High repetition frequency neutron source target station beam extraction structure design method based on genetic algorithm

Abstract

The invention belongs to the field of neutron source target station structure design, and particularly relates to a high repetition frequency neutron source target station beam extraction structure design method based on a genetic algorithm. The method comprises the following steps: 1, initializing a genetic algorithm; 2, generating a Monte Carlo input file; 3, carrying out Monte Carlo program simulation transportation; 4, outputting an analog calculation result; 5, calculating the fitness; 6, selecting, crossing and mutating, and reserving an optimal individual; 7, judging whether a convergence condition is met or not; if yes, outputting an optimization parameter corresponding to the optimal individual; and if not, returning to the step 3 until the convergence condition is met. According to the method, through a genetic algorithm and Monte Carlo simulation coupling mode, the inner diameter and the outer diameter of a shielding collimator material are optimized, meanwhile, multi-target optimization is conducted on the inner diameter of a shielding pipeline and the distance between a composite target and the incident end of the shielding collimator, and therefore the overall optimal design of the high-repetition-frequency neutron source target station beam extraction structure is achieved.

Description

Technical Field

[0001] The invention relates to a neutron source target station beam extraction structure design method, in particular to a high repetition rate neutron source target station beam extraction structure design method based on genetic algorithm. Background Art

[0002] Neutron sources play a vital role in the fields of nuclear science, material science and life science. Their applications include the study of irradiation characteristics of new materials, neutron capture therapy, thermal neutron imaging technology and transmutation of nuclear waste. The main advantages of isotope neutron sources (such as 252Cf sources) are small size and simple structure, but energy diffusion is prone to occur, resulting in energy response errors of neutrons. Reactor neutron source devices (such as HFIR) can obtain high-intensity neutron radiation and carry out several experiments at the same time, but due to their high cost and special site selection requirements, they cannot be widely used in industry. Accelerator neutron sources mainly consist of two parts: the part that produces bombarding particles and the target material part. As a neutron source generation device of appropriate scale, accelerator neutron sources play an important role in the practical application of neutrons.

[0003] In order to meet the demand for high energy resolution neutron beams in nuclear detection technology and nuclear data research, it is necessary to build a high repetition rate short pulse neutron source device with high neutron yield, narrow pulse width (1ns level), high repetition rate (40kHz) and high neutron energy resolution. As an important component of the device, the neutron target station needs to be specially designed to meet the requirements of high energy resolution neutron beam extraction and radiation shielding. Currently, commonly used neutron sources, such as spallation neutron sources (such as ISIS, CSNS), Japan's Compact Accelerator Neutron Source (RANS), the United States Low Energy Neutron Source (LENS), Tsinghua University's Small Accelerator Neutron Source (CPHS) and Peking University's Neutron Imaging Facility (PKUNIFTY), are all devices with repetition rates below 100 Hz. Their target station designs all adopt a target-moderator-reflector (TMR) structure to obtain a certain intensity of cold / hot neutron flux. However, the traditional target-moderator-reflector (TMR) structure cannot usually be used when designing a high repetition rate short pulse neutron source target station. This is because the structure uses a large amount of moderator and surrounding reflector materials, which amplifies the influence of scattering in the neutron extraction process and has a negative effect on the extraction of high energy resolution neutron beams.

[0004] The existing design methods for beam extraction structures of high-repetition-rate neutron source target stations mainly focus on the design research of a single collimator, and directly simulate and calculate the material and inner and outer diameters of the collimator based on experience (see LüW H, Guo HP, LüN, et al. Design of a shielding collimator device for a small-angle monoenergetic neutron source [J]. Nuclear Science and Techniques, 2018, 29(8): 119.), or only optimize the single material component of the collimator (see YAN MF, HU G, LIU B, et al. Optimization design of a fast neutron imaging collimator by genetic algorithm [J]. Journal of Instrumentation, 2020, 15(12): P12002.). This design method can usually only achieve local structural optimization, but it is difficult to achieve overall structural optimization. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problem that the existing design methods mainly carry out design research on a single collimator, directly simulate and calculate the material and inner and outer diameters of the collimator based on experience, or only optimize the single material component of the collimator, which usually can only achieve the optimal local structure and is difficult to achieve the optimal overall structure. A high repetition rate neutron source target station beam extraction structure design method based on genetic algorithm is provided.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is:

[0007] A method for designing a beam extraction structure of a high repetition rate neutron source target station based on a genetic algorithm is characterized in that it comprises the following steps:

[0008] Step 1, initialize the genetic algorithm;

[0009] Step 2, randomly generating multiple groups of initial optimization parameters through a genetic algorithm, and using the randomly generated multiple groups of initial optimization parameters to establish multiple high repetition rate neutron source target station beam extraction structures to be optimized, and generating multiple Monte Carlo input files;

[0010] Step 3, based on all current Monte Carlo input files, simulate the neutron, photon, proton and photon coupling transport process through the Monte Carlo program, and record the non-colliding neutron flux, the reacting neutron flux and the total flux around the reacting neutron, so as to obtain the simulation calculation results of all Monte Carlo input files;

[0011] Step 4: Output all simulation calculation results through Monte Carlo program;

[0012] Step 5: Calculate the fitness of all simulation results and obtain their respective fitness values;

[0013] Step 6: Select all high repetition rate neutron source target station beam extraction structures whose fitness values ​​meet preset conditions, and use genetic algorithms to cross-adjust their optimization parameters to generate multiple new high repetition rate neutron source target station beam extraction structures, and retain the high repetition rate neutron source target station beam extraction structure with the highest fitness value obtained in step 5;

[0014] Step 7, determine whether the convergence condition is met; if so, output a set of optimization parameters corresponding to the high repetition rate neutron source target station beam extraction structure with the highest current fitness value; if not, generate a new Monte Carlo input file based on all current high repetition rate neutron source target station beam extraction structures, and return to step 3 until the convergence condition is met.

[0015] Furthermore, in step 5, the fitness is calculated according to the following formula to obtain the fitness value F(x):

[0016] F(x)=-(A·U0(x)-B·ΔU(x)-C·U t (x));

[0017] Where: U0(x), ΔU(x) and U t (x) are the non-colliding neutron flux, the reacting neutron flux and the total flux around the reacting neutrons; A, B and C are weight coefficients.

[0018] Furthermore, the values ​​of A, B and C are:

[0019] A=B=C=1;

[0020] Alternatively, A=B=1 and C=0.

[0021] Furthermore, the optimization parameters include:

[0022] The three-layer structure lengths L1, L2 and L3, and the inner diameters R1, R2 and R3 of the shielding collimator in the beam extraction structure of the high repetition rate neutron source target station;

[0023] The inner diameter R4 of the shielding pipe in the beam extraction structure of the high repetition rate neutron source target station;

[0024] The distance D between the incident end of the shielding collimator and the composite target in the beam extraction structure of the high repetition rate neutron source target station.

[0025] Furthermore, in step 2, the expression of the optimization parameter is:

[0026] X=X(R1,R2,R3,R4,L1,L2,L3,D);

[0027] Where: X is the optimization parameter; R1, R2, R3 and R4 are the inner diameters of the three-layer structure and the inner diameter of the shielding pipe in the shielding collimator, respectively, in cm; L1, L2, L3 are the lengths of the three-layer structure in the shielding collimator along the emission direction, respectively, in cm; D is the distance between the incident end of the shielding collimator and the composite target, in cm;

[0028] R1, R2, R3, R4, L1, L2, L3 and D satisfy the following constraints:

[0029]

[0030] Furthermore, in step 1, initializing the genetic algorithm specifically includes: assigning values ​​to parameters in the genetic algorithm, the parameters including the maximum number of generations;

[0031] In step 7, the convergence condition is that the maximum number of generations is greater than the preset number of generations.

[0032] The beneficial effects of the present invention are:

[0033] The present invention performs multi-parameter optimization on the overall target station by coupling genetic algorithm with Monte Carlo simulation, that is, while optimizing the inner and outer diameters of the shielding collimator material, multi-objective optimization is also performed on the inner diameter of the shielding pipe and the distance between the composite target and the incident end of the shielding collimator. In this way, more influencing factors can be considered at the same time, and the uncertainty of the overall structure caused by artificial decoupling is avoided, thereby achieving the overall optimal design of the beam extraction structure of the high repetition rate neutron source target station. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the beam extraction structure of the existing high repetition rate neutron source target station;

[0035] Figure 2 is a flow chart of an embodiment of the present invention;

[0036] Figure 3 is a comparison diagram of the uncollided neutron flux of the embodiment of the present invention and the existing solution;

[0037] Figure 4 It is a graph of the proportion of uncollided neutrons in the embodiment of the present invention and the existing solution;

[0038] Figure 5 It is a comparison diagram of the total neutron flux of the measurement points of the embodiment of the present invention and the existing solution;

[0039] Figure 6 It is a comparison diagram of the total neutron flux around 5m of the embodiment of the present invention and the existing solution;

[0040] Figure 7 is a schematic diagram of the XY cross-sectional position of the detector in the optimized target station model in an embodiment of the present invention;

[0041] Figure 8 is a schematic diagram of the YZ cross-sectional position of the detector in the optimized target station model in an embodiment of the present invention;

[0042] Fig. 9 is the total dose rate distribution at different positions along the X-axis from the target in the embodiment of the present invention;

[0043] Fig.10 is the neutron dose rate distribution at different positions along the X-axis from the target in the embodiment of the present invention;

[0044] Fig.11 is the photon dose rate distribution at different positions along the X-axis from the target in the embodiment of the present invention;

[0045] Fig.12 It is the spatial distribution of the neutron flux at 5 m along the coordinate axis in the embodiment of the present invention.

[0046] In the figure: 1-composite target, 2-shielded collimator, 3-shielded pipe, 4-detector, 5-vacuum pipe, 6-multi-layer shielding structure. DETAILED DESCRIPTION

[0047] In order to make the purpose, advantages and features of the present invention more clear, the following is a further detailed description of a high repetition rate neutron source target station beam extraction structure design method based on genetic algorithm proposed by the present invention in conjunction with the accompanying drawings and specific embodiments. According to the following specific implementation methods, the advantages and features of the present invention will be more clear.

[0048] The existing high repetition rate short pulse neutron source uses protons with a repetition rate of up to 40kHz, an average incident current of 50μA, and an energy of 3MeV to bombard the beryllium target to produce a high repetition rate neutron beam. The specific geometric model of the beam extraction structure of the existing high repetition rate neutron source target station is as follows: Figure 1As shown, the vacuum chamber is a cylindrical structure with a radius of 15 cm, the inner diameter of the incident pipe is 5 cm, and the right side of the target station uses a multilayer shielding structure 6 composed of 16 cm thick boron-containing polyethylene (BPE) + 4 cm thick lead (Pb) + 16 cm thick boron-containing polyethylene (BPE) + 2 cm thick lead (Pb). The shielding collimator 2 is made of stainless steel (Steel), BPE and Pb. This is because heavy metal materials can preferentially slow down fast neutrons into medium-energy neutrons. After the medium-energy neutrons are slowed down by hydrogen-containing substances, the energy is reduced to the thermal neutron energy region, and further absorption by materials with large thermal neutron absorption cross-sections is required. Therefore, boron-containing polyethylene material (BPE) is selected as the second layer. In the process of interaction between neutrons and matter, gamma rays will be generated, so heavy metal materials (such as Pb) need to be added as the third layer to shield gamma rays. Around the neutron exit pipe, an additional layer of boron carbide (B4C) material with a thickness of 15 cm is added as a shielding pipe 3 to reduce the radiation dose outside the target station. The detector 4 is fixed at the center of the vacuum pipe 5 of the composite target 1. The neutron flux around the measuring point is outside the diameter range of Φ50mm. During the optimization process, the total length of the vacuum pipe 5 (the position of the measuring point) changes with the position of the composite target 1 in the vacuum chamber.

[0049] Genetic Algorithm (GA) is a random optimization algorithm based on Darwin's theory of evolution. It simulates the selection, crossover and mutation processes in nature to find the optimal solution to the problem. The genetic algorithm initializes a set of potential solutions (i.e., population), and then evaluates the quality of each individual according to the fitness function, thereby selecting individuals with higher fitness for genetic recombination and mutation to generate a new generation of individuals. After multiple generations of evolution, the individuals in the population gradually tend to the optimal solution to the problem. Genetic algorithms can provide high-quality solutions to a variety of complex problems, and have demonstrated strong adaptability and robustness in practical applications. They have been widely used in scientific research and engineering practice.

[0050] See also Figure 2 The high repetition rate neutron source target station beam extraction structure design method based on genetic algorithm in this embodiment specifically includes the following steps:

[0051] Step 1: Initialize the parameters of the genetic algorithm; the parameters mainly include population parameters and maximum number of generations, etc. In optimization design, the genetic algorithm is mainly used to initialize and adjust the optimization parameters.

[0052] Step 2: Genetically generate multiple groups of initial optimization parameters, and use the randomly generated multiple groups of initial optimization parameters to establish multiple high repetition rate neutron source target station beam extraction structures to be optimized, and generate multiple Monte Carlo input files.

[0053] Step 3: Based on all current Monte Carlo input files, the Monte Carlo program is used to simulate the neutron, photon, proton and photon coupling transport processes, and the uncollided neutron flux, the reacting neutron flux and the total flux around the reacting neutron are recorded, so as to obtain the simulation calculation results of all Monte Carlo input files.

[0054] Step 4: Output all simulation calculation results through the Monte Carlo program.

[0055] Step 5: Calculate the fitness of all simulation results and obtain their respective fitness values;

[0056] The software simulation uses ENDF / B-VII as the cross-section database. By repeatedly calling the Monte Carlo procedure and the evolutionary strategy of excellent individuals, the GA algorithm can ensure that the optimization design evolves in the direction of convergence. During the optimization process, the change curve of the optimal fitness function value during the iteration process is simultaneously drawn to intuitively obtain the convergence trend. The fitness function of the neutron extraction pipeline of the target station is as follows:

[0057] F(x)=-(A·U0(x)-B·ΔU(x)-C·U t (x)

[0058] X=X(R1,R2,R3,R4,L1,L2,L3,D)

[0059] Where: U0(x), ΔU(x) and U t (x) are the uncollided neutron flux, the reacted neutron flux and the surrounding total flux respectively. X is the parameter to be optimized for the neutron extraction pipeline. Among them, R1, R2, R3 and R4 are the inner diameter of the three-layer structure of the shielding collimator and the inner diameter of the shielding pipeline respectively, in cm, L1, L2, L3 are the length of the three-layer structure of the shielding collimator along the exit direction respectively, in cm, and D represents the distance between the incident end of the shielding collimator and the composite target, in cm. The target parameters are all neutron flux and the order of magnitude of the counts is consistent, so there is no need to normalize. A, B and C are weight coefficients. In the optimization design of this article, two sets of weight coefficients are used as needed: ①A=B=C=1; ②A=B=1 and C=0.

[0060] The overall optimization goal is to obtain the minimum fitness function F(x) and satisfy the following constraints:

[0061]

[0062] The basis for selecting the constraint value is: the minimum inner diameter (R1) of the shielding collimator should be larger than the area where protons and the composite target interact, so as to obtain as many neutron counts in the 0° direction as possible. According to the design idea of ​​the shielding collimator, the inner diameters of the three layers of materials (R1, R2 and R3) should satisfy the law of gradual increase, and the maximum shielding collimator inner diameter (R3) should be constrained according to the actual detection needs (diameter Φ50mm). At the same time, in order to avoid falling into the local optimal solution, the total thickness (L1+L2+L3) of the three-layer structure of the shielding collimator and the inner diameter (R4) of the shielding pipe are limited according to the existing empirical examples. Finally, since the composite target is placed in a 30cm cavity, the maximum distance from the exit pipe to the composite target should be less than 27.5cm, excluding the thickness of the overall composite target structure itself.

[0063] Step 6: Select all high repetition rate neutron source target station beam extraction structures whose fitness values ​​meet the preset conditions, and use the genetic algorithm to cross-adjust their optimization parameters to generate multiple new high repetition rate neutron source target station beam extraction structures, and retain the high repetition rate neutron source target station beam extraction structure with the highest fitness value.

[0064] Step 7, determine whether the convergence condition is met; if so, output a set of optimization parameters corresponding to the high repetition rate neutron source target station beam extraction structure with the highest current fitness value; if not, generate a new Monte Carlo input file based on all current high repetition rate neutron source target station beam extraction structures, and return to step 3 until the convergence condition is met.

[0065] In order to verify the effect of the proposed design method, this paper compares the optimized output results with the simulation results of the enumeration method. Figure 3-6 The optimized parameters and optimized target values ​​are given respectively. "GA1" and "GA2" are obtained by optimizing the fitness function with different weight coefficients respectively, "Ref" is the initial structural parameters obtained by empirical method, and "Enu1", "Enu2" and "Enu3" are optimized by enumeration method based on the initial structure.

[0066] From Table 1 and Figure 3-6 It can be seen that the comprehensive performance of the "GA1" structural scheme is significantly better than that of the other five types of structural combinations. The neutron flux around the measurement point of the "GA1" scheme is 254.80n / cm 2 , which is the smallest compared to the other five schemes. Compared with the initial scheme, the total thickness of the shielding collimator material used in the "GA1" structure was reduced by about 30%, the inner diameter of the vacuum pipe was reduced by 7.17cm, and the proportion of uncollided neutrons was increased by 0.8% while the total neutron flux around was reduced by 3.6%. 2When the total neutron flux and the percentage of uncollided neutron flux of the "GA1" scheme are slightly lower than those of the optimal "Enu1", this is because the inner diameter of the vacuum pipe increases, and the scattering effect of the shielding pipe on the outgoing neutrons is weakened. However, the total thickness of the three layers of materials of its shielding collimator is reduced by about 22cm compared with "Enu1", and the inner diameter of the vacuum pipe is also reduced by about 20.8%, which reduces the construction cost of the vacuum environment. Without considering the neutron flux around the measurement point, the percentage of uncollided neutron flux of the optimization algorithm "GA2" is 0.3% higher than that of "GA1". This is because the shielding pipe is the key to affecting the percentage of uncollided neutrons, and as a vacuum boundary, it has a significant scattering effect on neutrons. From the comparison of the target values ​​of "Enu1" and "Enu2", it can be seen that the distance between the composite target and the shielding collimator will affect the percentage of uncollided neutrons and the value of the surrounding neutron flux at the same time. Optimizing this parameter can not only obtain more uncollided neutron percentages, but also enhance the collimation effect of the device and reduce the number of outgoing neutrons at surrounding angles.

[0067] Table 1 Comparison of optimization parameters of each scheme

[0068]

[0069]

[0070] Taking the scheme "GA1" as the final optimization scheme, a row of point detectors were set at a distance of 10 cm from the outermost shielding structure and at intervals of 20 cm from the composite target structure to 5 m. The distribution of the surrounding dose rate after the optimization of the target station extraction structure was statistically analyzed. The schematic diagram is shown in the figure. Figure 7-8 The statistical results are shown in Figure 9-11 After shielding and collimation, the spatial distribution of the neutron flux at 5m is as follows: Fig.12 shown.

[0071] from Figure 9-11 It can be seen that when the "GA1" scheme is adopted, the total dose rate around the shielding body is lower than 1μSv / h, which meets the dose requirements of radiation protection. When the distance from the composite target is between 0 and 100 cm, the total dose rate peak in the figure is mainly affected by the photon dose rate, and the total dose rate and the photon dose rate both increase first and then decrease. This is because the first 20cm measurement point is located outside the multi-layer shielding material of boron-containing polyethylene and lead, and neutrons and photons are fully absorbed, while the second measurement point is located outside the B4C material. The optimized collimation structure almost only has the neutron shielding material boron-containing polyethylene, which cannot cause further attenuation of the surrounding photon dose rate. At a distance of 100cm to 150cm from the composite target, the photon dose rate and the total dose rate have a slow upward trend. This process is caused by the transition from the presence to absence of the outer shielding layer of the pipeline. Starting from the position of 150cm from the composite target, the total dose rate gradually decreases with the increase of the distance from the composite target. From Fig.12It can be seen that the neutron beam after collimation is mainly distributed between -10cm and 10cm, and maintains a high neutron flux and a relatively uniform distribution within the range of Φ50mm at the center point.

Claims

1. A method for designing the beam extraction structure of a high repetition rate neutron source target station based on genetic algorithm. It is characterized in that The following steps are involved: Step 1, initialize the genetic algorithm; Step 2, randomly generating multiple groups of initial optimization parameters through a genetic algorithm, and using the randomly generated multiple groups of initial optimization parameters to establish multiple high repetition rate neutron source target station beam extraction structures to be optimized, and generating multiple Monte Carlo input files; Step 3, based on all current Monte Carlo input files, simulate the neutron, photon, proton and photon coupling transport process through the Monte Carlo program, and record the non-colliding neutron flux, the reacting neutron flux and the total flux around the reacting neutron, so as to obtain the simulation calculation results of all Monte Carlo input files; Step 4: Output all simulation calculation results through Monte Carlo program; Step 5: Calculate the fitness of all simulation results and obtain their respective fitness values; Step 6, selecting all high repetition rate neutron source target station beam extraction structures whose fitness values ​​meet preset conditions, cross-adjusting their optimization parameters using a genetic algorithm to generate multiple new high repetition rate neutron source target station beam extraction structures, and retaining the high repetition rate neutron source target station beam extraction structure with the highest fitness value; Step 7: Determine whether the convergence condition is met; If yes, then a set of optimization parameters corresponding to the high repetition rate neutron source target station beam extraction structure with the highest fitness value is output; if not, a new Monte Carlo input file is generated according to all current high repetition rate neutron source target station beam extraction structures, and the process returns to step 3 until the convergence condition is met.

2. The method for designing a high repetition rate neutron source target station beam extraction structure based on a genetic algorithm according to claim 1, It is characterized in that In step 5, the fitness is calculated according to the following formula to obtain the fitness value F(x): F(x)=-(A·U 0 (x)-B·ΔU(x)-C·U t (x); Where: U 0 (x), ΔU(x) and U t (x) are the non-colliding neutron flux, the reacting neutron flux and the total flux around the reacting neutrons; A, B and C are weight coefficients.

3. The method for designing a high repetition rate neutron source target station beam extraction structure based on genetic algorithm according to claim 2, It is characterized in that The values ​​of A, B and C are: A=B=C=1; Alternatively, A=B=1 and C=0.

4. The method for designing a high repetition rate neutron source target station beam extraction structure based on a genetic algorithm according to any one of claims 1 to 3, It is characterized in that In step 2, the optimization parameters include: The length L of the three-layer structure of the shielding collimator in the beam extraction structure of the high repetition rate neutron source target station 1 , L 2 and L 3 , and inner diameter R 1 , R 2 and R 3 ; The inner diameter R of the shielding pipe in the beam extraction structure of the high repetition rate neutron source target station 4 ; The distance D between the incident end of the shielding collimator and the composite target in the beam extraction structure of the high repetition rate neutron source target station.

5. The method for designing a high repetition rate neutron source target station beam extraction structure based on a genetic algorithm according to claim 4, It is characterized in that In step 2, the expression of the optimization parameter is: X=X(R 1 ,R 2 ,R 3 ,R 4 ,L 1 ,L 2 ,L 3 ,D); Where: X is the optimization parameter; R 1 , R 2 , R 3 and R 4 are the inner diameters of the three-layer structure and the inner diameter of the shielding pipe in the shielding collimator, in cm; L 1 , L 2 , L 3 are the lengths of the three-layer structure in the shielding collimator, in cm; D is the distance between the incident end of the shielding collimator and the composite target, in cm; The R 1 , R 2 , R 3 , R 4 , L 1 , L 2 , L 3 and D satisfy the following constraints:

6. The method for designing a beam extraction structure of a high repetition rate neutron source target station based on a genetic algorithm according to claim 1, It is characterized in that The specific steps for initializing the genetic algorithm in step 1 are: Parameter assignment in genetic algorithms, including the maximum number of generations; In step 7, the convergence condition is that the maximum number of generations is greater than the preset number of generations.

7. The method for designing a beam extraction structure of a high repetition rate neutron source target station based on a genetic algorithm according to claim 6, Features: The preset number of generations is 100.

Images