Method and system for predicting state of reactor core

Abstract

Embodiments of the present application relate to the technical field of nuclear reactors, and disclose a method and system for predicting the state of a reactor core, a device, and a medium. The method comprises: acquiring first reactor core state measurement parameters of a target reactor core in a first time step, the first time step being a time step obtained by dividing a combustion process of the target reactor core, and the first reactor core state measurement parameters comprising state measurement parameters of the target reactor core in a plurality of physical fields; when the first reactor core state measurement parameters are inputted into a data fusion model, by means of the data fusion model, performing fusion processing on the state measurement parameters of the plurality of physical fields, and obtaining a target data matrix; and when the target data matrix is inputted into a state prediction model, by means of the state prediction model, performing prediction to obtain first reactor core state information, the first reactor core state information being used for representing a reactor core state of the target reactor core in a second time step, and the second time step being later than the first time step.

Description

Method and system for predicting reactor core state TECHNICAL FIELD

[0001] The present application relates to the technical field of nuclear reactors, and particularly relates to a method and system for predicting a reactor core state, a device and a medium. BACKGROUND

[0002] A high-fidelity numerical reactor model provides a tool for comprehensively understanding and evaluating the behavior of a nuclear energy system, which is widely used in the field of nuclear energy by accurately describing and simulating the physical processes and behaviors of a reactor. However, the calculation time of the high-fidelity numerical reactor model is usually in days or months, while the refresh frequency of the reactor measured data is usually in seconds. Obviously, the high-fidelity numerical reactor model cannot accurately reflect the changes in the reactor measured data. SUMMARY

[0003] Therefore, one of the purposes of the present application is to provide a method and system for predicting a reactor core state, and an electronic device and a computer readable storage medium, which can accurately predict the reactor core state.

[0004] In a first aspect, an embodiment of the present application provides a method for predicting a reactor core state, comprising:

[0005] obtaining a first core state measurement parameter of a target reactor core at a first time step, wherein the first time step is one time step obtained by dividing a burnup process of the target reactor core, and the first core state measurement parameter comprises a state measurement parameter of the target reactor core in a plurality of physical fields;

[0006] in a case where the first core state measurement parameter is input to a data fusion model, performing fusion processing on the state measurement parameters of the plurality of physical fields by the data fusion model to obtain a target data matrix;

[0007] in a case where the target data matrix is input to a state prediction model, predicting a first core state information by the state prediction model, wherein the first core state information is used to represent a core state of the target reactor core at a second time step, and the second time step is later than the first time step.

[0008] In a possible implementation, in a case where the target data matrix is input to the state prediction model, the first core state information is predicted by the state prediction model, comprising:

[0009] determining a prediction covariance matrix according to the state prediction model;

[0010] performing initialization processing on the first core state measurement parameter to obtain a state estimation vector and a state covariance matrix;

[0011] determining a target state particle at the first time step according to the state estimation vector and the state covariance matrix;

[0012] determining a target error particle at the first time step according to the prediction covariance matrix;

[0013] determining the first core state information according to the target state particle and the target error particle.

[0014] In a possible implementation, the determining of the first core state information according to the target state particle and the target error particle comprises:

[0015] propagating the target state particle to a second time step according to the preset state transition equation and the target error particle to obtain the core state information.

[0016] In a possible implementation, after the first core state information is predicted by the state prediction model in the case that the target data matrix is input to the state prediction model, the method further comprises:

[0017] In the case that the prior probability is determined according to the core state information, the target state particle and the target error particle, the second core state measurement parameter at the second time step of the target reactor core is acquired;

[0018] determining a second target state particle at the second time step according to the second core state measurement parameter and the prior probability;

[0019] mapping the second target state particle to a measurement space to obtain a measurement hardware value and a covariance matrix of the measurement hardware value;

[0020] updating the first core state information according to the measurement hardware value and the covariance matrix of the measurement hardware value.

[0021] In a possible implementation, the updating of the first core state information according to the measurement hardware value and the covariance matrix of the measurement hardware value comprises:

[0022] In the case that the measurement value and the uncertainty at the second time step are acquired, determining a residual according to the measurement value at the second time step and the measurement hardware value;

[0023] determining a gain according to the covariance matrix of the measurement hardware value and the uncertainty;

[0024] updating the first core state information according to the residual and the gain.

[0025] In a possible implementation, the target data matrix is obtained by performing fusion processing on the state measurement parameters of the plurality of physical fields through a data fusion model, comprising:

[0026] The state measurement parameters of the plurality of physical fields are fused by a data fusion model to obtain a first data matrix.

[0027] The first data matrix is reduced in rank to obtain a second data matrix, and the target data matrix includes the second data matrix.

[0028] In a possible implementation, the state measurement parameters of the plurality of physical fields are fused by a data fusion model to obtain a first data matrix, including:

[0029] The spatial distribution data of the target reactor core under each physical field is determined to obtain a plurality of initial data matrices.

[0030] The plurality of initial data matrices are spliced to obtain the first data matrix.

[0031] In a possible implementation, the first data matrix is reduced in rank to obtain a second data matrix, including:

[0032] In a case where the first data matrix is reshaped according to a preset reshaping function, a covariance matrix of the first data matrix is determined.

[0033] An initial principal component of the covariance matrix of the first data matrix and an eigenvalue corresponding to the initial principal component are obtained.

[0034] In a case where the variance contribution rate of each initial principal component is determined, a target principal component is determined according to a preset variance contribution rate threshold.

[0035] The first data matrix after reshaping is reconstructed according to the target principal component and the eigenvalue corresponding to the target principal component to obtain the second data matrix.

[0036] In a possible implementation, an initial principal component of the covariance matrix of the first data matrix and an eigenvalue corresponding to the initial principal component are obtained, including:

[0037] The initial principal component and the eigenvalue corresponding to the initial principal component are determined according to a principal component analysis method or a singular value decomposition method.

[0038] In a second aspect, an embodiment of the present application provides a system for predicting a state of a reactor core, including:

[0039] The obtaining module is configured to obtain a first core state measurement parameter of a target reactor core at a first time step, where the first time step is one time step obtained by dividing a burning process of the target reactor core, and the first core state measurement parameter includes state measurement parameters of the target reactor core in a plurality of physical fields.

[0040] a fusion module, configured to perform fusion processing on the state measurement parameters of the plurality of physical fields by the data fusion model to obtain a target data matrix in a case where the first core state measurement parameters are input to the data fusion model;

[0041] a prediction module, configured to obtain the first core state information by the state prediction model in a case where the target data matrix is input to the state prediction model, wherein the first core state information is used to represent a core state of the target reactor core at a second time step, and the second time step is later than the first time step.

[0042] In a third aspect, an electronic device is provided, and the electronic device includes a memory and a processor. The memory has stored thereon a computer program, and the computer program, when executed by the processor, implements the method for predicting a reactor core state according to the first aspect.

[0043] In a fourth aspect, a computer-readable storage medium is provided, and the computer-readable storage medium has stored thereon a computer program, and the computer program, when executed by one or more processors, implements the method for predicting a reactor core state according to the first aspect.

[0044] The method for predicting a reactor core state provided in the embodiments of the present application can obtain the first core state measurement parameters of the target reactor core at the first time step, perform fusion processing on the state measurement parameters of the plurality of physical fields by the data fusion model in a case where the first core state measurement parameters are input to the data fusion model, obtain a target data matrix, reduce the mutual coupling and mutual influence between different physical fields, and improve the consistency between different physical fields. The first core state information can be accurately obtained by the state prediction model in a case where the target data matrix is input to the state prediction model, and the accurate prediction of the target reactor core state is realized. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. It should be understood that the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0046] FIG. 1 is a method flowchart of a method for obtaining a reactor core state according to an embodiment of the present application;

[0047] FIG. 2 is a functional module schematic diagram of a system for obtaining a reactor core state according to the present application;

[0048] FIG. 3 is an internal structure diagram of an electronic device according to an embodiment of the present application.

[0049] BRIEF DESCRIPTION OF DRAWINGS

[0050] System 200 for acquiring a reactor core state, acquisition module 210, fusion module 220, prediction module 230. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0054] It should be noted that: similar reference numbers and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0055] In various embodiments of the present application, the expression “or” or “at least one of A or / and B” includes any combination or all combinations of the listed terms. For example, the expression “A or B” or “at least one of A or / and B” can include A, can include B, or can include both A and B.

[0056] In the description of the present application, it should be noted that if the terms “upper”, “lower”, “inner”, “outer” and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the present application is usually placed, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the indicated device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0057] In addition, if the terms "first", "second" and the like are used herein, they are merely used to distinguish one entity from another, and do not imply or suggest relative importance.

[0058] It should be noted that the features in the embodiments of the present application can be combined with each other without conflict.

[0059] Please refer to Fig. 1, which is a method flow chart of a method for obtaining a reactor core state provided by an embodiment of the present application. The method for obtaining a reactor core state can be applied to an electronic device in the following embodiments, including a personal computer, a server, a mobile device, a cloud computing platform, a supercomputer and the like. The method for obtaining a reactor core state specifically includes the following steps:

[0060] In step 110, a first core state measurement parameter of a target reactor core at a first time step is obtained, wherein the first time step is a time step obtained by dividing a burnup process of the target reactor core, and the first core state measurement parameter includes a state measurement parameter of the target reactor core in multiple physical fields.

[0061] In step 120, in a case where the first core state measurement parameter is input to a data fusion model, the state measurement parameters of the multiple physical fields are fused by the data fusion model to obtain a target data matrix.

[0062] In step 130, in a case where the target data matrix is input to a state prediction model, a first core state information is predicted by the state prediction model, wherein the first core state information is used to represent a core state of the target reactor core at a second time step, and the second time step is later than the first time step.

[0063] The method for predicting a reactor core state provided by the embodiment of the present application can obtain a first core state measurement parameter of a target reactor core at a first time step, fuse the state measurement parameters of the multiple physical fields by a data fusion model in a case where the first core state measurement parameter is input to the data fusion model to obtain a target data matrix, thereby reducing the mutual coupling and mutual influence between different physical fields and improving the consistency between different physical fields. In a case where the target data matrix is input to a state prediction model, the first core state information can be accurately obtained by the state prediction model, thereby realizing accurate prediction of the target reactor core state.

[0064] The following will be a detailed description of each step of the method in Fig. 1.

[0065] In step 110, a first core state measurement parameter of the target reactor core at a first time step can be acquired by the electronic device. The first time step is a time step obtained by dividing a burnup process of the target reactor core, and the first core state measurement parameter includes a state measurement parameter of the target reactor core in a plurality of physical fields.

[0066] The target reactor core can be any one of a plurality of reactor cores in at least one nuclear power plant, or at least two reactor cores in the at least one nuclear power plant. The number of target reactor cores can be selected according to actual needs. For the purpose of clear and simple introduction, the target reactor core in the embodiment and the following embodiments can be regarded as any one of a plurality of reactor cores in at least one nuclear power plant.

[0067] At present, common nuclear reactor types in the world include pressurized water reactors, boiling water reactors, heavy water reactors, gas cooled reactors, and molten salt reactors. The method for acquiring a reactor core state provided in the embodiment can be applied to any type of nuclear reactor, and has a wide application range.

[0068] The first time step can also be referred to as a first burnup step. The burnup step is a main process of nuclear fuel depletion and fission product accumulation in the reactor core. The process can be divided into a plurality of time steps or burnup steps, and the first time step is any one of the plurality of time steps.

[0069] The plurality of physical fields are physical fields related to disciplines of operating parameters of the target reactor core. For example, temperature distribution, neutron flux distribution, flow field distribution, stress distribution, power distribution, product deposition distribution, and the like.

[0070] Corresponding to the above examples, the first core state measurement parameter can include temperature, neutron flux, mechanical stress, thermal stress, oxide, carbonate, silicate, speed of flow medium such as water, pressure, temperature, and the like.

[0071] In step 120, when the first core state measurement parameter is input to the data fusion model, the state measurement parameters of the plurality of physical fields are fused by the data fusion model to obtain a target data matrix.

[0072] The state measurement parameters in the plurality of physical fields can be collected by a plurality of detectors in the target reactor core. However, the state measurement parameters collected by each detector can be different, and the collection accuracy is not completely the same, and even has a large difference.

[0073] The state measurement parameters of multiple physical fields are fused by the data fusion model in this embodiment to obtain a target data matrix. The target data matrix is applied to predict the core state of the target reactor core in the following embodiments, which can reduce the influence between different physical fields, obtain a comprehensive and multi-angle prediction of the core state, reduce the uncertainty of information, and further improve the accuracy of predicting the core state of the target reactor core.

[0074] In step 130, the first core state information is predicted by the state prediction model when the target data matrix is input to the state prediction model, wherein the first core state information is used to represent the core state of the target reactor core at the second time step, which is later than the first time step.

[0075] Specifically, the target data matrix is obtained based on the first core state measurement parameter at the first time step. Based on the target data matrix, the first core state information at the second time step can be further predicted by the state prediction model in this embodiment to determine the core state of the target reactor core at the second time step.

[0076] The state prediction model is trained based on the historical core state information in the training sample and the historical data matrix corresponding to the training sample. The training sample also includes historical core state measurement parameters, and the historical data matrix is obtained by fusing the historical core state measurement parameters based on the data fusion model.

[0077] In one possible implementation, the target data matrix is obtained by fusing the state measurement parameters of multiple physical fields by the data fusion model, comprising:

[0078] The state measurement parameters of multiple physical fields are fused by the data fusion model to obtain the first data matrix.

[0079] The first data matrix is reduced to obtain the second data matrix, and the target data matrix includes the second data matrix.

[0080] The data fusion model can fuse the state measurement parameters of multiple physical fields in this embodiment, which can reduce the influence between different physical fields, obtain a comprehensive and multi-angle prediction of the core state, reduce the uncertainty of information, and further improve the accuracy of predicting the core state of the target reactor core.

[0081] Further, the first data matrix is reduced in this embodiment, which can reduce the amount of data to be processed, reduce the complexity of solving and predicting the state prediction model, and be beneficial to improve the prediction efficiency of the state prediction model.

[0082] In some embodiments, the target data matrix includes a first data matrix, and the data fusion model can only fuse the state measurement parameters of the plurality of physical fields in the process of processing to meet special prediction accuracy requirements.

[0083] In some embodiments, the consistency of the state measurement parameters of different physical fields can be achieved according to a preset coupling strategy in the process of fusing the state measurement parameters of the plurality of physical fields. Specifically, consistency of each physical field at each time step can be forced.

[0084] In some embodiments, the consistency of the state measurement parameters of different physical fields can also be achieved by iterative solving until convergence. Specifically, an equation set for describing the coupling relationship between different physical fields can be established first, and the initial values of all physical fields are set. In each iteration, the coupling equation set is updated according to the current value of the physical field, and it is checked whether the value of the physical field has converged to a preset accuracy range. If it converges to the preset accuracy range, the iteration can be stopped, otherwise, the iteration continues until the value of the physical field converges to the preset accuracy range.

[0085] In a possible implementation, the state measurement parameters of the plurality of physical fields are fused by the data fusion model to obtain a first data matrix, including:

[0086] The spatial distribution data of the target reactor core under each physical field is determined to obtain a plurality of initial data matrices;

[0087] The plurality of initial data matrices are spliced to obtain the first data matrix.

[0088] The embodiment can select a splicing mode according to requirements, including any one of horizontal splicing, vertical splicing, block splicing, diagonal line splicing, etc.

[0089] In a possible implementation, the first data matrix is reduced to obtain a second data matrix, including:

[0090] In the case of remodeling the first data matrix according to a preset remodeling function, a covariance matrix of the first data matrix is determined;

[0091] An initial principal component of the covariance matrix of the first data matrix and an eigenvalue corresponding to the initial principal component are obtained;

[0092] In the case of determining the variance contribution rate of each initial principal component, a target principal component is determined according to a preset variance contribution rate threshold;

[0093] The first data matrix after remodeling is reconstructed according to the target principal component and the eigenvalue corresponding to the target principal component to obtain the second data matrix.

[0094] The embodiment realizes the order reduction processing of the first data matrix, simplifies data analysis, and further improves the prediction efficiency of the state prediction model by screening the initial principal components of the covariance matrix of the first data matrix to obtain target principal components.

[0095] In a possible implementation, the initial principal components of the covariance matrix of the first data matrix and the eigenvalues corresponding to the initial principal components are obtained, and the method comprises the following steps:

[0096] The initial principal components and the eigenvalues corresponding to the initial principal components are determined according to the principal component analysis method or the singular value decomposition method.

[0097] The embodiment can realize order reduction according to the principal component analysis method or the singular value decomposition method. Taking the principal component analysis method as an example, the original data in the covariance matrix of the first data matrix is mapped to a new low-dimensional space, and each dimension in the new space is a principal component of the original data. By retaining the most important principal components, the dimensionality of the data can be reduced, thereby simplifying data analysis and improving the prediction efficiency of the state prediction model.

[0098] In addition, the principal component analysis can also identify redundant information and noise in the data, and extract the most representative features, thereby realizing the simplification and refinement of the data. By retaining the main components, the essential features of the data can be captured more effectively, and the prediction accuracy of the state prediction model can be improved.

[0099] In order to more clearly describe the specific implementation process in the above embodiment, as a process for determining the first data matrix, the following is an example:

[0100] 1.1 Determine the number of physical fields as M1.

[0101] 1.2 Divide the target reactor core into discrete small units to form a calculation network, and the number of spatial grids of the target reactor core is M2.

[0102] 1.3 Define initial conditions for each physical field. The initial conditions refer to the initial state of the physical field at the beginning of the calculation, and the parameters related to the initial state, such as the rod position of the control rod in the target reactor core, the critical boron concentration, the core power level, the burnup level, the power operation history, etc. Each of the above state parameters is discretized and finally forms different state permutations and combinations to obtain M3 core initial states.

[0103] 1.4 Set the coupling strategy. Set the total running time T and the time step dt, and the number of time points for each simulation is M4, and M4*dt=T.

[0104] 1.5 Multiple physical field iteration. For M3 core initial states, iterative calculation of multiple physical fields is performed by means of nuclear reactor design software such as, and the spatial distribution of each field in the nuclear reactor multiple physical field is recorded at each time step. Finally, multiple initial data matrices X1 are formed, with dimensions (M3, M4, M1, M2).

[0105] 1.6 Splicing of multiple physical field data. The splicing of multiple initial data matrices is completed by means of the reshape function in the open source matrix calculation library such as numpy (np), to obtain the first data matrix X2 as follows: X2 = np.reshape (X1, (M3, M4, M1*M2)) Equation 1

[0106] Taking the above example as an example, a process example for determining the second data matrix is as follows:

[0107] 2.1 Reshape the first data matrix X2 into a two-dimensional matrix as follows: X2 reshaped = np.reshape (X2, (M3*M4, M1*M2)) Equation 2

[0108] 2.2 Calculate the covariance matrix of the reshaped first data matrix X2 as follows: cov_M = np.cov (X2 reshaped , rowvar = False) Equation 3

[0109] Wherein, cov_M is the calculated covariance, and rowvar = False is the input keyword of the np.cov function.

[0110] 2.3 Principal component analysis of the covariance matrix of the reshaped first data matrix X2. Principal component analysis (PCA) is performed on the covariance matrix to obtain principal components and their corresponding eigenvalues. Optionally, the PCA module of the open source machine learning library sklearn is imported by using the decomposition module, and the principal component decomposition and corresponding eigenvalues can be output as follows: eig val , eig vec = PCA (cov_M) Equation 4

[0111] Wherein, eig val is a list of eigenvalues (also known as characteristic values), and the eigenvalues are usually arranged in a list from large to small, represented as (eig val,1 ≥ eig val,2 ≥…≥ eig val,neig ), neig is the number of eigenvalues. eig veca list of eigenvectors corresponding to the eigenvalues (also called eigenvector).

[0112] 2.4 Cumulative variance contribution rate calculation. For each initial principal component, calculate its corresponding variance contribution rate. The variance contribution rate represents the percentage of total variance explained by the initial principal component. r,i = eig val,i / np.sum(eig val ) Formula 5 cum_var r = np.cumsum(var r ) Formula 6

[0113] where var r represents the list of normalized eigenvalues, var r,i is the i-th variable value in the list. cum_var r represents the cumulative value list obtained by cumulating var r,i , where the last value in the list is 1.

[0114] 2.5 Determining target principal components and their eigenvalues. According to the cumulative variance contribution rate, determine the number of principal components to be retained M5. Traverse the variance contribution rate of the principal components until the cumulative variance contribution rate reaches a preset threshold such as 0.99. Generally, since the variance contribution rate of the principal components decreases exponentially, only a small number of principal components need to be retained to achieve a retention accuracy of 0.99 or above. Therefore, M5 << M1 * M2. Where M5 = np.argmax(cum_var r > = 0.99) + 1 Formula 7 W = eig vec [:, M5] Formula 8

[0115] where W represents the dimension reduction matrix, with dimensions (M1 * M2, M5).

[0116] 2.6 Reconstruction of the second data matrix X3. Using the retained target principal components and corresponding eigenvalues, reconstruct the two-dimensional matrix: X2 recon with dimensions (M3 * M4, M5), where X2 recon = np.dot(X2 reshaped , W) Formula 9

[0117] Further, the two-dimensional matrix X2 reconThe reconstruction is a three-dimensional matrix X3 with dimensions (M3, M4, M5).

[0118] It can be seen that the dimension of the time series data matrix can be reduced while the cumulative variance contribution rate of the PCA principal component (target principal component) is maintained at 0.99 or above. In this way, the dimension can be reduced while maintaining high data interpretation capability, thereby improving prediction efficiency.

[0119] In some embodiments, the second time step is the next adjacent time step of the first time step.

[0120] In one possible implementation, when the target data matrix is input into the state prediction model, the first core state information is predicted by the state prediction model, including:

[0121] According to the state prediction model, a prediction covariance matrix is determined.

[0122] The first core state measurement parameter is initialized to obtain a state estimation vector and a state covariance matrix.

[0123] According to the state estimation vector and the state covariance matrix, a target state particle at the first time step is determined.

[0124] According to the prediction covariance matrix, a target error particle at the first time step is determined.

[0125] According to the target state particle and the target error particle, the first core state information is determined.

[0126] The embodiment combines the state prediction model and the Bayesian filter to further improve the accuracy of determining the first core state information, and thus predict the core distribution with high accuracy.

[0127] In one possible implementation, according to the target state particle and the target error particle, the first core state information is determined, including:

[0128] According to the preset state transition equation and the target error particle, the target state particle is propagated to the second time step to obtain the core state information.

[0129] Similarly, in order to more clearly illustrate the specific implementation process in the above embodiments, the above process examples 1.1-1.6 and 2.1-2.6 are taken as an example process for determining a state prediction model as follows:

[0130] 3.1 Build a transformer model. The input is a matrix X in with time series and reduced dimension d and M5, and the output is a matrix X out with time series and reduced dimension k and M5.

[0131] According to the characteristics of the nonlinear dynamics of nuclear reactors, the time sequence value of d can be between 2-100, and d≤M4. Optionally, the value of k is selected according to the length of the desired prediction, usually 1≤k≤d.

[0132] 3.1.1 Build input encoder: (1) Embedding encoding of input, mapping each element to a low-dimensional feature space. An embedding layer can be used to convert the input into a lower-dimensional feature representation. (2) To capture the timing information, position encoding can be added to introduce the sequential information of the timing dimension into the model. in

[0133] 3.1.2 Build Transformer encoding layer: composed of multiple Transformer encoder layers, each encoder layer contains a self-attention mechanism (Self-Attention) and a feed-forward neural network (Feed-Forward Neural Network). The self-attention mechanism can learn the dependency between elements in the time series data and model the timing information. The feed-forward neural network is used for nonlinear transformation of time series data.

[0134] 3.1.3 Build Transfomer decoding layer: similar to the input encoder, the decoder layer is responsible for converting the features learned by the model into output time series data. Multiple layers of Transformer decoder layers can be used to reconstruct and map the features.

[0135] 3.1.4 Build output decoder: the output decoder maps the output of the decoder layer back to the target dimension k and the spatial dimension M5.

[0136] 3.2 Selection of state prediction model parameters.

[0137] 3.2.1 Number of Transformer layers (n layers ): the number of Transformer layers in the encoder and decoder. A deeper model can provide more complex representation ability, but also increases the computational cost and training difficulty. The present application generally takes the value of 4-12.

[0138] 3.2.2 Feed-forward neural network hidden layer dimension (d model ): specifies the dimension of the hidden layer in the feed-forward neural network. Larger hidden layer dimensions can provide stronger nonlinear modeling capabilities. It also becomes the dimension of the feature vector, generally taking the value of 128-1024.

[0139] 3.2.3 Number of attention heads (n heads ​): Specifies the number of multi-head attention heads in the attention mechanism. Each attention head can focus on different parts of the input sequence, thus providing better representation and modeling capabilities. A typical value is 4-16. A common choice is to set the number of attention heads to the number of layers in the model (n). layers The square root of ) can be used to balance the complexity of the model and computational efficiency to some extent.

[0140] 3.2.4 Embedding Dimension (e) dim ): Specifies the dimension of the input embedding layer and positional encoding. Typically, the embedding dimension should be large enough to capture rich features of the input data. A common value is d. model / n heads .

[0141] 3.2.5 Activation function type, generally takes the value "ReLU", but this application does not impose any restrictions.

[0142] 3.2.6 Learning Rate (LR): Specifies the learning rate used during model training, controlling the step size for parameter updates. The choice of learning rate has a significant impact on the training performance and convergence speed of the model. A typical value is between 0.001 and 0.0001.

[0143] 3.3 Constructing the Training Data Set. From the second data matrix X3, construct the training set and the test set. Let the number of working condition samples corresponding to the training set be M31, and the number of working condition samples corresponding to the test set be M32, where M3 = M31 + M32. Then the input to the training set is X_train. in The dimensions are (M31, M4-dk, d, M5), and the corresponding output set is X_train. out The dimensions are (M31, M4-dk, k, M5), where (M4-dk) represents the maximum number of segments that can be extracted from a time series. Similarly, X_test is obtained. in and X_test out Their dimensions are (M31, M4-dk, d, M5) and (M32, M4-dk, k, M5), respectively.

[0144] Taking the training set as an example:

[0145] (1) Determine the number of training samples: Since the length of each time series segment is M4, after truncation, the remaining segment is M4-dk. Therefore, the number of training samples is M3*(M4-dk).

[0146] (2) Initialize X_train in and X_train out Create an empty X_train in and X_train outMatrixes with dimensions (M31, M4-d-k, d, M5) and (M31, M4-d-k, k, M5) respectively.

[0147] (3) Construct training samples: for each sample (M3 time series), iterate according to the dimensions of the time series. For each time series, from the 1st time step, sequentially intercept time series segments with length M4-d-k. For each time series segment, take the data of the first d time steps as input and the data of the next k time steps as output. Add the intercepted time series segment to X_train in and X_train out in the corresponding position.

[0148] Similarly, construct the test set X_test in and X_test out No longer on display.

[0149] 3.4 Model training and model testing.

[0150] According to the sample set constructed by X_train in and X_train out , model training is carried out, and appropriate loss functions such as mean squared error (Mean Squared Error) or mean absolute error (Mean Absolute Error) are used to measure the difference between the model output model_train out and the true output X_train out . Gradient descent and other optimization algorithms can be used to minimize the loss and update the model parameters in 3.2.

[0151] Apply the input X_test in of the test set to the transformer model to obtain model_test out , if model_test out meets the preset conditions, a trained state prediction model is obtained, otherwise the process of 3.4 is repeated until model_test out meets the preset conditions.

[0152] In order to improve the accuracy and prediction ability of the state prediction model, by understanding the uncertainty in the model, the limitations of the state prediction model can be identified and appropriate improvement measures can be taken to improve the accuracy and reliability of the model prediction. Full-order model construction and uncertainty quantification such as 4.1-4.4 example process are required.

[0153] 4.1 Full order model construction and uncertainty quantification. Since the dimension reduction matrix W in the aforementioned example is an orthogonal matrix (each column is unit orthogonal), its inverse matrix is equal to its transpose matrix. The inverse transformation matrix W inv of dimension (M5, M1*M2) is calculated. Using the linalg.inv function in the numpy library, we can get: W inv = np.linalg.inv(W) Equation 10

[0154] 4.2 Multiply the reduced order matrix model_test out predicted by the Transformer with the inverse transformation matrix W inv , to get the restored prediction matrix full_model_test out as follows: full_model_test out = np.dot(model_test out , W inv .T) Equation 11

[0155] 4.3 Obtain the true value of the original data. Take the data matrix X2 in step 1, obtain the full order multi-physical field distribution value under the same time sequence input and output of the same sample condition as model_test out . Here, through a simple correspondence, we can get: X2 full of dimension X2 (M3_2, M4-d-k, k, M5).

[0156] 4.4 Evaluate the error of the state prediction model and the reduced order model of the model Transformer.

[0157] The error of the prediction model is:

[0158] where i = 1, …, k; j = 1, …, M5.

[0159] Based on the above process example, as a process example for determining the core state information, as follows:

[0160] 5.1 Combine the state prediction model with the Bayesian filter to perform real-time estimation of the multi-physical coupling state of the target reactor core based on the measured values. At this time, since real-time numerical twin is only needed to predict the state of the next time step, in the current example process, k = 1 is specified. At this time, the covariance matrix of the model prediction can be: Q = Diag(MSE(1, j), j = 1, …, M5) Equation 13

[0161] 5.2 Monte Carlo sampling.

[0162] In a multi-variate Gaussian distribution Assume N state particles are sampled using Monte Carlo sampling under the following assumption:

[0163] Similarly, N model error particles are sampled as follows:

[0164] 5.3 Target reactor core state prediction at time t+1 (second time step):

[0165] Each state particle is propagated to the next time step according to the following equation:

[0166] where F represents a mathematical model describing how the system state evolves over time, i.e., the state transition equation. It is used to predict the state of the target reactor core system at the next time step (second time step) based on the historical state and input at the current time step (first time step).

[0167] where, in the present application, F is formed from the complete calculation process of the above-mentioned example processes 2.1-4.4.

[0168] Therefore, the predicted core state at the next time step (second time step) (first core state information) can be obtained as follows:

[0169] In one possible implementation, in the case of inputting the target data matrix to the state prediction model, after predicting the first core state information by the state prediction model, the method further comprises:

[0170] In the case of determining the prior probability according to the core state information, target state particles, and target error particles, obtaining a second core state measurement parameter of the target reactor core at the second time step;

[0171] Determining a second target state particle at the second time step according to the second core state measurement parameter and the prior probability;

[0172] Mapping the second target state particle to the measurement space to obtain a measurement hardware value and a covariance matrix of the measurement hardware value;

[0173] Updating the first core state information according to the measurement hardware value and the covariance matrix of the measurement hardware value.

[0174] In one possible implementation, updating the first core state information according to the measurement hardware value and the covariance matrix of the measurement hardware value comprises:

[0175] In the case of obtaining the measurement value and uncertainty at the second time step, the residual is determined according to the measurement value and the measurement hardware value at the second time step;

[0176] The gain is determined according to the covariance matrix of the measurement hardware value and the uncertainty;

[0177] The first core state information is updated according to the residual and the gain.

[0178] The embodiment realizes the correction and update of the first core state information, and further improves the accuracy and reliability of the predicted first core state information.

[0179] Similarly, in order to more clearly set forth the specific implementation process in the above embodiment, the above process examples 5.1-5.3 are taken as an example process for determining a state prediction model as follows:

[0180] 5.4 Obtain the measurement value at t+1 time and the prior probability.

[0181] Specifically, the prior probability is determined according to the following formula:

[0182] Suppose that ND1 temperature probes, ND2 pressure probes, ND3 flow rate probes, ND4 neutron probes, etc. are arranged inside and outside the nuclear reactor. The total number of probes is represented as NDet=ND1+ND2+ND3+ND4+….

[0183] The observation function is a mathematical function used to describe the relationship between the system state and the observation value. In the state estimation problem, we usually cannot directly observe the true state of the system, but infer the state of the system through a series of observation values. The observation function defines how to map the state of the system to the observation value space.

[0184] The observation function is usually represented as h(x), where x is the state vector of the system. It accepts the state of the system as input and outputs the corresponding observation value. The observation function can be nonlinear and can contain noise terms to reflect the uncertainty of the actual observation.

[0185] The instrument control system or data centralized processing system of the nuclear power plant, etc. obtains the measurement value z t+1 and the related uncertainty R mes at t+1 time (second time step). The related instrument control system is standard, and the corresponding measurement signal is generally easy to obtain in the nuclear power plant. The covariance matrix R mes of the measurement value, with a dimension of (NDet, NDet), where NDet represents the total number of probes of the nuclear reactor.

[0186] 5.5 Predict the detector measurement value corresponding to the target reactor core state at time t+1:

[0187] After predicting the new state and the prior probability Then, based on the multi-dimensional Gaussian assumption, sample N state particles:

[0188] Then map to the measurement space to get:

[0189] Where h is the observation function in process 5.4.

[0190] The measurement hardware value and its covariance under the predicted condition are as follows:

[0191] 5.6 Measurement update of the target reactor core state at time t+1 (the second time step):

[0192] The residual between the measurement and the prediction can be expressed as y:

[0193] The corresponding gain calculation is K:

[0194] According to the gain and the residual, the predicted reactor state can be corrected or updated to the best estimated state S t+1 As follows:

[0195] The posterior probability Pt+1 is as follows:

[0196] 5.7 Let t=t+1 (update the second time step to the next time step of the second time step), repeat the above example processes 5.1-5.6 until the target reactor core stops running.

[0197] Corresponding to the above method embodiment, the embodiment of the application also provides a system for obtaining a reactor core state, please see Figure 2, Figure 2 is a functional module schematic diagram of a system for obtaining a reactor core state provided by the embodiment of the application, wherein the system 200 for obtaining a reactor core state comprises:

[0198] The acquisition module 210 is configured to acquire a first core state measurement parameter of a target reactor core at a first time step, wherein the first time step is one time step obtained by dividing the burning process of the target reactor core, and the first core state measurement parameter comprises a state measurement parameter of the target reactor core in a plurality of physical fields.

[0199] The fusion module 220 is configured to perform fusion processing on the state measurement parameters of the plurality of physical fields by using the data fusion model to obtain a target data matrix, in a case where the first core state measurement parameters are input into the data fusion model.

[0200] The prediction module 230 is configured to predict the first core state information by using the state prediction model, in a case where the target data matrix is input into the state prediction model, wherein the first core state information is used to represent a core state of the target reactor core at a second time step, and the second time step is later than the first time step.

[0201] In some embodiments, the prediction module 230 is further configured to:

[0202] determine a prediction covariance matrix according to the state prediction model;

[0203] perform initialization processing on the first core state measurement parameters to obtain a state estimation vector and a state covariance matrix;

[0204] determine a target state particle at the first time step according to the state estimation vector and the state covariance matrix;

[0205] determine a target error particle at the first time step according to the prediction covariance matrix;

[0206] determine the first core state information according to the target state particle and the target error particle.

[0207] In some embodiments, the system 200 for obtaining a reactor core state further comprises a determination submodule, which is configured to:

[0208] propagate the target state particle to the second time step to obtain the core state information according to the preset state transition equation and the target error particle.

[0209] In some embodiments, the system 200 for obtaining a reactor core state further comprises an acquisition submodule, which is configured to:

[0210] After the first core state information is predicted by using the state prediction model in a case where the target data matrix is input into the state prediction model, the method further comprises:

[0211] determine a second core state measurement parameter of the target reactor core at the second time step according to the core state information, the target state particle, and the target error particle;

[0212] determine a second target state particle at the second time step according to the second core state measurement parameter and the prior probability;

[0213] mapping the second target state particle to a measurement space to obtain a measurement hardware value and a covariance matrix of the measurement hardware value;

[0214] updating the first core state information according to the measurement hardware value and the covariance matrix of the measurement hardware value.

[0215] In some embodiments, the obtaining submodule is further configured to:

[0216] In a case where the measurement value and the uncertainty at the second time step are obtained, determining a residual according to the measurement value at the second time step and the measurement hardware value;

[0217] determining a gain according to the covariance matrix of the measurement hardware value and the uncertainty;

[0218] updating the first core state information according to the residual and the gain.

[0219] In some embodiments, the fusion module is further configured to:

[0220] performing fusion processing on the state measurement parameters of the plurality of physical fields by using a data fusion model to obtain a first data matrix;

[0221] performing order reduction processing on the first data matrix to obtain a second data matrix, wherein the target data matrix comprises the second data matrix.

[0222] In some embodiments, the fusion module is further configured to:

[0223] determining spatial distribution data of the target reactor core under each physical field to obtain a plurality of initial data matrices;

[0224] splicing the plurality of initial data matrices to obtain the first data matrix.

[0225] In some embodiments, the fusion module is further configured to:

[0226] in a case where the first data matrix is reshaped according to a preset reshaping function, determining a covariance matrix of the first data matrix;

[0227] obtaining initial principal components of the covariance matrix of the first data matrix and eigenvalues corresponding to the initial principal components;

[0228] in a case where the variance contribution rate of each initial principal component is determined, determining a target principal component according to a preset variance contribution rate threshold;

[0229] reconstructing the first data matrix after reshaping to obtain a second data matrix according to the target principal component and the eigenvalues corresponding to the target principal component.

[0230] In some embodiments, the fusion module is further configured to:

[0231] According to the principal component analysis method or the singular value decomposition method, the initial principal component and the eigenvalue corresponding to the initial principal component are determined.

[0232] The system for predicting the reactor core state provided in the embodiments of the present application can realize each process realized by the method embodiments in Figure 1 and achieve similar or the same technical effects. To avoid repetition, details are not described herein.

[0233] The present application also provides an electronic device, please see Figure 3, Figure 3 is an internal structure of an electronic device provided by the embodiments of the present application. Among them, the electronic device includes a processor, a memory and a network interface connected by a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the electronic device stores an operating system, and can also store a computer program, which, when executed by the processor, can enable the processor to implement the method for obtaining the burnup distribution of the above-mentioned embodiments applied to the electronic device. The internal memory can also store a computer program, which, when executed by the processor, can enable the processor to execute the method for obtaining the burnup distribution. Those skilled in the art can understand that the structure shown in Figure 3 is only a block diagram of part of the structure related to the present application scheme, and does not constitute a limitation on the electronic device to which the present application scheme is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0234] The embodiments of the present application also disclose a computer readable storage medium, and the computer readable storage medium stores a computer program, and the computer program is executed by a processor to realize the method for predicting the reactor core state in the method embodiments.

[0235] The embodiments of the present application provide a computer program product, which is stored in a storage medium, and the program product is executed by at least one processor to realize each process of the embodiments of the method for predicting the reactor core state, and can achieve similar or the same technical effects. To avoid repetition, details are not described herein.

[0236] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing relevant hardware, and the program can be stored in a non-volatile computer readable storage medium. When the program is executed, the program can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. The non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. The volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0237] The technical features of the above embodiments can be combined in any manner. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not contradict, they should be considered as the scope of the present disclosure.

Claims

1. A method of predicting a state of a reactor core, characterized by, The method comprises: obtaining a first core state measurement parameter of a target reactor core at a first time step, wherein the first time step is a time step divided by a burn process of the target reactor core, and the first core state measurement parameter comprises a state measurement parameter of the target reactor core in a plurality of physical fields; in a case of inputting the first core state measurement parameter into a data fusion model, performing fusion processing on the state measurement parameters of the plurality of physical fields by the data fusion model to obtain a target data matrix; in a case of inputting the target data matrix into a state prediction model, predicting a first core state information by the state prediction model, wherein the first core state information is used to represent a core state of the target reactor core at a second time step, and the second time step is later than the first time step.

2. The method of claim 1, wherein, The method further comprises: determining a prediction covariance matrix according to the state prediction model; performing initialization processing on the first core state measurement parameter to obtain a state estimation vector and a state covariance matrix; determining a target state particle at the first time step according to the state estimation vector and the state covariance matrix; determining a target error particle at the first time step according to the prediction covariance matrix; determining the first core state information according to the target state particle and the target error particle.

3. The method of claim 2, wherein, The method further comprises: propagating the target state particle to the second time step according to a preset state transition equation and the target error particle to obtain the core state information.

4. The method of claim 2, wherein, After the first core state information is predicted by the state prediction model in the case of inputting the target data matrix into the state prediction model, the method further comprises: obtaining a second core state measurement parameter of the target reactor core at the second time step according to the core state information, the target state particle and the target error particle; determining a second target state particle at the second time step according to the second core state measurement parameter and the prior probability; mapping the second target state particle to a measurement space to obtain a measurement hardware value and a covariance matrix of the measurement hardware value; updating the first core state information according to the measurement hardware value and the covariance matrix of the measurement hardware value.

5. The method of claim 4, wherein, The method further comprises: in a case of obtaining a measurement value and an uncertainty at the second time step, determining a residual according to the measurement value at the second time step and the measurement hardware value; determining a gain according to the covariance matrix of the measurement hardware value and the uncertainty; updating the first core state information according to the residual and the gain.

6. The method of claim 1, wherein, The fusion processing of the state measurement parameters of the multiple physical fields through the data fusion model comprises: The fusion processing of the state measurement parameters of the multiple physical fields through the data fusion model comprises: The order reduction processing of the first data matrix comprises:

7. The method of claim 6, wherein, The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises:

8. The method of claim 6, wherein, The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises:

9. The method of claim 8, wherein, The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises:

10. A system for predicting a state of a reactor core, characterized by, The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the first data matrix comprises: The order reduction processing of the

Images