Heat transfer solution method, device, equipment and storage medium for changing physical conditions

Abstract

The present application discloses a heat transfer solution method, device, equipment and storage medium for varying physical conditions. The method includes obtaining the physical conditions of a heat transfer object; inputting the physical conditions into a trained solution network model; determining the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model; determining an integral solution expression based on the Green's function and the Green's function gradient, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression. The present application combines the integral form of the Green's function with deep learning technology, and utilizes the analytical characteristics of the Green's function to reduce the data requirements for training the solution network model, thereby reducing the cost of solving heat transfer problems. At the same time, by combining the integral form of the Green's function with deep learning technology, the solution efficiency and solution accuracy of heat transfer problems can also be improved, which can better meet the real-time heat transfer analysis requirements.

Description

Technical Field

[0001] The present application relates to the technical field of heat transfer, and particularly relates to a heat transfer solution method, device, equipment, and storage medium with variable physical conditions. Background Art

[0002] In nature and various production technology fields, heat energy transfer caused by temperature differences is an extremely common physical phenomenon. Especially in the manufacturing process of precision products such as electronic devices, how to effectively achieve heat transfer to improve the thermodynamic performance of the structure has become an important aspect of product design. And the premise and key to such problems is to first determine the distribution law of the structural temperature field. With the development of computer technology, numerical solution techniques represented by the finite element method have been continuously improving their status in the research of heat transfer problems and are playing an increasingly important role in the engineering field. However, numerical methods face problems of high computational cost and low efficiency when dealing with complex geometric shapes, inhomogeneous material properties, and variable boundary conditions.

[0003] In recent years, deep learning technology has provided new ideas for solving heat transfer problems. AI uses technical means such as deep learning to independently mine the feature relationships in complex data, effectively breaking through the limitations of traditional physical simulations and empirical models. It can not only maintain efficient operation but also ensure the accuracy of the results. However, due to the complexity of heat transfer problems lying in the diversity of their boundary conditions and source terms, especially in complex geometric shapes and non-uniform media, a large amount of training data is required to ensure the accuracy and generalization ability of the model. This is not only time-consuming and costly in terms of calculation but may also lead to overfitting of the data, reducing the robustness of the model in practical applications.

[0004] Therefore, the existing technology still needs to be improved. Summary of the Invention

[0005] The technical problem to be solved by the present application is to provide a heat transfer solution method, device, equipment, and storage medium with variable physical conditions in view of the deficiencies of the existing technology.

[0006] To solve the above technical problem, the first aspect of the present application provides a heat transfer solution method with variable physical conditions, wherein the heat transfer solution method with variable physical conditions specifically includes:

[0007] Obtain the physical conditions of the heat transfer object;

[0008] Input the physical conditions into a trained solution network model, and determine the Green's function and the gradient of the Green's function corresponding to the heat transfer object through the solution network model, where the solution network model is a deep learning network constructed based on the integral form of the Green's function and includes a function approximation module and a gradient approximation module;

[0009] Determine an integral solution expression based on the Green's function and the gradient of the Green's function, and determine the heat transfer solution result corresponding to the heat transfer object through the integral solution expression.

[0010] The heat transfer solution method with variable physical conditions, wherein the physical conditions include geometric shape data, boundary condition data, material property data, and source term distribution data.

[0011] The heat transfer solution method with variable physical conditions, wherein the construction process of the solution network model specifically includes:

[0012] Construct an integral solution expression for the heat transfer solution problem with variable object conditions based on the Green's function, where the integral solution expression is used to determine the solution of the heat transfer solution problem with variable object conditions;

[0013] Construct a function approximation module for approximating the Green's function in the integral solution expression and a gradient approximation module for approximating the gradient of the Green's function in the integral solution expression;

[0014] Construct a solution network model to be trained based on the function approximation module and the gradient of the Green's function, and train the solution network model to be trained to obtain a solution network model.

[0015] The heat transfer solution method with variable physical conditions, wherein the training of the solution network model to be trained to obtain a solution network model specifically includes:

[0016] Obtain a training data set, wherein the training data in the training data set is obtained through data simulation or experiment, and it includes a number of training data, and each training data corresponds to a true solution;

[0017] Input the training data in the training data set into the solution network model to be trained, and output a predicted solution through the solution network model to be trained;

[0018] Construct a loss function based on the predicted solution and the true solution corresponding to the training data, and train the solution network model to be trained based on the loss function to obtain a solution network model.

[0019] The heat transfer solution method with variable physical conditions, wherein the heat transfer solution result is one or more of a temperature field distribution, a heat flux density, a temperature contour plot, and a heat flux vector plot.

[0020] The heat transfer solution method with variable physical conditions, wherein after determining the integral solution expression based on the Green's function and the gradient of the Green's function, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression, the method further includes:

[0021] The heat transfer solution results are displayed through a graphical interface.

[0022] The heat transfer solution method with variable physical conditions, wherein the integral solution expression is:

[0023] ,

[0024] wherein, represents the solution of the heat transfer problem, represents the position information of the heat transfer object, represents the volume discretization weight, represents the area discretization weight, represents the discrete number of the source term, represents n integration points, represents the m-th source term of the m-th integration point, represents the m-th Green's function of the m-th integration point, represents the discrete number of the boundary conditions, represents n integration points, represents the m-th first boundary condition of the m-th integration point, represents the m-th Green's function of the m-th integration point, represents the gradient of the Green's function at the m-th integration point, represents the second and third boundary conditions at the m-th integration point, represents the boundary weight function at the m-th integration point, represents the outward normal vector of the boundary at the m-th integration point.

[0025] The second aspect of the present application provides a heat transfer solution device with variable physical conditions, wherein the heat transfer solution device with variable physical conditions specifically includes:

[0026] A heat transfer problem determination module, configured to obtain the physical conditions of the heat transfer object;

[0027] A deep learning model module, configured to input the physical conditions into a trained solution network model, and determine the Green's function and the gradient of the Green's function corresponding to the heat transfer object through the solution network model, wherein the solution network model is a deep learning network constructed based on the integral form of the Green's function, and includes a function approximation module and a gradient approximation module;

[0028] A heat transfer solution module, configured to determine an integral solution expression based on the Green's function and the gradient of the Green's function, and determine a heat transfer solution result corresponding to the heat transfer object through the integral solution expression.

[0029] In a third aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in any one of the above-described heat transfer solution methods with variable physical conditions.

[0030] In a fourth aspect of the present application, a terminal device is provided, which includes: a processor and a memory;

[0031] The memory stores a computer-readable program executable by the processor;

[0032] When the processor executes the computer-readable program, the steps in any one of the above-described heat transfer solution methods with variable physical conditions are implemented.

[0033] Beneficial effects: Compared with the prior art, the present application provides a heat transfer solution method, device, equipment and storage medium with variable physical conditions. The method includes obtaining the physical conditions of a heat transfer object; inputting the physical conditions into a trained solution network model; determining the Green's function and the gradient of the Green's function corresponding to the heat transfer object through the solution network model; determining an integral solution expression based on the Green's function and the gradient of the Green's function, and determining a heat transfer solution result corresponding to the heat transfer object through the integral solution expression. The present application combines the integral form of the Green's function and deep learning technology, converts the heat transfer problem into the integral form of the Green's function, and then determines the Green's function and the gradient of the Green's function in the integral form of the Green's function through a solution network model constructed by a deep learning network. Finally, the heat transfer solution result is determined based on the Green's function and the gradient of the Green's function. In this way, the analytical characteristics of the Green's function can be utilized to reduce the data requirements for training the solution network model, thereby reducing the cost of solving the heat transfer problem. At the same time, by combining the integral form of the Green's function and deep learning technology, the solution efficiency and solution accuracy of the heat transfer problem can also be improved, and the real-time heat transfer analysis requirements can be better met. Description of the Drawings

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0035] Figure 1It is a flowchart of the heat transfer solution method with variable physical conditions provided by an embodiment of the present application.

[0036] Figure 2 It is an architecture diagram of the solution network model.

[0037] Figure 3 It is a display diagram of the heat transfer solution result and the actual result.

[0038] Figure 4 It is a principle block diagram of the heat transfer solution device with variable physical conditions provided by an embodiment of the present application.

[0039] Figure 5 It is a principle block diagram of the terminal device provided by an embodiment of the present application. Specific implementation manners

[0040] An embodiment of the present application provides a heat transfer solution method, device, equipment and storage medium with variable physical conditions. To make the purpose, technical solution and effect of the present application clearer and more definite, the following further details the present application with reference to the attached drawings and by way of examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0041] Those skilled in the art of this technology can understand that unless specifically stated, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "including" used in the specification of the present application means the presence of the described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more related listed items.

[0042] Those skilled in the art of this technology can understand that unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those of ordinary skill in the art to which the present application belongs. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0043] It should be understood that the sequence numbers and magnitudes of the steps in this embodiment do not imply the order of execution. The execution order of each process is determined by its function and internal logic, and should not impose any limitation on the implementation process of the embodiments of the present application.

[0044] Through research, it has been found that in nature and various fields of production technology, heat transfer caused by temperature differences is an extremely common physical phenomenon. Especially in the manufacturing process of precision products such as electronic devices, how to effectively achieve heat transfer to improve the thermodynamic performance of the structure has become an important aspect of product design. The premise and key to such problems are to first determine the distribution law of the structural temperature field. With the development of computer technology, numerical solution techniques represented by the finite element method have been playing an increasingly important role in the research of heat transfer problems in the engineering field. However, numerical methods face problems of high computational cost and low efficiency when dealing with complex geometric shapes, non-homogeneous material properties, and variable boundary conditions.

[0045] In recent years, deep learning technology has provided new ideas for solving heat transfer problems. AI uses technical means such as deep learning to autonomously discover the feature relationships in complex data, effectively breaking through the limitations of traditional physical simulations and empirical models. It can not only maintain efficient operation but also ensure the accuracy of the results. Currently, the AI solution methods adopted in the field of heat transfer are mainly divided into three categories, namely, methods for directly learning the solutions of heat conduction equations (such as physics-informed neural networks, PINN), methods for learning physical operators (such as deep operator networks, DeepONet), and operator learning methods combined with physical constraints (such as physics-informed deep operator networks, PI-DeepONet). Among them, PINN embeds partial differential equations into the loss function of the neural network and uses automatic differentiation to iteratively solve the equations without relying on training data. However, when dealing with complex boundary conditions or high-dimensional problems, PINN has a slow convergence speed and low computational efficiency, and for new boundary conditions and source terms, the network needs to be retrained. DeepONet can quickly predict the solutions under different parameters by learning the mapping relationship between input parameters and solutions; however, its accuracy and generalization ability will significantly decrease when dealing with complex geometric shapes or parameters beyond the training range. PI-DeepONet further combines the physical information of PINN and the efficiency of DeepONet to improve the accuracy and generalization ability of the model by minimizing the residual of the equation. However, due to the need to calculate the physical loss simultaneously, its training cost is high, which limits its application in large-scale problems.

[0046] As can be seen from the above, in the solution of heat transfer problems, although the existing deep learning-based methods have improved the computational efficiency, they still face some technical problems that need to be solved urgently. First, the complexity of heat transfer problems lies in the diversity of their boundary conditions and source terms. Especially in complex geometries and non-uniform media, existing deep learning frameworks are difficult to adapt to the changes in multiple boundary conditions and source terms simultaneously, resulting in limited generalization ability of the model. Second, when traditional methods are used to handle heat transfer problems in three-dimensional complex domains, they often require a large amount of computational resources, especially in cases where multiple solutions are needed, such as parameter optimization or uncertainty analysis, which limits their application efficiency in practical engineering. Finally, most existing methods lack interpretability of physical processes and it is difficult to directly extract physical insights from the model, which to some extent hinders the wide application of deep learning technology in the field of heat transfer.

[0047] Therefore, the embodiments of the present application provide a heat transfer solution method, device, equipment and storage medium with variable physical conditions. The method includes obtaining the physical conditions of a heat transfer object; inputting the physical conditions into a trained solution network model; determining the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model; determining an integral solution expression based on the Green's function and the Green's function gradient, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression. The present application combines the integral form of the Green's function and deep learning technology, converts the heat transfer problem into the integral form of the Green's function, and then determines the Green's function and the Green's function gradient in the integral form of the Green's function through a solution network model constructed by a deep learning network. Finally, the heat transfer solution result is determined based on the Green's function and the Green's function gradient. In this way, the analytical characteristics of the Green's function can be used to reduce the data requirements for training the solution network model, thereby reducing the cost of solving heat transfer problems. At the same time, by combining the integral form of the Green's function and deep learning technology, the solution efficiency and accuracy of heat transfer problems can be improved, which can better meet the real-time heat transfer analysis requirements.

[0048] The following further illustrates the content of the application through the description of embodiments in conjunction with the accompanying drawings.

[0049] This embodiment provides a heat transfer solution method with variable physical conditions, and the heat transfer solution method can be applied to the heat conduction of an electronic heat dissipation device. That is to say, in an application scenario of this application, the electronic heat dissipation device can be used as the target with variable physical conditions, and then the physical conditions of the electronic heat dissipation device are obtained (the physical conditions can be collected under different heat sources and different heat exchange conditions of the electronic heat dissipation device), the physical conditions are input into the trained solution network model, the Green's function and the Green's function gradient corresponding to the heat transfer object are determined through the solution network model, and finally the integral solution expression is determined based on the Green's function and the Green's function gradient, and the heat transfer solution result corresponding to the heat transfer object is determined through the integral solution expression. In this way, the heat exchange capacity of the electronic heat dissipation device under different heat sources and different heat exchange conditions can be quickly detected, and the design research of the electronic heat dissipation device can be accelerated.

[0050] As Figure 1 shown, a heat transfer solution method with variable physical conditions provided by an embodiment of this application specifically includes:

[0051] S10. Obtain the physical conditions of the heat transfer object.

[0052] Specifically, the heat transfer object is a target with heat transfer ability. For example, an electronic heat dissipation device, a heat exchanger, etc. And the entire heat transfer object can be used as the domain to be solved, or a local area of the heat transfer object can be used as the domain to be solved, and then heat transfer solution is performed on the domain to be solved. For example, the heat transfer object can be divided into grid regions through grid division, and then a local area composed of one grid region or multiple continuous grid regions is used as a domain to be solved, or multiple local areas composed of one grid region or multiple continuous grid regions are synchronously selected, and each selected local area is used as a domain to be solved to solve and compare multiple concerned regions, etc.

[0053] The physical conditions are used to reflect the physical properties of the transfer object, and the physical conditions can be input in the form of a graphical interface or input text. The physical conditions may include geometric shape data, boundary condition data, material property data, and source term distribution data. Among them, the geometric shape data is used to provide the calculation domain of the area to be solved, which can be a two-dimensional or three-dimensional model and supports user-defined input. The boundary condition data is used to describe the thermal behavior on the system boundary. The boundary conditions include the first kind of boundary condition, the second kind of boundary condition, and the third kind of boundary condition. The first kind of boundary condition is the temperature on the boundary, the second kind of boundary condition is the heat flux density on the boundary, and the third kind of boundary condition is that the boundary exchanges heat with the surrounding fluid through heat convection. Of course, the boundary conditions can also include that the boundary exchanges heat with the environment through thermal radiation, etc. The material property data is used to describe the physical properties of the material of the heat transfer object. For example, thermal conductivity, specific heat capacity, thermal diffusivity, etc.

[0054] S20. Input the physical conditions into the trained solution network model, and determine the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model.

[0055] Specifically, the solution network model is a deep learning network constructed based on the integral form of the Green's function. The Green's function can transform heat transfer problems (such as heat source distribution and boundary condition problems, etc.) into integral forms to simplify calculations, thereby significantly reducing data requirements and improving the model learning efficiency. At the same time, the present application combines the Green's function with a deep learning network to improve the solution efficiency and solution accuracy of heat transfer problems through a deep learning architecture, enabling second-level response and high-precision prediction to meet the needs of real-time heat transfer analysis.

[0056] Furthermore, the Green's function is the impulse response solution of a linear partial differential equation and satisfies the following equation:

[0057] ,

[0058] wherein, represents the coordinate point in the computational domain, represents the integration point, represents the linear differential operator, represents the Dirac function, represents the computational domain, represents the boundary.

[0059] Based on the Green's function, the solution of the heat transfer problem in integral form can be expressed as:

[0060] ,

[0061] wherein, represents the solution of the heat transfer problem, represents the source term, represents the first kind of boundary condition, represents the second and third kinds of boundary conditions, represents the boundary weight function, represents the outer normal vector of the boundary, represents the integration point on the boundary.

[0062] It should be noted that for multi-physics problems, the Green's function can be extended to a matrix form to support the solution of heat transfer problems in complex systems, that is, to support representing the solution of heat transfer problems in complex systems in integral form. At the same time, different types of heat transfer problems can be transformed into integral form through the Green's function. For example, steady-state and unsteady heat conduction, convective heat transfer, etc. That is to say, the heat transfer solution method with variable physical conditions in the system of the embodiments of the present application can solve different types of heat transfer problems, such as steady-state and unsteady heat conduction problems, convective heat transfer problems, etc.

[0063] Based on this, when combining the Green's function with a deep learning structure to construct a solution network model, a function approximation module and a Green's function gradient can be constructed. The main network is used to approximate the Green's function, and the branch network is used to approximate the gradient of the Green's function. Correspondingly, the construction process of the solution network model specifically includes:

[0064] Construct an integral solution expression for the heat transfer solution problem with variable object conditions based on the Green's function, where the integral solution expression is used to determine the solution of the heat transfer solution problem with variable object conditions;

[0065] Construct a function approximation module for approximating the Green's function in the integral solution expression and a gradient approximation module for approximating the gradient of the Green's function in the integral solution expression;

[0066] Construct a to-be-trained solution network model based on the function approximation module and the Green's function gradient, and train the to-be-trained solution network model to obtain a solution network model.

[0067] Specifically, both the function approximation module and the gradient approximation module can adopt a binary tree network. Multiple sub-networks are used to learn local features respectively, and finally combined into a complete solution. This structure not only improves the convergence speed of the model, but also enhances the ability to capture the singularities of the Green's function, thereby improving the efficiency and accuracy of the model. Among them, when solving heat transfer problems through the solution network model, as Figure 2 shown, the Green's function and the gradient of the Green's function corresponding to the solution of each coordinate point can be solved through the solution network, and then the solution of this coordinate point is determined based on the Green's function and the gradient of the Green's function. To this end, when determining the solution of each coordinate point through the solution network model, the position information of this coordinate point and the integration points can be input into the function approximation module and the gradient approximation module. The Green's function value and the gradient of the Green's function corresponding to this coordinate point are output through the function approximation module to determine the solution of this coordinate point. In this way, the solution of each coordinate point in the heat transfer object can be determined, and then the heat transfer solution result of the heat transfer object can be determined.

[0068] In the embodiments of the present application, a deep learning model is constructed based on an integral solution expression to obtain a solution network model for predicting an approximate solution of the Green's function. The solution network model can automatically extract the features of the heat transfer problem, learn the mapping relationship between the input data and the output solution, so as to realize the rapid solution of the heat transfer problem. At the same time, the solution network model also supports the optimization and adjustment of the model, improves the accuracy of the solution network model, and can directly adapt to complex heat transfer conditions without retraining, improving the generalization ability of the solution network model. In addition, from the construction process of the solution network model, it can be seen that the solution network model has concise model parameters and does not require a large amount of technical resources, enabling the solution network model to support multi-platform deployment and reducing the deployment cost of the solution network model; and it simplifies the operation process and can be used without professional knowledge, reducing the usage threshold of the solution network model.

[0069] In one implementation, training the solution network model to be trained to obtain the solution network model may include:

[0070] Obtain a training data set;

[0071] Input the training data in the training data set into the solution network model to be trained, and output a predicted solution through the solution network model to be trained;

[0072] Construct a loss function based on the predicted solution and the true solution corresponding to the training data, and train the solution network model to be trained based on the loss function to obtain the solution network model.

[0073] Specifically, the training data set includes a number of training data, and each training data corresponds to a true solution. Among them, a number of training data can all be obtained through data simulation, or all be obtained through experiments, or part be obtained through data simulation and part be obtained through experiments, which is not specifically limited here. In the present application, by using data simulation or experimental methods to obtain training data, the diversity and accuracy of the data can be ensured.

[0074] In addition, when training the solution network model to be trained, various training algorithms and optimization strategies can be used. For example, the Adam optimizer or other optimization algorithms can be used to optimize the model parameters by minimizing the error between the predicted solution and the true solution. Correspondingly, the loss function can be expressed as:

[0075] ,

[0076] where represents the predicted solution, represents the true solution, represents the number of training data.

[0077] S30. Determine an integral solution expression based on the Green's function and the gradient of the Green's function, and determine the heat transfer solution result corresponding to the heat transfer object through the integral solution expression.

[0078] Specifically, after obtaining the Green's function and the gradient of the Green's function, an integral solution expression can be determined based on the Green's function and the gradient of the Green's function. The integral solution expression can be a discrete form of the solution to the heat transfer problem. That is to say, the solution network model can be discretized from the integral form of the solution to the heat transfer problem determined based on the Green's function, that is, the integral solution expression can be expressed as:

[0079] ,

[0080] where, represents the solution to the heat transfer problem, represents the position information of the heat transfer object, represents the volume discretization weight, represents the area discretization weight, represents the discrete number of the source term, represents integration points, represents the th source term of the integration point, represents the th Green's function of the integration point, represents the discrete number of the boundary conditions, represents integration points, represents the th first boundary condition of the integration point, represents the th Green's function of the integration point, represents the th gradient of the Green's function of the integration point, represents the th second and third type boundary conditions of the integration point, represents the th boundary weight function of the integration point, represents the th boundary outer normal vector of the integration point.

[0081] Further, the representation form of the heat transfer solution result can be determined according to user requirements. For example, the heat transfer solution result is a temperature field distribution, heat flux density, isotherm diagram, heat flux vector diagram, etc., and the heat transfer solution result can also include multiple forms among the temperature field distribution, heat flux density, isotherm diagram, and heat flux vector diagram. In addition, after obtaining the heat transfer solution result, the heat transfer solution result can also be displayed in a graphical interface. Before displaying the heat transfer solution result in a graphical interface, the heat transfer solution result can also be analyzed, and the analysis result can be synchronously displayed with the heat transfer solution result. For example, the heat transfer solution result is displayed in a graphical interface as Figure 3 shown. Of course, after obtaining the heat transfer solution result, the heat transfer solution result can be saved or exported in a common format (such as CSV, VTK, etc.) for further research and analysis by users.

[0082] In summary, this embodiment provides a heat transfer solution method with variable physical conditions. The method includes obtaining the physical conditions of a heat transfer object; inputting the physical conditions into a trained solution network model; determining the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model; determining an integral solution expression based on the Green's function and the Green's function gradient, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression. This application combines the integral form of the Green's function and deep learning technology, converts the heat transfer problem into the integral form of the Green's function, then determines the Green's function and the Green's function gradient in the integral form of the Green's function through a solution network model constructed by a deep learning network, and finally determines the heat transfer solution result based on the Green's function and the Green's function gradient. In this way, the analytical characteristics of the Green's function can be utilized to reduce the data requirements for training the solution network model, thereby reducing the cost of solving the heat transfer problem. At the same time, by combining the integral form of the Green's function and deep learning technology, the solution efficiency and solution accuracy of the heat transfer problem can also be improved, which can better meet the real-time heat transfer analysis requirements.

[0083] At the same time, through actual test verification of the heat transfer solution method with variable physical conditions provided in this application embodiment, it is concluded through actual test verification that the advantages of the heat transfer solution method with variable physical conditions of this application are: 1. It maintains stable performance under different geometric shapes, boundary conditions, and material properties, and the overall prediction accuracy exceeds 80%; 2. The solution time for a single heat transfer problem is about 1 s, and only conventional GPU computing resources are required to achieve it; 3. It significantly reduces the dependence on high-performance computing devices and realizes fast prediction while ensuring high accuracy.

[0084] Based on the above heat transfer solution method with variable physical conditions, this embodiment provides a heat transfer solution device with variable physical conditions, as Figure 4 shown. The heat transfer solution device with variable physical conditions specifically includes:

[0085] The heat transfer problem determination module 100 is configured to obtain the physical conditions of the heat transfer object;

[0086] The deep learning model module 200 is configured to input the physical conditions into a trained solution network model, and determine the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model, where the solution network model is a deep learning network constructed based on the integral form of the Green's function, and includes a function approximation module and a gradient approximation module;

[0087] The heat transfer solution module 300 is configured to determine an integral solution expression based on the Green's function and the Green's function gradient, and determine the heat transfer solution result corresponding to the heat transfer object through the integral solution expression.

[0088] Based on the above heat transfer solution method with variable physical conditions, this embodiment provides a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the heat transfer solution method with variable physical conditions as described in the above embodiment.

[0089] Based on the above heat transfer solution method with variable physical conditions, the present application further provides a terminal device, as Figure 5 shown, which includes at least one processor 20; a display screen 21; and a memory 22, and may further include a communication interface 23 and a bus 24. Among them, the processor 20, the display screen 21, the memory 22, and the communication interface 23 can complete mutual communication through the bus 24. The display screen 21 is set to display a preset user guidance interface in the initial setting mode. The communication interface 23 can transmit information. The processor 20 can call the logical instructions in the memory 22 to execute the method in the above embodiment.

[0090] In addition, when the logical instructions in the above memory 22 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0091] The memory 22, as a computer-readable storage medium, can be set to store software programs and computer-executable programs, such as the program instructions or modules corresponding to the method in the embodiment of the present disclosure. The processor 20 executes functional applications and data processing by running the software programs, instructions, or modules stored in the memory 22, that is, implements the method in the above embodiment.

[0092] The memory 22 may include a program storage area and a data storage area. Among them, the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory 22 may include a high-speed random access memory and may also include a non-volatile memory. For example, various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, may also be a transient storage medium.

[0093] In addition, the specific processes of loading and executing multiple instructions in the above storage medium and the terminal device have been described in detail in the above method, and will not be repeated here one by one.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A heat transfer solution method for varying physical conditions, characterized in that The heat transfer solution method for varying physical conditions specifically includes: Obtaining the physical conditions of the heat transfer object; Inputting the physical conditions into a trained solution network model, and determining the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model, where the solution network model is a deep learning network constructed based on the integral form of the Green's function, and includes a function approximation module and a gradient approximation module; Determining an integral solution expression based on the Green's function and the Green's function gradient, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression; The integral solution expression is: ， Among them, represents the solution to the heat transfer problem, represents the position information of the heat transfer object, represents the volume discretization weight, represents the area discretization weight, represents the discrete number of the source term, represents integration points, represents the th integration point of the source term, represents the th integration point of the Green's function, represents the discrete number of the boundary conditions, represents integration points, represents the th integration point of the first boundary condition, represents the th integration point of the Green's function, represents the th integration point of the gradient of the Green's function, represents the th integration point of the second and third types of boundary conditions, represents the th integration point of the boundary weight function, represents the th integration point of the outer normal vector of the boundary.

2. The heat transfer solution method for varying physical conditions according to claim 1, characterized in that The physical conditions include geometric shape data, boundary condition data, material property data, and source term distribution data.

3. The heat transfer solution method for changing physical conditions according to claim 1, characterized in that, The construction process of the solution network model specifically includes: Constructing an integral solution expression for the heat transfer solution problem with varying object conditions based on the Green's function, where the integral solution expression is used to determine the solution of the heat transfer solution problem with varying object conditions; Constructing a function approximation module for approximating the Green's function in the integral solution expression and a gradient approximation module for approximating the Green's function gradient in the integral solution expression; Constructing a solution network model to be trained based on the function approximation module and the Green's function gradient, and training the solution network model to be trained to obtain a solution network model.

4. The heat transfer solution method for varying physical conditions according to claim 3, characterized in that The training of the solution network model to be trained to obtain a solution network model specifically includes: Obtaining a training data set, where the training data in the training data set is obtained through data simulation or experiments, and includes a number of training data, and each training data corresponds to a true solution; Inputting the training data in the training data set into the solution network model to be trained, and outputting a predicted solution through the solution network model to be trained; Constructing a loss function based on the predicted solution and the true solution corresponding to the training data, and training the solution network model to be trained based on the loss function to obtain a solution network model.

5. The heat transfer solution method for changing physical conditions according to claim 1, characterized in that The heat transfer solution result is one or more of a temperature field distribution, a heat flux density, a temperature line graph, and a heat flux vector graph.

6. The heat transfer solution method for changing physical conditions according to claim 1, wherein After determining the integral solution expression based on the Green's function and the Green's function gradient, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression, the method further includes: Displaying the heat transfer solution result through a graphical interface.

7. A heat transfer solution device for changing physical conditions, characterized in that, The heat transfer solution device for varying physical conditions specifically includes: A heat transfer problem determination module for obtaining the physical conditions of the heat transfer object; A deep learning model module for inputting the physical conditions into a trained solution network model, and determining the Green's function and the Green's function gradient corresponding to the heat transfer object through the solution network model, where the solution network model is a deep learning network constructed based on the integral form of the Green's function, and includes a function approximation module and a gradient approximation module; A heat transfer solution module for determining an integral solution expression based on the Green's function and the Green's function gradient, and determining the heat transfer solution result corresponding to the heat transfer object through the integral solution expression; The integral solution expression is: ， Among them, represents the solution to the heat transfer problem, represents the position information of the heat transfer object, represents the volume discretization weight, represents the area discretization weight, represents the discrete quantity of the source term, represents integration points, represents the th integration point of the source term, represents the th integration point of the Green's function, represents the discrete quantity of the boundary conditions, represents integration points, represents the th integration point of the first boundary condition, represents the th integration point of the Green's function, represents the th integration point of the gradient of the Green's function, represents the th integration point of the second and third type boundary conditions, represents the th integration point of the boundary weight function, represents the th integration point of the outer normal vector of the boundary.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the heat transfer solution method for varying physical conditions as described in any one of claims 1-6.

9. A terminal device, characterized in that, Including: A processor and a memory; A computer-readable program executable by the processor is stored on the memory; When the processor executes the computer-readable program, the steps in the heat transfer solution method for varying physical conditions as described in any one of claims 1-6 are implemented.

Images