x i , which can minimize the regularization objective function; ||·||1 is the sum of the absolute values of the elements; ||·||2 is the sum of the square values of the elements.
5. The neutron spectrum analysis method using multiple prior information as claimed in claim 1, characterized in that: In step 3, the response matrix M is obtained by detector calibration and Monte Carlo simulation, and the detector count A is obtained by experimental measurement. Based on the response matrix M and the detector count A, the neutron energy spectrum is sparsely reconstructed:
6. A neutron spectrum analysis method that can use multiple prior information as claimed in claim 5, characterized in that: In step 3, the annealing algorithm and the iterative regularization algorithm are used to obtain The minimum value of value.
7. A neutron spectrum analysis method that can use multiple prior information as claimed in claim 1, characterized in that: In step 4, the sparse expression matrix D is used with Solve the neutron spectrum by multiplication; 8. A neutron spectrum analysis method that can use multiple prior information as claimed in claim 7, characterized in that: In step 4 The meaning of the value is the representation of the neutron energy spectrum φ in the sparse expression matrix D.
9. The neutron spectrum analysis method using multiple prior information as claimed in claim 1, characterized in that: The spectrum decomposition result in step 5 is a set of vectors, the number of vector elements is the same as the number of energy group divisions, the value of each element represents the neutron flux measurement result of the energy group, and the sum of each element represents the total neutron flux measurement result.