Search and optimization method for refueling loading scheme

Abstract

The present invention relates to a method for searching and optimizing a refueling loading scheme, which includes the following steps: S1: Generate several common new component layouts based on several historical refueling schemes; and obtain the information of old components in the historical refueling schemes to generate an old component database; S2: Select old components from the old component database according to the common new component layouts, and randomly load them into the common new component layouts to generate K loading schemes, and select K1 kinds from the K loading schemes to form training samples; where K1 < K; S3: Use the training samples to train and generate a prediction model library that meets the accuracy requirements; the prediction model library contains several key parameter prediction models; S4: According to the refueling notice information of the current unit, select the key parameter prediction model corresponding to the refueling notice information from the prediction model library, and combine it with the fuel assembly information of the current unit to generate a feasible loading scheme. It has the advantages of saving the search time of the refueling scheme, shortening the refueling design time, effectively improving the safety and fuel economy of the power station, etc.

Description

Technical Field

[0001] The present invention relates to the field of nuclear fuel assembly refueling, and in particular to a refueling loading scheme search and optimization method. Background Art

[0002] At present, there are dozens of large pressurized water reactors in operation in my country. Each pressurized water reactor needs to refuel its nuclear fuel assemblies every 12-18 months. In theory, the search combination of refueling schemes can reach 10 13 The above complexity far exceeds Go.

[0003] Currently, reactor core loading plans are determined by refueling engineers through manual experience and trial and error. The different plans selected by different engineers are related to user effects. Generally, searching for a mature plan requires one person per day, and multiple plans are often searched to find the optimal one. In particular, emergency refueling requires even more time to search for a suitable refueling plan. In recent years, many researchers and scholars have attempted to optimize this refueling search using mathematical methods, but no method has yet been found that is suitable for engineering applications. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for searching and optimizing a material replacement loading scheme.

[0005] The technical solution adopted by the present invention to solve the technical problem is to construct a material replacement loading plan search optimization method, including the following steps:

[0006] S1: Generate several commonly used new component layouts based on several historical material replacement plans; and obtain old component information in the historical material replacement plans to generate an old component database;

[0007] S2: According to the commonly used new component layout, select the old components from the old component database and randomly load them into the commonly used new component layout to generate K loading schemes, and select K1 from the K loading schemes to form training samples; where K1 <K;

[0008] S3: Using the training samples to train and generate a prediction model library that meets the accuracy requirements; the prediction model library contains several key parameter prediction models;

[0009] S4: According to the refueling notification information of the current unit, the key parameter prediction model corresponding to the refueling notification information is selected from the prediction model library, and combined with the fuel assembly information of the current unit to generate a feasible loading plan.

[0010] Preferably, in step S1, based on the layout characteristics of the fuel assemblies of the core, a local area of ​​the core is selected as the area for the common new assembly layout.

[0011] Preferably, the local area is a 1 / 8 core area of ​​the core, and eight 1 / 8 core areas are symmetrically arranged to form a complete area of ​​the core.

[0012] Preferably, the old component information includes old position information of each old component in the historical refueling plan in which it is located and in the core;

[0013] In step S2, according to the position loading rule, the old components matching the old position information are selected from the old component database and randomly loaded into the commonly used new component layout.

[0014] Preferably, the position loading rules include placing the old components that were at the 1 / 4 position in the historical material replacement plan into the central component position and the 1 / 4 position; and placing the old components that were at the 1 / 8 position in the historical material replacement plan into the central component position, the 1 / 4 position, and the 1 / 8 position.

[0015] Preferably, the step S2 further includes: selecting K2 loading schemes from the K loading schemes to form test samples; wherein K1+K2=K.

[0016] Preferably, in step S3, deep neural network training is performed on the training samples with the same number of new components to obtain the key parameter prediction model corresponding to the same number of components.

[0017] Preferably, the deep neural network includes: an input layer, a network layer and an output layer;

[0018] The input layer is used to input kinf eigenvalues, gd eigenvalues, enrichment eigenvalues ​​and four-corner burnup eigenvalues; wherein, n input nodes;

[0019] The network structure layer includes 18 network layers, with 7 nodes in each layer. n Each network layer uses full connection, Sigmoid activation function, L2 regularization and Batch Normalization;

[0020] The output layer is used to output parameter prediction values ​​to determine whether the requirements are met, and the parameter prediction values ​​include the maximum enthalpy rise factor and the cycle length;

[0021] in, n is the number of old components placed in the training sample.

[0022] Preferably, step S4 includes the following steps:

[0023] S4.1: Selecting the corresponding key parameter prediction model based on the number of new components to be replaced in the refueling notification information; and performing calculations based on the cycle length in the refueling notification information and the selected key parameter prediction model to select appropriate available old components for the current unit;

[0024] S4.2: Based on the selected available old components, a genetic algorithm is used to generate several loading schemes to be verified, and the cycle length and maximum enthalpy rise factor of each loading scheme to be verified are predicted using the selected key parameter prediction model to search for a loading scheme that requires all core refueling parameter calculations.

[0025] Preferably, the step S4.1 includes the following steps:

[0026] S4.1.1: Obtain information on all available old components of the current unit;

[0027] S4.1.2: Select the 1 / 4 position component, the 1 / 8 position component, and the center component according to the component reactivity kinf from large to small, and load them into the key parameter prediction model;

[0028] S4.1.3: Calculate the cycle length of the key parameter prediction model after loading is completed;

[0029] S4.1.4: Determine whether the calculated cycle length meets the material change notification requirements; if so, end the process; otherwise, reselect components with smaller reactivity kinf from the remaining components of the current unit to replace components with larger reactivity kinf from the selected components until the cycle length requirements are met and the cycle ends.

[0030] Preferably, the step S4.2 includes the following steps:

[0031] S4.2.1: Random Generation m An initial population is generated by crossover and mutation using a set of digitally encoded chromosomes to simulate each loading scenario. 2*m offspring;

[0032] S4.2.2: Use the trained prediction model to predict the key parameters 2*m The maximum enthalpy rise factor and cycle length of each loading scheme are searched for a loading scheme whose maximum enthalpy rise factor and cycle length meet the requirements, and the loading scheme is saved;

[0033] S4.2.3: Simulate chromosome crossover and / or mutation operations to eliminate individuals with low fitness generation by generation and generate new individuals with high fitness according to the law of natural selection where the fittest survive. The fitness includes cycle length and maximum enthalpy rise factor.

[0034] S4.2.4: The search ends when the number of saved loading solutions is greater than m or after k iterations;

[0035] S4.2.5: Select the best 0.01*m loading schemes and recalculate the maximum enthalpy rise factor and cycle length. Select the best 0.001*m loading schemes from these and calculate all core refueling parameters.

[0036] Among them, m is the initial population size of the loading scheme, and k is the number of iterations of genetic search.

[0037] Preferably, in step S4.2.3, the simulated chromosome crossover and / or mutation operation is only performed on components within the same region.

[0038] The implementation of the present invention has the following beneficial effects: a prediction model library is generated based on historical refueling plan training, and based on the refueling notification information of the current unit, a key parameter prediction model corresponding to the refueling notification information is selected from the prediction model library, and combined with the fuel assembly information of the current unit, a feasible loading plan is generated, avoiding the time-consuming and complex problem of manual search, and has the advantages of saving refueling plan search time, shortening refueling design time, and effectively improving the safety and fuel economy of the power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0040] Figure 1 1 is a flow chart of the method for searching and optimizing the material replacement and loading scheme of the present invention;

[0041] Figure 2 is a schematic diagram of the layout of a core of an embodiment;

[0042] Figure 3 yes Figure 2 Schematic diagram of filling 1 / 8 of the core area with new components;

[0043] Figure 4 is a schematic diagram of the location classification of old components of a reactor core according to an embodiment;

[0044] Figure 5 1 is a flow chart of an embodiment of step S4.1 of the method for searching and optimizing a material replacement loading plan according to the present invention;

[0045] Figure 6 This is a flow chart of an embodiment of step S4.2 of the method for searching and optimizing a material replacement loading plan according to the present invention;

[0046] Figure 7 A schematic diagram comparing the maximum enthalpy rise factors of a loading scheme generated by the material change loading scheme search optimization method of the present invention and a historical cycle scheme;

[0047] Figure 8 Schematic diagram comparing the cycle lengths of a loading scheme generated by the material change loading scheme search optimization method of the present invention and a historical cycle scheme. DETAILED DESCRIPTION

[0048] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, the specific embodiments of the present invention are now described in detail with reference to the accompanying drawings. In the following description, it should be understood that the directions or positional relationships indicated by "front", "back", "up", "down", "left", "right", "longitudinal", "horizontal", "vertical", "horizontal", "top", "bottom", "inside", "outside", "head", "tail", etc. are based on the directions or positional relationships shown in the accompanying drawings and are constructed and operated in specific directions. They are only for the convenience of describing the technical solution and do not indicate that the devices or components referred to must have specific directions. Therefore, they should not be understood as limiting the present invention.

[0049] It should also be noted that, unless otherwise clearly specified and limited, terms such as "installed", "connected", "connected", "fixed", and "set" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integrated connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. When an element is referred to as being "on" or "under" another element, the element can be "directly" or "indirectly" located on the other element, or there may be one or more intervening elements. The terms "first", "second", "third", etc. are only for the convenience of describing the present technical solution, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0050] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0051] The present invention proposes a method for searching and optimizing refueling and loading schemes suitable for a reactor core. This method uses fuel assembly parameters as input, constructs a neural network model to predict parameters such as cycle length and maximum enthalpy rise factor, and uses a genetic algorithm to automatically search for refueling schemes. This method offers at least the following advantages: 1. It saves time searching for refueling schemes and shortens refueling design time. Generally, it takes 5 minutes to calculate the maximum enthalpy rise factor and cycle length for a loading scheme. Using this method, searching for 10,000 loading schemes takes only 1.5 seconds (software implementation), and the total loading scheme search takes about 0.5 hours (parallel), while manual work generally takes more than one man-day. To find a better solution, multiple searches are usually performed in parallel.

[0052] 2. A better solution can be searched. The characteristic parameters of the core loading solution searched by this method are generally better than those searched manually, which effectively improves the safety and fuel economy of the power plant.

[0053] 3. Compared to the feasible solution, the optimized fuel replacement solution may be safer and more economical. For example, it may have a smaller power peak and a longer cycle length, which ideally could result in a fuel saving of four groups.

[0054] 4. The analysis process does not require human intervention, reducing human errors.

[0055] like Figure 1 As shown, in one embodiment of the material change loading scheme search optimization method of the present invention, it includes step S1: generating several commonly used new component layouts according to several historical material change schemes; and obtaining old component information in the historical material change schemes to generate an old component database.

[0056] The old component information includes old position information of each old component in its historical refueling plan and its position in the core.

[0057] Currently, the commonly used new assembly layouts in power plant refueling designs are considered different based on the number of new assemblies, their placement, and the number of gadolinium rods. To reduce the complexity of the search, local areas of the core were selected as the common new assembly layout areas based on the layout characteristics of the core fuel assemblies.

[0058] In this embodiment, the local area is selected from the 1 / 8 core area of ​​the core, such as Figure 2 As shown, the new components can be arranged in this 1 / 8 core area to secure their position. Eight 1 / 8 core areas are symmetrically arranged to form a complete core area. It is understood that the entire core area, or a 1 / 4 or other partial area, can also be used for arrangement, depending on requirements.

[0059] In the case of determining the layout, such as Figure 3Taking the new material layout shown as an example, without considering rotational symmetry, there are over 4 million possible layout options. Experience-based consolidation reduces the number of possible layout combinations. After determining the placement of the old components, to ensure better adaptability of the sample training model and to cover as many component types as possible, this example uses all available old components from units with a wide variety of component types.

[0060] S2: Based on the commonly used new component layout, select old components from the old component database and load them into the commonly used new component layout to generate K loading schemes. Then, import the approximately K randomly generated loading schemes into the core design software for calculation to obtain key parameters such as cycle length and maximum enthalpy rise factor for the corresponding loading scheme.

[0061] In order to be compatible with time and model accuracy, K1 loading schemes are selected from K loading schemes to form training samples; <K。

[0062] Furthermore, K2 loading schemes can be selected from the K loading schemes to form a test sample, where K1 + K2 = K. In one embodiment, K1 = 0.8 * K; K2 = 0.2 * K. Of course, the specific ratio can also be adjusted according to actual needs.

[0063] The loading of old components can be performed according to the position loading rules. Old components with matching old position information are selected from the old component database and randomly loaded into the commonly used new component layout to generate K loading schemes.

[0064] The position loading rules include placing the old components at the 1 / 4 position in the historical material replacement plan in the center component position or the 1 / 4 position; and placing the old components at the 1 / 8 position in the historical material replacement plan in the center component position, the 1 / 4 position, or the 1 / 8 position. Figure 4 shown.

[0065] S3: Use the training samples to generate a prediction model library that meets accuracy requirements; the prediction model library contains several key parameter prediction models. This step uses the generated training samples (0.8*K loading scenarios) to train a deep neural network. The constructed neural network key parameter prediction model is capable of evaluating the cycle lengths and maximum enthalpy rise factors of more than 10,000 loading scenarios per second.

[0066] Because the maximum enthalpy rise factor occurs at different burnup conditions at the beginning and middle of its lifecycle, its variability complicates neural network training. Using different numbers of new components as samples, it's impossible to train a suitable model using the component characteristics of the 1 / 8 core region as input. In this example, a deep neural network is trained on training samples with the same number of new components to obtain a key parameter prediction model corresponding to the same number of components.

[0067] When the component features of the 1 / 8 core area are used as input, a deep neural network (DNN) is used. Each layer of the network adopts full connection, sigmoid activation function, L2 regularization and batch normalization (BN). By adjusting the input features and hidden layer structure, the training and testing results of the model are within an acceptable range.

[0068] Deep neural network includes: input layer, network layer and output layer;

[0069] The input layer is used to input the kinf (reactivity) eigenvalue, gd (gadolinium rod number) eigenvalue, enrichment eigenvalue, and four-corner burnup eigenvalue; each layer has n input nodes.

[0070] The network structure layer includes 18 network layers, with 7*n nodes in each layer. Each layer adopts full connection, Sigmoid activation function, L2 regularization and Batch Normalization (BN).

[0071] The output layer is used to output parameter prediction values ​​to determine whether they meet the requirements. The parameter prediction values ​​include the maximum enthalpy rise factor and the cycle length.

[0072] in, n is the number of old components placed in the training sample.

[0073] S4: Based on the refueling notification information of the current unit, a key parameter prediction model corresponding to the refueling notification information is selected from the prediction model library, and combined with the fuel assembly information of the current unit, a feasible loading plan is generated.

[0074] Step S4 includes the following steps:

[0075] S4.1: According to the number of new components to be replaced in the refueling notification information, select the corresponding key parameter prediction model; and calculate based on the cycle length in the refueling notification information and the selected key parameter prediction model to select the appropriate available old components of the current unit. The specific steps for selecting available old components are as follows: Figure 5 As shown, the following steps are included:

[0076] S4.1.1: Obtain information on all available old components of the current unit, excluding damaged components and artificially selected center components.

[0077] S4.1.2: Select the 1 / 4 position components, 1 / 8 position components, and center components from largest to smallest component reactivity kinf, and load them into the key parameter prediction model. 1 / 4 position components are selected only from all 1 / 4 position components, and 1 / 8 position components are selected only from all 1 / 8 position components. If the center component is manually selected, it is not selected in this step; otherwise, it is selected.

[0078] S4.1.3: Calculate the cycle length of the key parameter prediction model after loading is completed; place the components with the smallest reactivity kinf at the 1 / 8 position on the periphery of the core and calculate the cycle length.

[0079] S4.1.4: Determine whether the calculated cycle length meets the material change notification requirements; if so, end the process; otherwise, reselect components with smaller reactivity kinf from the remaining components of the current unit to replace components with larger reactivity kinf from the selected components until the cycle length requirements are met and the cycle ends.

[0080] After completing the selection of old components, execute step S4.2: Based on the selected available old components, use the genetic algorithm to generate several loading schemes to be verified, and use the selected key parameter prediction model to predict the cycle length and maximum enthalpy rise factor of each loading scheme to be verified, so as to search for the loading scheme that requires all core refueling parameter calculations.

[0081] In this embodiment, the genetic search process of loading schemes is mainly operated by genetic chromosomes, including selection, crossover and mutation operations; the cycle length and maximum enthalpy rise factor of each loading scheme are predicted by neural network to calculate the fitness. Figure 6 As shown, the specific steps include:

[0082] S4.2.1: Random Generation m An initial population is generated by crossing over and mutation, using a set of digitally encoded chromosomes to simulate each loading scenario. 2*m offspring.

[0083] S4.2.2: Use the trained key parameter prediction model to make predictions 2*m The maximum enthalpy rise factor and cycle length of each loading scheme are searched out, and the loading scheme with the maximum enthalpy rise factor and cycle length meeting the requirements is saved. In this embodiment, each generation only saves the optimal 0.01*m A plan.

[0084] S4.2.3: Simulate chromosome crossover and / or mutation operations to eliminate individuals with low fitness and generate new individuals with high fitness, following the natural selection principle of survival of the fittest. Fitness includes cycle length and maximum enthalpy rise factor. In this embodiment, simulated chromosome crossover and / or mutation operations are performed only on components within the same region.

[0085] S4.2.4: When the saved loading plan is greater than m or iteration k End the search after times.

[0086] S4.2.5: Select the best 0.01*m Recalculate the maximum enthalpy rise factor and cycle length for each loading scheme and select the optimal one 0.001*m All core refueling parameters are calculated for each loading scheme.

[0087] in, m is the initial population size of the loading scheme, k is the number of iterations of the genetic search. m and k are considered to be computationally expensive, and are generally set to 500 and 110, respectively. Of course, other values ​​can be selected based on actual needs.

[0088] Compared with the historical solutions, the method using the specific embodiment of the present invention is obviously better than the historical solutions.

[0089] The search results of the 5-cycle a~e schemes of units 1 and 2 of a nuclear power plant are compared with their actual historical scheme results. Figure 7 、 8 From the comparison of the results of the five cyclic search schemes, it can be seen that the characteristic parameters of the core loading scheme searched by this method are better than those searched manually.

[0090] This method was validated during actual refueling design for Unit 4 of a nuclear power plant during its 10th cycle and Unit 2 of a second nuclear power plant during its 7th cycle. In Unit 2 of the second nuclear power plant, 68 new components were planned for its 7th cycle. Using this method, 64 new component solutions were found to meet the plant's refueling notification requirements. Based on core metrics, these 64 new component solutions met all required parameters and could generate the economic benefits of nearly four new components.

[0091] Comparison of the manual search scheme and the search scheme of this method for the 10th cycle of Unit 4 of a nuclear power plant

[0092]

[0093] Comparison of manual search scheme and this method's search scheme for the 7th cycle of Unit 2 of a nuclear power plant

[0094]

[0095] After comparative analysis, the material replacement scheme search method adopted in this method can successfully search for a scheme that meets the requirements. The searched scheme can make full use of the old components of the previous cycle and some key characteristic parameters are better than manual search, which has great engineering application prospects.

[0096] It is understandable that the above embodiments only express the preferred implementation modes of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the patent scope of the present invention. It should be pointed out that for ordinary technicians in this field, without departing from the concept of the present invention, the above technical features can be freely combined, and several deformations and improvements can be made, all of which fall within the scope of protection of the present invention. Therefore, all equivalent changes and modifications made to the scope of the claims of the present invention should fall within the scope of coverage of the claims of the present invention.

Claims

1. A method for searching and optimizing a material change loading scheme, characterized in that: The following steps are involved: S1: generating a plurality of commonly used new component layouts based on a plurality of historical refueling plans; and obtaining old component information in the historical refueling plans to generate an old component database; the old component information includes old position information of each old component in the core in the historical refueling plans; S2: According to the commonly used new component layout, select the old components from the old component database, randomly load them into the commonly used new component layout to generate K loading schemes, and select K1 from the K loading schemes to form training samples; Among them K1 <K; In step S2, according to the position loading rule, the old components matching the old position information are selected from the old component database and randomly loaded into the common new component layout; The position loading rule includes: the old component at the 1 / 4 position in the historical material replacement plan is selectively placed in the central component position and the 1 / 4 position; the old component at the 1 / 8 position in the historical material replacement plan is selectively placed in the central component position, the 1 / 4 position, and the 1 / 8 position; S3: using the training samples to train and generate a prediction model library that meets the accuracy requirements; the prediction model library contains several key parameter prediction models; wherein the key parameters include cycle length and maximum enthalpy rise factor; S4: According to the refueling notification information of the current unit, the key parameter prediction model corresponding to the refueling notification information is selected from the prediction model library, and combined with the fuel assembly information of the current unit to generate a feasible loading plan.

2. The method for searching and optimizing the material reloading scheme according to claim 1, characterized in that: In step S1, based on the layout characteristics of the fuel assemblies in the core, a local area of ​​the core is selected as the area for the common new assembly layout.

3. The method for searching and optimizing the material reloading scheme according to claim 2, characterized in that: The local area is a 1 / 8 core area of ​​the core, and eight 1 / 8 core areas are symmetrically arranged to form a complete area of ​​the core.

4. The method for searching and optimizing the material reloading scheme according to claim 1, characterized in that: The step S2 further includes: selecting K2 loading schemes from the K loading schemes to form test samples; wherein K1+K2=K.

5. The method for searching and optimizing the material reloading scheme according to claim 1, characterized in that: In step S3, deep neural network training is performed on the training samples with the same number of new components to obtain the key parameter prediction model corresponding to the same number of components.

6. The method for searching and optimizing the material reloading scheme according to claim 5, characterized in that: The deep neural network includes: an input layer, a network layer and an output layer; The input layer is used to input kinf eigenvalues, gd eigenvalues, enrichment eigenvalues ​​and four-corner burnup eigenvalues; wherein, n input nodes; The network layer includes 18 network layers, with 7 nodes in each layer. n Each network layer uses full connection, Sigmoid activation function, L2 regularization and Batch Normalization; The output layer is used to output parameter prediction values ​​to determine whether the requirements are met, and the parameter prediction values ​​include the maximum enthalpy rise factor and the cycle length; in, n is the number of old components placed in the training sample.

7. The method for searching and optimizing the material reloading scheme according to claim 1, characterized in that: The step S4 comprises the following steps: S4.1: Selecting the corresponding key parameter prediction model based on the number of new components to be replaced in the refueling notification information; and performing calculations based on the cycle length in the refueling notification information and the selected key parameter prediction model to select appropriate available old components for the current unit; S4.2: Based on the selected available old components, a genetic algorithm is used to generate several loading schemes to be verified, and the cycle length and maximum enthalpy rise factor of each loading scheme to be verified are predicted using the selected key parameter prediction model to search for a loading scheme that requires all core refueling parameter calculations.

8. The method for searching and optimizing the material reloading scheme according to claim 7, characterized in that: The step S4.1 includes the following steps: S4.1.1: Obtain information on all available old components of the current unit; S4.1.2: Select the 1 / 4 position component, the 1 / 8 position component, and the center component according to the component reactivity kinf from large to small, and load them into the key parameter prediction model; S4.1.3: Calculate the cycle length of the key parameter prediction model after loading is completed; S4.1.4: Determine whether the calculated cycle length meets the material change notification requirements; if so, end the process; otherwise, reselect components with smaller reactivity kinf from the remaining components of the current unit to replace components with larger reactivity kinf from the selected components until the cycle length requirements are met and the cycle ends.

9. The method for searching and optimizing the material reloading scheme according to claim 8, characterized in that: The step S4.2 includes the following steps: S4.2.1: Random Generation m An initial population is generated by crossover and mutation using a set of digitally encoded chromosomes to simulate each loading scenario. 2*m offspring; S4.2.2: Use the trained prediction model to predict the key parameters 2*m The maximum enthalpy rise factor and cycle length of each loading scheme are searched for a loading scheme whose maximum enthalpy rise factor and cycle length meet the requirements, and the loading scheme is saved; S4.2.3: Simulate chromosome crossover and / or mutation operations to eliminate individuals with low fitness generation by generation and generate new individuals with high fitness according to the law of natural selection where the fittest survive. The fitness includes cycle length and maximum enthalpy rise factor. S4.2.4: When the saved loading plan is greater than m or iteration k End the search after times; S4.2.5: Select the best 0.01*m Recalculate the maximum enthalpy rise factor and cycle length for each loading scheme and select the optimal one 0.001*m Calculate all core refueling parameters based on the loading scheme; in, m is the initial population size of the loading scheme, k is the number of iterations of genetic search.

10. The method for searching and optimizing the material reloading scheme according to claim 9, characterized in that: In step S4.2.3, the simulated chromosome crossover and / or mutation operations are only performed on components within the same region.

Images