A multi-parameter nuclear data adjustment method for fast reactor

Abstract

The application discloses a kind of multi-parameter nuclear data adjustment methods of fast reactor, first, for the multiple critical measurement experiments of existing zero-power experimental device, control rod value measurement experiment and cavitation reactivity measurement experiment, the χ coefficient and IS factor of all experiments are calculated respectively, and with a certain experiment as benchmark experiment, the similarity coefficient of all experiments and benchmark experiment is calculated, according to the screening criterion corresponding to similarity coefficient, χ coefficient and IS factor, select the experiment that can be used for nuclear data adjustment, second, using the measurement results of the experiment screened out, including effective neutron multiplication factor, control rod value and cavitation reactivity, based on Bayes theorem, nuclear data is adjusted, finally, three judgment criteria are used to screen the adjusted nuclear data, and the unreasonable nuclear data is discarded or limited range, finally, the adjusted nuclear data is obtained.The more accurate nuclear data is obtained, so as to improve the core physical calculation precision and confidence of fast reactor.

Description

Technical Field

[0001] This invention relates to the field of nuclear reactor physics analysis and calculation technology, specifically to a method for adjusting multi-parameter nuclear data in a fast reactor. Background Technology

[0002] Nuclear data is one of the key input parameters for reactor core physics calculation programs. However, due to factors such as measuring instruments and methods, current nuclear data inevitably contains uncertainties, especially in the high-energy neutron reaction cross-section, where uncertainties are significant. Since the neutron reaction energies of fast reactors are mostly in the mid-to-high energy range, the uncertainty of nuclear data leads to significant uncertainties in the core neutronics calculation results for fast reactors. With the continuous improvement of reactor core numerical calculation methods, the approximation error of theoretical models has become increasingly smaller, and the uncertainty of nuclear data has become the main source of uncertainty in core neutronics calculation programs. Therefore, adjusting nuclear data in reverse using experimental measurements is an important means to correct nuclear data and improve the agreement between numerical calculations and experimental measurements.

[0003] However, existing nuclear data adjustment work is based on the measurement results of the effective neutron multiplication factor obtained from integral experiments. But adjusting based on a single response can only improve the calculation accuracy of the effective neutron multiplication factor, and it is difficult to improve the calculation accuracy and confidence of other key physical parameters, such as control rod value. Summary of the Invention

[0004] To address the problems existing in the prior art, the present invention aims to provide a method for adjusting multi-parameter nuclear data in fast reactors. By utilizing criticality measurement experiments, control rod value measurement experiments, and cavitation reactivity measurement experiments in zero-power experiments, the nuclear data is adjusted to obtain more accurate nuclear data, thereby improving the accuracy and confidence of core physics calculations in fast reactors.

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

[0006] A method for adjusting multi-parameter nuclear data in a fast reactor first calculates the χ² coefficient and IS factor for multiple critical measurement experiments, control rod value measurement experiments, and cavitation reactivity measurement experiments on an existing zero-power experimental setup. Then, using a specific experiment as a baseline, the similarity coefficient between all experiments and the baseline experiment is calculated. Experiments suitable for nuclear data adjustment are selected based on screening criteria corresponding to the similarity coefficient, χ² coefficient, and IS factor. Next, the nuclear data is adjusted based on Bayes' theorem using the measurement results of the selected experiments, including effective neutron multiplication factor, control rod value, and cavitation reactivity. Finally, the adjusted nuclear data is further screened according to the criteria to obtain the final nuclear data. The method includes the following steps:

[0007] Step 1: For multiple critical measurement experiments, control rod value measurement experiments, and cavitation reactivity measurement experiments of the existing zero-power experimental setup, calculate the χ coefficient and IS factor for each experiment. Then, using a certain experiment as the benchmark experiment, calculate the similarity coefficient between all experiments and the benchmark experiment. Based on the screening criteria corresponding to the similarity coefficient, χ coefficient, and IS factor, select the experiments that can be used for nuclear data adjustment.

[0008] Using experiment j as the baseline, the expression for calculating the similarity coefficient of experiment i relative to experiment j is as follows:

[0009]

[0010] In the formula:

[0011] ρ i,j —The similarity coefficient between experiment i and experiment j characterizes the degree of consistency between the effects of changes in kernel data on experiment i and experiment j;

[0012] S i —The sensitivity coefficient vector of the kernel data in Experiment i;

[0013] S i T —The transpose of the sensitivity coefficient vector of the kernel data in Experiment i;

[0014] S j —The sensitivity coefficient vector of the nuclear data in experiment j;

[0015] S j T —The transpose of the sensitivity coefficient vector of the nuclear data in experiment j;

[0016] M σ —The covariance matrix of the kernel data;

[0017] When ρ i,j =1 indicates that the sources of covariance between the two experiments are completely consistent, when ρ i,j =-1 indicates that the covariance sources of the two experiments are completely opposite. Experiments with high similarity play similar roles in the adjustment process, and the adjusted results of the kernel data are similar. Experiments with low similarity can complement each other, which helps to avoid compensatory effects in the adjustment process.

[0018] The expression for calculating the χ coefficient is:

[0019]

[0020] In the formula:

[0021] χ i—The χ coefficient of experiment i refers to the proportion of the calculation error in one uncertainty, and is used to measure the consistency between the experimental calculation error, calculation uncertainty and measurement uncertainty;

[0022] E i —Experimental measurement results of Experiment i;

[0023] C i —Numerical calculation results of Experiment i;

[0024] M E —The covariance matrix of the experimental measurement results;

[0025] (M E ) i,i —Covariance matrix M of experimental measurement results E The value in the i-th row and i-th column represents the uncertainty of the measurement result in experiment i.

[0026] When χ i If the value is greater than 1, it means that the calculation result exceeds the range of one uncertainty, and there is a possibility that the calculation result of experiment i is inconsistent with the uncertainty.

[0027] The expression for calculating the IS factor is:

[0028]

[0029] In the formula:

[0030] IS i —The IS factor of experiment i, which characterizes the ratio of the uncertainty of the result caused by the kernel data to the uncertainty of the experimental measurement;

[0031] The final screening criteria used in the experiment were:

[0032] (1)ρ i,j The proportion of experiments with values ​​<0 should be greater than 50% of the total number of experiments to avoid compensatory effects;

[0033] (2)χ i ≤1, ensuring the calculation result is within one degree of uncertainty;

[0034] (3)IS i ≥1, meaning that the uncertainty of the nuclear data in the calculation results is greater than the experimental uncertainty, thus ensuring that experiment i can effectively adjust the nuclear data;

[0035] Step 2: Using the experimental measurement results of the three categories of effective neutron multiplication factor, control rod value, and cavitation reactivity from the selected experiments, adjust the nuclear data based on Bayes' theorem;

[0036] Using the kernel data covariance data provided in the evaluation kernel database, the probability density function expression for the kernel data can be written as follows:

[0037]

[0038] In the formula:

[0039] σ — kernel data;

[0040] σ0 — The nominal value of the kernel data given in the kernel database;

[0041] p(σ) — the probability density function of kernel data;

[0042] Similarly, the probability density function of the experimental measurements can be written as:

[0043]

[0044] In the formula:

[0045] E – Experimental measurement value;

[0046] E0—the nominal value measured in the actual experiment;

[0047] p(E) — the probability density function of the experimental measurement;

[0048] According to Bayes' theorem, we can obtain the conditional probability density function expression for kernel data when the experiment represents the true value:

[0049]

[0050] In the formula:

[0051] p(σ|E)——Conditional probability density function of kernel data when the experiment is the true value;

[0052] p(E|σ)——Conditional probability density function of experimental results when the kernel data are true values;

[0053] Since the experimental measurements are distributed around the results of numerical calculations using kernel data as input parameters, p(E|σ) can be written as follows:

[0054]

[0055] In the formula:

[0056] C(σ) — The result of a numerical calculation program that takes kernel data as input parameters;

[0057] Therefore, the conditional probability density function of the kernel data has the following expression:

[0058]

[0059] For the conditional probability density function p(σ|E) of the kernel data to reach its maximum value, the value of the kernel data σ should minimize the absolute value of the exponential term in equation (8), that is, to find the kernel data σ such that J in the following equation reaches its minimum value:

[0060]

[0061] In the formula:

[0062] J — Loss function;

[0063] According to the extreme value theorem, the loss function reaches its minimum value where the derivative with respect to the kernel data is zero. The posterior kernel data at this point is:

[0064] σ post =σ0+M σ S[S T M σ S+M E ] -1 [EC(σ0)] (10)

[0065] In the formula:

[0066] S—The sensitivity coefficient matrix of the experiment to nuclear data;

[0067] σ post —Adjusted post-verification data;

[0068] Meanwhile, based on the Woodbury matrix identity, the expression for the posterior covariance matrix is:

[0069]

[0070] In the formula:

[0071] M σ,post —The adjusted posterior covariance matrix, whose elements are the adjusted values ​​of the kernel data covariance obtained by combining experimental measurement results;

[0072] In summary, the adjusted posterior kernel data σ is finally obtained through formulas (10) and (11). post With the adjusted posterior covariance matrix M σ,post The minimum adjustment amount of the loss function cannot be obtained through a single adjustment calculation. Therefore, iterative calculation using equations (10) and (11) is required. The calculation formula for the i-th iteration is as follows:

[0073]

[0074]

[0075] In the formula:

[0076] σ i —The kernel data value obtained after the i-th iteration adjustment;

[0077] σ i-1 —The kernel data value obtained after the (i-1)th iteration adjustment;

[0078] —The covariance matrix of the kernel data obtained after the i-th iteration adjustment;

[0079] —The covariance matrix of the kernel data obtained after the (i-1)th iteration adjustment;

[0080] Step 3: Use three judgment criteria to filter the adjusted kernel data, discard or limit the range of unreasonable kernel data, and finally obtain the adjusted kernel data;

[0081] The adjusted nuclear data were screened according to the following criteria:

[0082] 1) The maximum adjustment amount is limited to one time the uncertainty of the initial kernel data; adjustments exceeding this limit are made according to the maximum adjustment amount.

[0083] 2) The adjusted kernel data should be non-negative; therefore, if a negative value appears, the original data before adjustment should be restored.

[0084] 3) Do not change the zero elements in the original kernel data covariance matrix.

[0085] Compared with the prior art, the present invention has the following advantages:

[0086] This invention utilizes multiple types of experimental data during the nuclear data adjustment process, including criticality measurement experiments, control rod value measurement experiments, and cavitation reactivity measurement experiments from zero-power experiments. This broadens the applicability of the adjusted nuclear data and improves the calculation accuracy of critical characteristics, control rod value, and cavitation reactivity. Simultaneously, a comprehensive screening criterion is employed to screen both the experimental data and the adjusted nuclear data, thereby avoiding compensation effects or overfitting and ensuring the reliability of the nuclear data adjustment results. Attached Figure Description

[0087] Figure 1 This is a flowchart of a multi-parameter core data adjustment method for a fast heap.

[0088] Figure 2 The similarity coefficient of the experiments.

[0089] Figure 3 χ² is the coefficient of the experiment.

[0090] Figure 4 The IS factor for the experiment.

[0091] Figure 5 The results are adjusted for the inelastic scattering cross section data of nuclide U-238. Detailed Implementation

[0092] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0093] like Figure 1 As shown, the present invention provides a method for adjusting multi-parameter core data of a fast heap, comprising the following steps:

[0094] Step 1: For existing zero-power experimental setups, such as the ZPPR fast neutron energy spectrum experimental setup, there are 47 critical measurement experiments, control rod value measurement experiments, and cavitation reactivity measurement experiments, as shown in Tables 1, 2, and 3 below.

[0095] Table 1 Critical Measurement Experiment

[0096]

[0097] Table 2. Experiment on the Measurement of Control Rod Value

[0098]

[0099]

[0100] Table 3. Measurement Experiment of Vapor Reactivity

[0101]

[0102] The χ² coefficient and IS factor of each of the 47 experiments were calculated. Taking experiment number 1 as the baseline experiment, the similarity coefficients of all experiments with the baseline experiment were calculated. Finally, 47 χ² coefficients, 47 IS factors and 47 similarity coefficients were obtained (the similarity coefficient of the baseline experiment itself was 1). Experiments that can be used for nuclear data adjustment were selected according to the screening criteria corresponding to the similarity coefficients, χ² coefficients and IS factors.

[0103] Using experiment j as the baseline, the similarity coefficient of experiment i relative to experiment j is calculated as follows. If experiment number 1 is taken as the baseline experiment, then the similarity coefficient of all experiments relative to experiment 1 is calculated, and the similarity coefficient of the baseline experiment relative to itself is 1.

[0104]

[0105] In the formula:

[0106] ρ i,j —The similarity coefficient between experiment i and experiment j characterizes the degree of consistency between the effects of changes in kernel data on experiment i and experiment j;

[0107] S i —The sensitivity coefficient vector of the kernel data in Experiment i;

[0108] S i T —The transpose of the sensitivity coefficient vector of the kernel data in Experiment i;

[0109] S j —The sensitivity coefficient vector of the nuclear data in experiment j;

[0110] S j T —The transpose of the sensitivity coefficient vector of the nuclear data in experiment j;

[0111] M σ —The covariance matrix of the kernel data;

[0112] When ρ i,j =1 indicates that the sources of covariance between the two experiments are completely consistent, when ρ i,j =-1 indicates that the covariance sources of the two experiments are completely opposite. Experiments with high similarity play similar roles in the adjustment process, and the adjusted kernel data results are similar. Experiments with low similarity can complement each other, which helps to avoid compensatory effects in the adjustment process. The similarity coefficients of the 47 experiments in Tables 1 to 3 are attached. Figure 2 As shown.

[0113] The expression for calculating the χ coefficient is:

[0114]

[0115] In the formula:

[0116] χ i —The χ coefficient of experiment i refers to the proportion of the calculation error in one uncertainty, and is used to measure the consistency between the experimental calculation error, calculation uncertainty and measurement uncertainty;

[0117] E i —Experimental measurement results of Experiment i;

[0118] C i —Numerical calculation results of Experiment i;

[0119] M E —The covariance matrix of the experimental measurement results;

[0120] (M E ) i,i —Covariance matrix M of experimental measurement results E The value in the i-th row and i-th column represents the uncertainty of the measurement result in experiment i.

[0121] When χ i If the value is greater than 1, it indicates that the calculated result exceeds the range of one unit of uncertainty. Therefore, experiment i may have a discrepancy between the calculated result and the uncertainty, and should not be used. The χ² coefficients for the 47 experiments in Tables 1 to 3 are attached. Figure 3 As shown.

[0122] The expression for calculating the IS factor is:

[0123]

[0124] In the formula:

[0125] IS i —The IS factor of experiment i, which characterizes the ratio of the uncertainty of the result caused by the kernel data to the uncertainty of the experimental measurement;

[0126] For example, the IS factors for the 47 experiments in Tables 1 to 3 are attached. Figure 4 As shown.

[0127] The final screening criteria used in the experiment were:

[0128] (1)ρ i,j The proportion of experiments with values ​​<0 should be greater than 50% of the total number of experiments to avoid compensatory effects;

[0129] (2)χ i ≤1, ensuring the calculation result is within one degree of uncertainty;

[0130] (3)IS i ≥1, meaning that the uncertainty of the nuclear data in the calculation results is greater than the experimental uncertainty, thus ensuring that experiment i can effectively adjust the nuclear data;

[0131] Using the above experimental screening criteria, experiments suitable for subsequent nuclear data adjustments were selected, ultimately narrowing down the list to 20 experiments from 47 (see appendix). Figure 3 (The experiment shown in the black box).

[0132] Step 2: Using the experimental measurements of effective neutron multiplication factor, control rod value, and cavitation reactivity from the selected experiments, nuclear data are adjusted based on Bayes' theorem. For the 20 experiments selected in Step 1, the nuclear data are adjusted using the corresponding measurements of effective neutron multiplication factor, control rod value, and cavitation reactivity.

[0133] Using the kernel data covariance data provided in the evaluation kernel database, the probability density function expression for the kernel data can be written as follows:

[0134]

[0135] In the formula:

[0136] σ — kernel data;

[0137] σ0 — The nominal value of the kernel data given in the kernel database;

[0138] p(σ) — the probability density function of kernel data;

[0139] Similarly, the probability density function of the experimental measurements can be written as:

[0140]

[0141] In the formula:

[0142] E – Experimental measurement value;

[0143] E0—the nominal value measured in the actual experiment;

[0144] p(E) — the probability density function of the experimental measurement;

[0145] According to Bayes' theorem, we can obtain the conditional probability density function expression for kernel data when the experiment represents the true value:

[0146]

[0147] In the formula:

[0148] p(σ|E)——Conditional probability density function of kernel data when the experiment is the true value;

[0149] p(E|σ)——Conditional probability density function of experimental results when the kernel data are true values;

[0150] Since the experimental measurements are distributed around the results of numerical calculations using kernel data as input parameters, p(E|σ) can be written as follows:

[0151]

[0152] In the formula:

[0153] C(σ) — The result of a numerical calculation program that takes kernel data as input parameters;

[0154] Therefore, the conditional probability density function of the kernel data has the following expression:

[0155]

[0156] For the conditional probability density function p(σ|E) of the kernel data to reach its maximum value, then the value of the kernel data σ should be...

[0157] To minimize the absolute value of the exponential term in equation (8), that is, to find the kernel data σ such that J in the following equation reaches its minimum value:

[0158]

[0159] In the formula:

[0160] J — Loss function;

[0161] According to the extreme value theorem, the loss function reaches its minimum value where the derivative with respect to the kernel data is zero. The posterior kernel data at this point is:

[0162] σ post =σ0+M σ S[S T M σ S+M E ] -1 [EC(σ0)] (10)

[0163] In the formula:

[0164] S—The sensitivity coefficient matrix of the experiment to nuclear data;

[0165] σ post — The adjusted posterior nuclear data, that is, the adjusted value of the nuclear data finally obtained by combining the experimental measurement results;

[0166] Meanwhile, based on the Woodbury matrix identity, the expression for the posterior covariance matrix is:

[0167]

[0168] In the formula:

[0169] M σ,post —The posterior covariance matrix, whose elements are the adjusted values ​​of the kernel data covariance obtained by combining experimental measurement results;

[0170] In summary, the adjusted posterior kernel data σ is finally obtained through formulas (10) and (11). post With the adjusted posterior covariance matrix M σ,post The minimum adjustment amount of the loss function cannot be obtained through a single adjustment calculation. Therefore, iterative calculation using equations (10) and (11) is required. The calculation formula for the i-th iteration is as follows:

[0171]

[0172]

[0173] In the formula:

[0174] σ i —The kernel data value obtained after the i-th iteration adjustment;

[0175] σ i-1—The kernel data value obtained after the (i-1)th iteration adjustment;

[0176] —The covariance matrix of the kernel data obtained after the i-th iteration adjustment;

[0177] —The covariance matrix of the kernel data obtained after the (i-1)th iteration adjustment;

[0178] Step 3: Use three judgment criteria to filter the adjusted kernel data, discard or limit the range of unreasonable kernel data, and finally obtain the adjusted kernel data;

[0179] To ensure the reliability of the adjusted nuclear data and avoid compensation effects or overfitting, the adjusted nuclear data were screened according to the following criteria:

[0180] 1) The maximum value of the adjustment is limited to one time the uncertainty of the initial nuclear data. If the value exceeds the limit, the adjustment shall be made according to the maximum value. For example, if the uncertainty of the initial neutron fission reaction cross section is 1 bar, and the calculated adjustment is 1.5 bar, then according to this rule, the neutron fission reaction cross section should only be increased by 1 bar.

[0181] 2) The nuclear data should be non-negative after adjustment. Therefore, if a negative value appears, the original data before adjustment should be restored. For example, if the initial neutron scattering reaction cross section is 1 bar and the calculated adjustment amount is -1.5 bar, if the adjustment is made, the neutron scattering reaction cross section will become -0.5 bar. According to this rule, the neutron scattering reaction cross section should remain at the original 1 bar.

[0182] 3) Do not change the zero elements in the original kernel covariance matrix. The original covariance matrix contains several zero elements; according to this rule, these values ​​will not be adjusted.

[0183] After performing steps 2 and 3 using the experiments selected in step 1, the results of adjusting the inelastic scattering cross section of nuclide U-238 are shown in the appendix. Figure 5 As shown, the uncertainty of the nuclear data is significantly reduced after the adjustment ranges from -30% to 15%.

Claims

1. A method for adjusting multi-parameter nuclear data of a fast reactor, characterized in that: Includes the following steps: Step 1: For the multiple critical measurement experiments, control rod value measurement experiments, and cavitation reactivity measurement experiments of the existing zero-power experimental setup, calculate the respective values ​​of all experiments. coefficients and sum IS Factors, and using a certain experiment as the baseline experiment, calculate the similarity coefficient between all experiments and the baseline experiment, and based on the similarity coefficient, coefficients and sum IS The screening criteria corresponding to the factors were used to select experiments that could be used for nuclear data adjustment; Based on the experiment j , the experiment i , the similarity coefficient relative to the experiment j The expression for calculating the similarity coefficient is: (1) In the formula: — Experiment i Compared to the experiment j The similarity coefficient characterizes the effect of changes in nuclear data on experiments. i and experiment j The degree of consistency of the impact; - Experiment i a sensitivity coefficient vector of the core data; - Experiment i the transpose of the sensitivity coefficient vector of the core data; - Experiment j a sensitivity coefficient vector of the core data; - Experiment j the transpose of the sensitivity coefficient vector of the core data; - a covariance matrix of the nuclear data; when This indicates that the sources of covariance between the two experiments are completely consistent, when This indicates that the covariance sources of two experiments are completely opposite, and the high degree of similarity (i.e., the similarity coefficient is close to 1) means that the kernel data adjustment results are similar, that is, the kernel data adjustment amounts tend to be more consistent. Experiments with low similarity, i.e., similarity coefficients close to -1, can complement each other, which helps to avoid compensatory effects during the adjustment process. The expression for calculating the coefficient is: (2) In the formula: — Experiment i of The coefficient refers to the proportion of the calculation error in one unit of uncertainty, and is used to measure the degree of consistency between experimental calculation error, calculation uncertainty, and measurement uncertainty; - experimental i measurements; - experimental i numerical results; - a covariance matrix of the experimental measurement results; — Covariance matrix of experimental measurement results No. i Line 1 i The values ​​in the column, i.e., the experiment i Uncertainty of measurement results; When the calculation result exceeds the range of one uncertainty degree, it is indicated that the experiment i may have a possibility that the calculation result is inconsistent with the uncertainty degree. IS The factor calculation expression is: (3) In the formula: — Experiment i of IS The factor characterizes the ratio of the uncertainty of the result caused by the kernel data to the uncertainty of the experimental measurement; The final screening criteria used in the experiment were: (1) The proportion of experimental numbers to total experimental numbers should be greater than 50% to avoid compensatory effects. (2) , to ensure that the result of the calculation is within one unit of uncertainty; (3) i.e. the uncertainty of the calculation result of the nuclear data is greater than the experimental uncertainty, thereby ensuring that the experimental i nuclear data can be effectively adjusted; Step 2: Using the experimental measurement results of the three categories of effective neutron multiplication factor, control rod value, and cavitation reactivity from the selected experiments, adjust the nuclear data based on Bayes' theorem; Using the kernel data covariance data provided in the evaluation kernel database, the probability density function expression for the kernel data can be written as follows: (4) In the formula: - nuclear data; - nominal values of nuclear data given in nuclear databases; - a probability density function of the nuclear data; Similarly, the probability density function of the experimental measurements can be written as: (5) In the formula: - experimental measurements; - the nominal value of the actual experimental measurement; - a probability density function of the experimental measurements; According to Bayes' theorem, we can obtain the conditional probability density function expression for kernel data when the experiment represents the true value: (6) In the formula: - the conditional probability density function of the nuclear data given that the experiment is the true value; - the conditional probability density function of the experimental results given that the nuclear data are the true values; Because the experimental measurements are distributed around the calculation results of a numerical calculation program that uses kernel data as input parameters, therefore Write it as follows: (7) In the formula: - the results of a numerical calculation program with the core data as input parameters; Therefore, the conditional probability density function of the kernel data has the following expression: (8) At this point, the conditional probability density function of the kernel data To obtain the maximum value, then the kernel data... The value of should minimize the absolute value of the exponent term in equation (8), that is, to find the kernel data. Make the following formula Reaching the minimum value: (9) In the formula: — a loss function; According to the extreme value theorem, the loss function reaches its minimum value where the derivative with respect to the kernel data is zero. The posterior kernel data at this point is: (10) In the formula: - the sensitivity coefficient matrix of the experiment to the nuclear data; —Adjusted post-verification data; Meanwhile, based on the Woodbury matrix identity, the expression for the posterior covariance matrix is: (11) In the formula: —The adjusted posterior covariance matrix, whose elements are the adjusted values ​​of the kernel data covariance obtained by combining experimental measurement results; In summary, the adjusted posterior verification data are finally obtained through formulas (10) and (11). With the adjusted posterior covariance matrix The minimum adjustment amount for the loss function cannot be obtained through a single adjustment calculation; therefore, iterative calculation using equations (10) and (11) is required. i The calculation formula for the next iteration is as follows: (12) (13) In the formula: - the first iteration adjustment of the core data value; and i the second iteration adjustment of the core data value. ——No. i The kernel data value obtained after -1 iterations of adjustment; ——No. i The covariance matrix of the kernel data obtained after the next iteration adjustment; -1 iteration adjustment of the core data; and i -1 iteration adjustment of the core data; and Step 3: Use three judgment criteria to filter the adjusted kernel data, discard or limit the range of unreasonable kernel data, and finally obtain the adjusted kernel data; The adjusted nuclear data were screened according to the following criteria: 1) The maximum adjustment amount is limited to one time the uncertainty of the initial kernel data; adjustments exceeding this limit are made according to the maximum adjustment amount. 2) The adjusted kernel data should be non-negative; therefore, if a negative value appears, the original data before adjustment should be restored. 3) Do not change the zero elements in the original kernel data covariance matrix.

Images