Rapid prediction method for thermal-mechanical coupling process of thermochemical energy storage carrier particles

Abstract

The invention discloses a rapid prediction method for a particle thermal-mechanical coupling process, and the method comprises the steps: scanning energy carrier particles based on a microfocus computed tomography technology, and generating a three-dimensional digital model; importing the three-dimensional digital model into a multi-physics field simulation platform, and establishing a transient full-order model; based on the transient full-order model, generating a training data set and a test data set under different working conditions and corresponding temperature fields and stress fields; performing decomposition calculation on the training data set by using an intrinsic orthogonal decomposition POD reduced-order model to obtain a POD primary function and a spectral coefficient; establishing a mapping model from an input working condition to an output spectral coefficient based on a BP neural network; inputting the target working condition into the mapping model, and outputting a POD spectral coefficient; and on the basis of the calculation result and the POD spectral coefficient, reconstructing full-field physical quantity distribution at any moment under the target working condition through linear combination. The method solves the problems that traditional numerical simulation is low in calculation efficiency and large in resource consumption.

Description

Technical Field

[0001] This invention relates to the field of thermochemical energy storage system technology, and in particular to a rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles. Background Technology

[0002] In high-temperature thermochemical solar energy storage systems, the energy carrier particles experience drastic temperature changes during cycling, easily generating complex thermo-mechanical loads. Their internal thermo-mechanical coupling behavior directly affects energy storage efficiency and particle structural stability. Therefore, achieving rapid and accurate prediction of the thermo-mechanical multiphysics coupling behavior of three-dimensional energy carrier particles during cycling is crucial for optimizing particle design and improving system performance. Currently, this field mainly relies on traditional numerical methods, such as the finite volume method and the finite element method, to simulate the heat conduction and stress evolution processes within the particles at all orders. Although these methods can provide relatively accurate simulation results, they consume significant computational resources and have slow solution speeds when dealing with three-dimensional particle models, nonlinear material behavior, and transient coupling processes, making it difficult to meet the real-time requirements for rapid evaluation and multi-parameter optimization in engineering applications.

[0003] To improve computational efficiency, a reduced-order modeling method based on Proper Orthogonal Decomposition (POD) has been introduced. POD can extract low-dimensional characteristic basis functions from full-order simulation data of high-temperature thermo-mechanical coupling processes, enabling the characterization of energy carrier particle behavior. However, in the traditional POD framework, solving for spectral coefficients still requires constructing and solving reduced-order differential equations online, which is complex and lacks adaptability to thermo-mechanical processes with strong nonlinearity and large parameter variations, limiting its application in rapid prediction. Summary of the Invention

[0004] The purpose of this invention is to provide a rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles. This method combines intrinsic orthogonal decomposition (POD) with a backpropagation neural network to establish an end-to-end mapping from operating parameters to POD spectral coefficients. This effectively solves the problems of low computational efficiency and high resource consumption in traditional numerical simulation, as well as the fact that traditional POD reduction methods still require real-time solution of differential equations and are not adaptable to complex operating conditions.

[0005] The objective of this invention is achieved through the following technical solution: A rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles, the method comprising: Step 1: Scan the energy carrier particles using microfocus computed tomography (CT) technology to generate a three-dimensional digital model that reflects the true geometric morphology of the energy carrier particles. Step 2: Import the three-dimensional digital model obtained in Step 1 into the multiphysics simulation platform to establish a transient full-order model describing the thermo-mechanical coupling process of the energy-carrying particles under solar irradiation. Step 3: Based on the transient full-order model established in Step 2, generate training and test datasets under different working conditions and corresponding temperature and stress fields. Step 4: Use the intrinsic orthogonal decomposition POD reduction model to decompose and calculate the training dataset to obtain the POD basis functions and spectral coefficients; Step 5: Based on the BP neural network, learn the mapping relationship between different working conditions and spectral coefficients, and establish a mapping model from input working conditions to output spectral coefficients; Step 6: Input the target operating conditions into the mapping model of Step 5, and output the predicted, dimension-reduced POD spectral coefficients. Step 7: Based on the calculation results of Step 4 and the POD spectral coefficients output in Step 6, the distribution of all physical quantities at any time under the target working condition is reconstructed by linear combination. Step 8: Verify the prediction accuracy using the test dataset generated in Step 3.

[0006] As can be seen from the technical solution provided by the present invention, the above method, by combining intrinsic orthogonal decomposition (POD) with a backpropagation neural network, establishes an end-to-end mapping from operating parameters to POD spectral coefficients. This effectively solves the problems of low computational efficiency and high resource consumption in traditional numerical simulation, as well as the shortcomings of traditional POD order reduction methods, which still require real-time solution of differential equations and are insufficiently adaptable to complex operating conditions. Compared with numerical simulation methods and traditional POD methods, it greatly improves the solution speed, providing dedicated and efficient real-time computational support for energy carrier particle design and system operation optimization. Attached Figure Description

[0007] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of a rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles provided in an embodiment of the present invention. Detailed Implementation

[0009] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments, and do not constitute a limitation of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0010] like Figure 1 The diagram shows a flowchart of a rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles provided in an embodiment of the present invention. The method includes: Step 1: Scan the energy carrier particles using microfocus computed tomography (μ-CT) to generate a three-dimensional digital model that reflects the true geometric morphology of the energy carrier particles. In this step, specifically, microfocus computed tomography (μ-CT) is used to perform high-resolution scanning on energy carrier particles with complex porous structures to obtain multiple two-dimensional grayscale slice images of the inside of the energy carrier particles. The acquired slice images were preprocessed using 3D image processing and reconstruction software, including: filtering and denoising to improve the signal-to-noise ratio; thresholding to distinguish between solid particles and pore spaces; and morphological operations to correct defects. Finally, a three-dimensional digital model that reflects the true geometric morphology of the energy carrier particles is generated, providing a real geometric basis for subsequent high-fidelity numerical simulations.

[0011] Step 2: Import the three-dimensional digital model obtained in Step 1 into the multiphysics simulation platform and establish a transient full-order model (FOM) describing the thermo-mechanical coupling process of the energy carrier particles under solar irradiation. In this step, the three-dimensional digital model is divided into unstructured meshes, and local meshing is performed in areas with large temperature gradients. Then, boundary conditions consistent with actual working conditions are applied, including applying a convection-radiation composite heat transfer boundary to the surface and constraining the displacement of the bottom of the energy-carrying particles. The thermal conductivity equation for the energy carrier particles is as follows: ; In the formula: r s The density of the energy carrier particles; C p,s The specific heat capacity of the energy carrier particles; T s The temperature of the energy carrier particles; l s The thermal conductivity of the energy carrier particles is given by t; t represents time; the subscript s indicates a solid. For the gaseous component, the Boussinesq assumption is used to approximate the density field. Except for the density in the buoyancy term of the momentum equation, where the density is a function of temperature, all other densities are treated as constants. Under these assumptions, the relationship between gas density and temperature is: ; In the formula: r g The density of the gas; r 0 represents the gas density at the reference temperature; β Δ is the coefficient of thermal expansion of the gas; T Actual temperature and reference temperature T Temperature difference between 0 and 0; subscript g indicates gas; The gas continuity equation is expressed as: ; In the formula: u g For gas velocity; The gas momentum equation is expressed as: ; In the formula: or Dynamic viscosity; p This refers to gas pressure; It is the acceleration due to gravity; The gas energy equation is expressed as: ; In the formula: C p,g Specific heat capacity of the gas; T g The gas temperature; l g is the thermal conductivity of the gas.

[0012] Step 3: Based on the transient full-order model established in Step 2, generate training and test datasets under different working conditions and corresponding temperature and stress fields. In this step, the different operating conditions P include: incident solar energy intensity I, ambient temperature, and so on. Surface convective heat transfer coefficient h and initial properties of particulate material; Generate k sets of combinations of working conditions {p1, p2, ..., p} that have spatial filling and representativeness. k}; For each set of working conditions p i If i ranges from 1 to k, run the transient full-order model established in step 2 and perform the complete transient coupling calculation; After the calculation is completed, a full-field data snapshot, including the temperature field and stress field, is output at fixed time intervals; All snapshot data are divided into training and testing datasets according to a set ratio; the training dataset is used to build POD basis functions and train neural networks; the testing dataset is used to verify the model's prediction accuracy.

[0013] Step 4: Use the intrinsic orthogonal decomposition POD reduction model to decompose and calculate the training dataset to obtain the POD basis functions and spectral coefficients; In this step, the global time mean field is first calculated from all S snapshot data (at different times and under different training conditions) in the training dataset; Among them, the average field of the temperature field for: ; In the formula: These are spatial coordinates, representing a position in three-dimensional space; For the first snapshot in the sequence j A point in time; The mean field It characterizes the dominant static features of the system; Subsequently, the pulsation field of each snapshot data is calculated. That is, deviation from the mean field The part is represented as: ; Pulsating field Key information about the dynamic evolution of the system was obtained; All S pulsating fields Arranged in columns to form a snapshot matrix , where M is the total number of spatial grid nodes; For the real number field; The goal of the intrinsic orthogonal decomposition (POD) reduced-order model is to find an optimal set of orthogonal bases. After linear reconstruction based on the first N basis vectors, the mean square error between the predicted field and the original field is minimized. Specifically, this involves adjusting the snapshot matrix... Perform singular value decomposition, represented as: ; in, column vectors These are POD spatial modes, i.e., POD basis functions, which constitute the optimal orthogonal coordinate system for describing the dynamics of the system; It is a diagonal matrix, and its diagonal elements are called singular values. The square of the singular value is the eigenvalue, which represents the energy or variance contribution carried by the corresponding POD spatial mode. The column vectors contain the time coefficient information of the spatial modes; To achieve order reduction, the first N modes are truncated. The truncation order N is determined by a preset energy percentage, expressed as: ; In the formula: For the first i The eigenvalues ​​of each POD spatial mode characterize the energy contribution of that mode; This represents a preset energy percentage threshold used to determine the cutoff order. N ; This leads to the truncated POD basis functions. pulsating field Approximately expressed as: ; Among them, coefficient These are called POD spectral coefficients, which constitute the original field, i.e., the coordinates of the previously known sample data in the low-dimensional space spanned by the POD basis functions. By pulsating field The calculation is performed by projecting onto the POD basis functions, and is expressed as follows: ; In the formula: Represents the inner product over the computational domain; For the first k One spatial grid node; For the first k The weight of each node; Perform this operation on all S snapshot data to obtain the time series of the corresponding temperature field's POD spectral coefficients. , is represented as: ; The superscript T indicates transpose; Perform the same operation on the stress field to obtain the POD basis functions and spectral coefficients of the stress field.

[0014] Step 5: Based on the BP neural network, learn the mapping relationship between different working conditions and spectral coefficients, and establish a mapping model from input working conditions to output spectral coefficients; In this step, the BP neural network adopts a multi-layer feedforward design, including an input layer, hidden layers, and an output layer; Each set of working conditions , i From 1 to k, where k represents the total number of operating conditions, it serves as an input sample for the BP neural network, and the corresponding output label is the operating condition. All spectral coefficients calculated below To reduce the complexity of network learning, dimensionality reduction is usually performed, ultimately forming training data pairs. ; Standardize all input and output data to accelerate training convergence; A supervised learning model is adopted, with mean squared error as the loss function. , is represented as: ; In the formula: This represents the number of training samples; For the first i The true spectral coefficient vector of each training sample; For the first i The neural network prediction spectral coefficient vector of each training sample; Represents the L2 norm; Calculate the loss function using the backpropagation algorithm. The gradient of the weights in the BP neural network is calculated, and an adaptive learning rate optimizer is used to update the weights. To prevent overfitting, regularization techniques and early stopping strategies are also needed. After training, a mapping model from input operating conditions to output spectral coefficients is obtained. , is represented as: .

[0015] Step 6: Input the target operating conditions into the mapping model of Step 5, and output the predicted, dimension-reduced POD spectral coefficients. In this step, if the output is a dimension-reduced feature, the complete spectral coefficient time series needs to be recovered through the corresponding inverse transformation (such as principal component reconstruction).

[0016] Step 7: Based on the calculation results of Step 4 and the POD spectral coefficients output in Step 6, the distribution of all physical quantities at any time under the target working condition is reconstructed by linear combination. In this step, the average field of the temperature field calculated in step 4 is used. The mean field of the stress field and truncated POD basis functions Combining the POD spectral coefficients output from step 6, the distribution of all physical quantities at any time under the target working condition is reconstructed through linear combination, where: Reconstructed temperature field Represented as: ; In the formula: The first prediction of the BP neural network i Each POD spectral coefficient; For the temperature field i One POD basis function; Reconstructed stress field Represented as: ; In the formula: For the stress field i One POD basis function.

[0017] The above reconstruction process involves only basic matrix-vector operations, which is highly efficient and can achieve true real-time prediction.

[0018] Step 8: Verify the prediction accuracy using the test dataset generated in Step 3.

[0019] In this step, the predicted field is compared with the real field calculated by the full-order model. The global maximum error, relative error and root mean square error of the predicted field and the real field values ​​of the nodes are calculated to quantitatively verify the prediction accuracy. At the same time, an error field is generated based on the relative error of each node.

[0020] It is worth noting that the contents not described in detail in the embodiments of the present invention belong to the prior art known to those skilled in the art.

[0021] This invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the method.

[0022] This invention also provides a computer storage medium storing a plurality of instructions adapted for loading and executing the method by a processor.

[0023] In summary, the method described in the embodiments of the present invention has the following advantages: 1. Significantly Improved Computational Efficiency. Traditional numerical simulation methods for processing 3D particle models consume significant computational resources and have slow solution speeds, making it difficult to meet the real-time requirements of engineering projects. This invention, however, uses POD (Programmable Observation) to reduce the order of high-dimensional multiphysics data, extracting low-dimensional feature basis functions and mapping the problem from a high-dimensional space to a low-dimensional space. Furthermore, a backpropagation (BP) neural network is used to establish an end-to-end mapping from operating parameters to POD spectral coefficients. The computation time is reduced from hours or days in traditional methods to mere seconds, achieving true real-time prediction. This effectively solves the problems of low computational efficiency and insufficient real-time performance, providing rapid response support for particle design and system operation optimization. 2. High prediction accuracy ensures reliable results. This invention generates training data based on a high-fidelity numerical model. The model reconstructs the real particle geometry through μ-CT scanning and applies boundary conditions consistent with actual working conditions, ensuring the accuracy of the physical process. During the POD order reduction process, modal truncation is performed based on the energy proportion criterion, preserving key information about the system's dynamic evolution, enabling the low-dimensional representation to approximate the full-order model with high accuracy. Simultaneously, the BP neural network is optimized through supervised learning using mean squared error as the loss function, and regularization and early stopping strategies are employed to prevent overfitting. Validation using a test dataset shows that the error between the predicted field and the full-order model reference field demonstrates that this method maintains high accuracy under various working conditions. 3. High adaptability, suitable for complex nonlinear operating conditions. Traditional POD methods have limited adaptability when dealing with highly nonlinear processes and large parameter variations. This invention utilizes the nonlinear mapping capability of BP neural networks to directly learn the complex relationship between operating parameters and POD spectral coefficients, without the need to directly solve mathematical and physical equations. This data-driven approach enables the method to flexibly adapt to various operating parameter variations, effectively improving the method's robustness and generalization ability. 4. Highly specialized, with optimized workflow for rapid prediction of energy carrier particles. Starting from real particle geometry, this invention undergoes high-fidelity modeling, data generation, order reduction decomposition, neural network mapping, and rapid prediction and verification, forming a complete, specialized, and efficient solution. This workflow is designed for the specific physical processes and engineering requirements of energy carrier particles in high-temperature thermochemical energy storage scenarios, achieving end-to-end transformation from complex physical problems to lightweight computational tools. It provides standardized analysis tools for this specific field, significantly improving the overall efficiency from design to optimization.

[0024] Furthermore, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware, and the corresponding program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0025] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims. The information disclosed in the background section is intended only to enhance the understanding of the overall background technology of the present invention and should not be construed as an admission or implication in any way that such information constitutes prior art known to those skilled in the art.

Claims

1. A rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles, characterized in that, The method includes: Step 1: Scan the energy carrier particles using microfocus computed tomography (CT) technology to generate a three-dimensional digital model that reflects the true geometric morphology of the energy carrier particles. Step 2: Import the three-dimensional digital model obtained in Step 1 into the multiphysics simulation platform to establish a transient full-order model describing the thermo-mechanical coupling process of the energy-carrying particles under solar irradiation. Step 3: Based on the transient full-order model established in Step 2, generate training and test datasets under different working conditions and corresponding temperature and stress fields. Step 4: Use the intrinsic orthogonal decomposition POD reduction model to decompose and calculate the training dataset to obtain the POD basis functions and spectral coefficients; Step 5: Based on the BP neural network, learn the mapping relationship between different working conditions and spectral coefficients, and establish a mapping model from input working conditions to output spectral coefficients; Step 6: Input the target operating conditions into the mapping model of Step 5, and output the predicted, dimension-reduced POD spectral coefficients. Step 7: Based on the calculation results of Step 4 and the POD spectral coefficients output in Step 6, the distribution of all physical quantities at any time under the target working condition is reconstructed by linear combination. Step 8: Verify the prediction accuracy using the test dataset generated in Step 3.

2. The rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles according to claim 1, characterized in that, In step 1, specifically, microfocus computed tomography is used to perform high-resolution scanning on energy carrier particles with complex porous structures to obtain multiple two-dimensional grayscale slice images of the inside of the energy carrier particles. The acquired slice images were preprocessed using 3D image processing and reconstruction software, including: filtering and denoising to improve the signal-to-noise ratio; thresholding to distinguish between solid particles and pore spaces; and morphological operations to correct defects. Finally, a three-dimensional digital model that reflects the true geometric shape of the energy carrier particles is generated.

3. The rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles according to claim 1, characterized in that, In step 2, the three-dimensional digital model is divided into unstructured meshes, and local meshing is performed in areas with large temperature gradients; Then, boundary conditions consistent with actual working conditions are applied, including applying a convection-radiation composite heat transfer boundary to the surface and constraining the displacement of the bottom of the energy-carrying particles. The thermal conductivity equation for the energy carrier particles is as follows: ； In the formula: ρ s The density of the energy carrier particles; C p,s The specific heat capacity of the energy carrier particles; T s The temperature of the energy carrier particles; λ s The thermal conductivity of the energy carrier particles is given by t; t represents time; the subscript s indicates a solid. For the gas component, the relationship between gas density and temperature is as follows: ； In the formula: ρ g The density of the gas; ρ 0 represents the gas density at the reference temperature; β Δ is the coefficient of thermal expansion of the gas; T Actual temperature and reference temperature T Temperature difference between 0 and 0; subscript g indicates gas; The gas continuity equation is expressed as: ； In the formula: u g For gas velocity; The gas momentum equation is expressed as: ； In the formula: η Dynamic viscosity; p This refers to gas pressure; It is the acceleration due to gravity; The gas energy equation is expressed as: ； In the formula: C p,g Specific heat capacity of the gas; T g The gas temperature; λ g is the thermal conductivity of the gas.

4. The rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles according to claim 1, characterized in that, In step 3, the different operating conditions P include: incident solar energy intensity I, ambient temperature, and so on. Surface convective heat transfer coefficient h and initial properties of particulate material; Generate k sets of combinations of working conditions {p1, p2, ..., p} that have spatial filling and representativeness. k }; For each set of working conditions p i If i ranges from 1 to k, run the transient full-order model established in step 2 and perform the complete transient coupling calculation; After the calculation is completed, a full-field data snapshot, including the temperature field and stress field, is output at fixed time intervals; All snapshot data are divided into training and testing datasets according to a set ratio; the training dataset is used to build POD basis functions and train neural networks; the testing dataset is used to verify the model's prediction accuracy.

5. The rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles according to claim 1, characterized in that, The process of step 4 is as follows: First, calculate the global time-averaged field from all S snapshots in the training dataset; where the average field of the temperature field is... for: ； In the formula: These are spatial coordinates, representing a position in three-dimensional space. For the first snapshot in the sequence j A point in time; The mean field It characterizes the dominant static features of the system; Subsequently, the pulsation field of each snapshot data is calculated. That is, deviation from the mean field The part is represented as: ； Pulsating field Key information about the dynamic evolution of the system was obtained; All S pulsating fields Arranged in columns to form a snapshot matrix , where M is the total number of spatial grid nodes; For the real number field; The goal of the intrinsic orthogonal decomposition (POD) reduced-order model is to find an optimal set of orthogonal bases. After linear reconstruction based on the first N basis vectors, the mean square error between the predicted field and the original field is minimized. Specifically, this involves adjusting the snapshot matrix... Perform singular value decomposition, represented as: ； in, column vectors These are POD spatial modes, i.e., POD basis functions, which constitute the optimal orthogonal coordinate system for describing the dynamics of the system; It is a diagonal matrix, and its diagonal elements are called singular values. The square of the singular value is the eigenvalue, which represents the energy or variance contribution carried by the corresponding POD spatial mode. The column vectors contain the time coefficient information of the spatial modes; The first N modes are truncated, and the truncation order N is determined by a preset energy percentage, expressed as: ； In the formula: For the first i The eigenvalues ​​of each POD spatial mode characterize the energy contribution of that mode; This represents a preset energy percentage threshold used to determine the cutoff order. N ; This leads to the truncated POD basis functions. pulsating field Approximately expressed as: ； Among them, coefficient These are called POD spectral coefficients, which constitute the original field, i.e., the coordinates of the previously known sample data in the low-dimensional space spanned by the POD basis functions. By pulsating field The calculation is performed by projecting onto the POD basis functions, and is expressed as follows: ； In the formula: Represents the inner product over the computational domain; For the first k One spatial grid node; For the first k The weight of each node; Perform this operation on all S snapshot data to obtain the time series of the corresponding temperature field's POD spectral coefficients. , is represented as: ； The superscript T indicates transpose; Perform the same operation on the stress field to obtain the POD basis functions and spectral coefficients of the stress field.

6. The rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles according to claim 1, characterized in that, In step 5, the BP neural network adopts a multi-layer feedforward design, including an input layer, a hidden layer, and an output layer; Each set of working conditions , i From 1 to k, where k represents the total number of operating conditions, it serves as an input sample for the BP neural network, and the corresponding output label is the operating condition. All spectral coefficients calculated below The final training data pairs ; Standardize all input and output data to accelerate training convergence; A supervised learning model is adopted, with mean squared error as the loss function. , is represented as: ； In the formula: This represents the number of training samples; For the first i The true spectral coefficient vector of each training sample; For the first i The neural network prediction spectral coefficient vector of each training sample; Represents the L2 norm; Calculate the loss function using the backpropagation algorithm. The gradient of the weights in the BP neural network is calculated, and an adaptive learning rate optimizer is used to update the weights. After training, a mapping model from input operating conditions to output spectral coefficients is obtained. , is represented as: 。 7. The rapid prediction method for the thermo-mechanical coupling process of thermochemical energy storage carrier particles according to claim 5, characterized in that, In step 7, the average field of the temperature field calculated in step 4 is used. The mean field of the stress field and truncated POD basis functions Combining the POD spectral coefficients output from step 6, the distribution of all physical quantities at any time under the target working condition is reconstructed through linear combination, where: Reconstructed temperature field Represented as: ； In the formula: The first prediction of the BP neural network i POD spectral coefficients; For the temperature field i One POD basis function; Reconstructed stress field Represented as: ； In the formula: For the stress field i One POD basis function.

8. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 7.

9. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions adapted for loading by a processor and executing the method of any one of claims 1 to 7.

Images