Prediction method of thermal stress and thermal deformation of layout board based on deep learning technology

Abstract

The present invention discloses a method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology, comprising: constructing a mathematical model for layout optimization design of the layout board based on the structure and layout of the layout board; obtaining a training data set based on the mathematical model, wherein the training data includes the temperature field of the layout board and its corresponding thermal stress and thermal deformation; constructing a deep learning model; training the deep learning model using the training data set to fit the mapping relationship between the temperature field, thermal stress, and thermal deformation of the layout board; and predicting the thermal stress and thermal deformation of the layout board using the trained deep learning model. The present invention can achieve rapid and accurate calculation of the thermal stress and thermal deformation corresponding to the temperature fields of different layout boards, greatly reducing the amount of calculation required for optimizing the layout board layout when considering the performance requirements of the layout board's structural stability and working accuracy stability under large temperature differences, thereby reducing the computational cost and time cost required for optimizing the layout board layout.

Description

Technical Field

[0001] The present invention relates to the technical field of component layout board design, and in particular to a method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology. Background Art

[0002] With the rapid development of space technology and industrialization, satellite design is currently facing a series of goals, including shortening design cycles, reducing development costs, ensuring design reliability, and achieving standardization, serialization, and generalization. Since the layout design of satellite components is a key part of the overall satellite design, achieving these goals requires the ability to achieve a "good, fast, and cost-effective" component layout design for the satellite component layout board during satellite design.

[0003] The design of a satellite component layout board requires comprehensive consideration of design requirements across multiple disciplines, including temperature fields, electromagnetic fields, and mass characteristics. This represents a typical complex multidisciplinary optimization design problem. Traditional layout design relies heavily on prior human experience to arrive at a relatively reasonable design. After a preliminary design is finalized, ground-based simulated space testing and simulation calculations are conducted to verify that it meets structural and operational stability requirements. If not, the layout must be redesigned, followed by a performance review, and this cycle continues.

[0004] The optimal design of satellite layout board assembly layouts requires extensive simulation of thermal deformation and thermal stress fields when considering the structural stability and operational accuracy requirements under wide temperature fluctuations. However, since calculating these thermal deformation and stress fields requires a complex finite element analysis process, each simulation is costly and time-consuming. Considering both thermal deformation and thermal stress during the optimization process requires numerous simulation program calls, significantly increasing the computational and time-consuming cost of solving the satellite layout optimization problem. This can even make it impossible to achieve the desired design results within the required timeframe. Summary of the Invention

[0005] In order to solve some or all of the technical problems existing in the above-mentioned prior art, the present invention provides a method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology.

[0006] The technical solutions of the present invention are as follows:

[0007] A method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology is provided, the method comprising:

[0008] According to the structure and layout of the layout board, a mathematical model for the layout optimization design of the layout board is constructed;

[0009] According to the mathematical model, a training data set is obtained, wherein the training data includes a temperature field of the layout board and its corresponding thermal stress and thermal deformation;

[0010] Build deep learning models;

[0011] Using the training dataset, a deep learning model is trained to fit the mapping relationship between the temperature field, thermal stress, and thermal deformation of the layout board.

[0012] The trained deep learning model is used to predict the thermal stress and thermal deformation of the layout board.

[0013] In some possible implementations, each component within the layout area of ​​the layout board is considered as a heat source, and the mathematical model for the layout optimization design of the layout board is:

[0014]

[0015] Where X represents the component layout scheme within the layout area of ​​the layout board, (x i ,y i ) represents the position coordinates of the i-th component in the layout area, N s represents the number of components, f1(X) and f2(X) represent the objective function, σ equivalent represents the equivalent thermal stress, u equivalent represents the equivalent thermal strain, (x, y) represents the position coordinates of a point in the layout area, T represents the temperature field in the layout area of ​​the layout board, φ i represents the strength of the i-th component, T0 represents the temperature value on the layout area boundary, k represents the thermal conductivity in the layout area, n represents the normal vector of the layout area boundary, h represents the surface heat transfer coefficient between the object on the layout area boundary and the surrounding fluid, u x and u y denote the thermal displacement in the x-direction and y-direction, (σ x ,σ y ,σ xy ) represents the thermal stress in three directions, E represents Young's modulus, α represents the linear expansion coefficient, and μ represents Poisson's ratio.

[0016] In some possible implementations, the training dataset is obtained in the following manner:

[0017] The layout area of ​​the layout board is meshed, and a corresponding number of grids are selected based on the number of components to place the components to obtain a layout scheme. The temperature field of the layout board under the current layout scheme is calculated using the finite element analysis method. Based on the obtained temperature field of the layout board and the constraint equations in the mathematical model, the thermal stress and thermal deformation corresponding to the temperature field of the current layout board are calculated using the finite element analysis method, thereby obtaining training data including the temperature field of the layout board and its corresponding thermal stress and thermal deformation;

[0018] The training data acquisition process is repeated until a preset amount of training data is obtained, thereby obtaining a training data set including the preset amount of training data.

[0019] In some possible implementations, a deep learning model is trained using a training dataset, including:

[0020] The temperature field of the layout board in the training data is used as input, and the thermal stress and thermal deformation in the training data are used as output to train the deep learning model.

[0021] In some possible implementations, the temperature field of the layout board in the training data is used as input, and the thermal stress and thermal deformation in the training data are used as output to train the deep learning model, including:

[0022] Step S401: inputting the temperature fields of the layout boards in the plurality of training data into the deep learning model in sequence to obtain the predicted thermal stress and predicted thermal deformation output by the deep learning model;

[0023] Step S402: comparing the predicted thermal stress and predicted thermal deformation output by the deep learning model with the thermal stress and thermal deformation in the training data, and calculating the prediction accuracy of the deep learning model;

[0024] Step S403, determine whether the prediction accuracy obtained for at least two consecutive times is greater than the preset accuracy threshold. If so, use the current deep learning model as the deep learning model to complete the training. If not, adjust the parameters of the deep learning model and return to step S401.

[0025] In some possible implementations, the method further includes:

[0026] Before using the training data set to train the deep learning model, the temperature field in the training data is normalized, and the thermal stress and thermal deformation in the training data are normalized.

[0027] In some possible implementations, the method further includes:

[0028] Build a layout optimization design problem for a layout board that integrates deep learning models;

[0029] Solve the layout optimization design problem of the layout board and obtain a layout solution of the layout board that can minimize the maximum equivalent thermal stress and the maximum equivalent thermal strain in the layout area of ​​the layout board.

[0030] In some possible implementations, the layout optimization design problem for the layout board integrated with the deep learning model is:

[0031]

[0032] Where X represents the component layout scheme within the layout area of ​​the layout board, f1(X) and f2(X) represent the objective functions, and σ equivalent represents the equivalent thermal stress, u equivalent represents the equivalent thermal strain, T represents the temperature field of the layout area of ​​the layout board, Indicates the corresponding relationship between the component layout scheme and the temperature field, Represents the predicted thermal stress and predicted thermal deformation corresponding to the temperature field T output by the deep learning model, It represents the corresponding relationship between thermal stress and thermal deformation and equivalent thermal stress and equivalent thermal strain.

[0033] The main advantages of the technical solution of the present invention are as follows:

[0034] The layout board thermal stress and thermal deformation prediction method based on deep learning technology of the present invention obtains training data including the temperature field of the layout board and its corresponding thermal stress and thermal deformation according to the mathematical model satisfied by the layout optimization design of the layout board, and uses the training data to train the deep learning model. The deep learning model can be used to realize fast and accurate calculation of the thermal stress and thermal deformation corresponding to the temperature field of different layout boards, greatly reducing the amount of calculation required for the layout optimization solution of the layout board when considering the performance requirements of the structural stability and working accuracy stability of the layout board under large temperature difference conditions, reducing the calculation cost and time cost required for the layout optimization solution of the layout board, and improving the efficiency of the layout optimization design of the layout board. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0036] Figure 1 This is a flowchart of a method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present invention and corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0038] The technical solutions provided by the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0039] refer to Figure 1 An embodiment of the present invention provides a method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology. The method includes the following steps S1-S5:

[0040] Step S1: constructing a mathematical model for layout optimization design of the layout board according to the structure and layout of the layout board.

[0041] In one embodiment of the present invention, the layout area of ​​the layout board can be regarded as a two-dimensional plane. When considering the layout optimization design problem of the layout board components such as thermal stress and thermal deformation and working stability indicators, each component on the layout board can be simplified as a heat source. The component layout optimization design of the layout board can be regarded as the layout optimization design of the heat source, that is, a certain number of heat source components are placed in a certain layout area, and by optimizing the position of the heat source components, the maximum equivalent thermal stress and the maximum equivalent thermal strain in the layout area are made as small as possible.

[0042] Specifically, the structure and layout of the layout board are assumed to meet the following conditions: the layout area of ​​the layout board is a two-dimensional plane, and each component within the layout area of ​​the layout board can be regarded as a heat source. Then, the mathematical model for constructing the layout optimization design of the layout board is constructed as follows:

[0043]

[0044] Where X represents the component layout scheme within the layout area of ​​the layout board, (x i ,y i ) represents the position coordinates of the i-th component in the layout area, N s represents the number of components, f1(X) and f2(X) represent the objective function, σ equivalent represents the equivalent thermal stress, u equivalent represents the equivalent thermal strain, (x, y) represents the position coordinates of a point in the layout area, T represents the temperature field in the layout area of ​​the layout board, φ i represents the strength of the i-th component, φ i Determined according to the actual working power of the component, T0 represents the temperature value on the boundary of the layout area, k represents the thermal conductivity in the layout area, n represents the normal vector of the layout area boundary, h represents the surface heat transfer coefficient between the object on the boundary of the layout area and the surrounding fluid, u x and u y denote the thermal displacement in the x-direction and y-direction, (σ x ,σ y ,σ xy ) represents the thermal stress in three directions, E represents Young's modulus, α represents the linear expansion coefficient, μ represents Poisson's ratio, (σ x ,σ y ,σxy ) constitutes the thermal stress of the layout area of ​​the layout board, (u x ,u y ) constitutes the thermal deformation of the layout area of ​​the layout board, and the equivalent thermal stress and equivalent thermal strain of the layout area are obtained by processing the thermal stress and thermal deformation of the layout area.

[0045] The two-dimensional coordinate system of the layout area can be set according to actual conditions.

[0046] In the above mathematical model, T=T0 represents the first type of boundary condition (Dirichlet boundary condition), represents the second type of boundary condition (Neumann boundary condition), The third type of boundary conditions (Robin boundary conditions) are shown. The boundary conditions satisfied by the layout area of ​​the layout board are determined according to the actual situation.

[0047] Represents the heat conduction equation satisfied by the layout area of ​​the layout board.

[0048] Represents the thermoelastic equation satisfied by the layout area of ​​the layout board.

[0049] Represents the thermal balance equation satisfied by the layout area of ​​the layout board.

[0050] The above-mentioned heat conduction equation, boundary conditions, thermoelasticity equation and thermal balance equation constitute the constraints when solving the layout optimization design of the layout board.

[0051] Step S2: obtaining a training data set according to the mathematical model, wherein the training data includes the temperature field of the layout board and its corresponding thermal stress and thermal deformation.

[0052] In one embodiment of the present invention, the training data set is obtained in the following manner:

[0053] The layout area of ​​the layout board is meshed, and a corresponding number of grids are selected based on the number of components to place the components to obtain a layout scheme. The temperature field of the layout board under the current layout scheme is calculated using the finite element analysis method. Based on the obtained temperature field of the layout board and the constraint equations in the mathematical model, the thermal stress and thermal deformation corresponding to the temperature field of the current layout board are calculated using the finite element analysis method, thereby obtaining training data including the temperature field of the layout board and its corresponding thermal stress and thermal deformation;

[0054] The training data acquisition process is repeated until a preset amount of training data is obtained, thereby obtaining a training data set including the preset amount of training data.

[0055] In one embodiment of the present invention, the training data is represented as (T, σ x ,σ y ,σ xy ,u x ,u y ), T represents the temperature field of the layout board, (σ x ,σ y ,σ xy ) represents thermal stress, (u x ,u y ) indicates thermal deformation.

[0056] In one embodiment of the present invention, the amount of training data is determined according to actual needs, for example, it can be 10000. Generally speaking, the more training data there is, the higher the prediction accuracy of the trained deep learning model.

[0057] Furthermore, in an embodiment of the present invention, when the layout area of ​​the layout board is divided into grids, the number of divided grids is set according to actual conditions so that each grid can just accommodate one component.

[0058] Furthermore, in one embodiment of the present invention, when calculating the temperature field of the layout board and its corresponding thermal stress and thermal deformation, the layout area can be divided into multiple finite element calculation grids. Based on the divided finite element calculation grids, the finite element analysis method is sequentially used to calculate the temperature field of the layout board and its corresponding thermal stress and thermal deformation under the current layout scheme. The number of divided finite element calculation grids is set according to actual conditions. The greater the number of divided finite element calculation grids, the higher the accuracy of the calculated temperature field and its corresponding thermal stress and thermal deformation, but the required calculation cost and time cost are also higher.

[0059] Step S3: Build a deep learning model.

[0060] In one embodiment of the present invention, the deep learning model adopts the U-Net network model.

[0061] Step S4: Using the training data set to train a deep learning model to fit the mapping relationship between the temperature field and thermal stress and thermal deformation of the layout board.

[0062] In one embodiment of the present invention, training a deep learning model using a training data set includes:

[0063] The temperature field of the layout board in the training data is used as input, and the thermal stress and thermal deformation in the training data are used as output to train the deep learning model.

[0064] Furthermore, in one embodiment of the present invention, the temperature field of the layout board in the training data is used as input, and the thermal stress and thermal deformation in the training data are used as output to train the deep learning model, including the following steps S401-S403:

[0065] In step S401 , the temperature fields of the layout boards in the plurality of training data are sequentially input into the deep learning model to obtain the predicted thermal stress and predicted thermal deformation output by the deep learning model.

[0066] In one embodiment of the present invention, the temperature field of the layout board in the training data is input from the input end of the deep learning model, processed in sequence by the parameters of each layer in the deep learning model, and output from the output end of the deep learning model. The information output from the output end is the predicted thermal stress and predicted thermal deformation corresponding to the temperature field of the layout board.

[0067] Step S402 : Compare the predicted thermal stress and predicted thermal deformation output by the deep learning model with the thermal stress and thermal deformation in the training data, and calculate the prediction accuracy of the deep learning model.

[0068] In one embodiment of the present invention, the difference between the predicted thermal stress and predicted thermal deformation corresponding to each training data and the thermal stress and thermal deformation in the training data, and the ratio of the thermal stress and thermal deformation in the training data can be calculated, and the average value of all ratios can be used as the prediction accuracy.

[0069] Step S403, determine whether the prediction accuracy obtained for at least two consecutive times is greater than the preset accuracy threshold. If so, use the current deep learning model as the deep learning model to complete the training. If not, adjust the parameters of the deep learning model and return to step S401.

[0070] In one embodiment of the present invention, if the prediction accuracy obtained at least twice in a row exceeds a preset accuracy threshold, the current deep learning model is deemed to have completed training, and deep learning model training ends. If the prediction accuracy obtained at least twice in a row is not greater than the preset accuracy threshold, the parameters of the deep learning model are adjusted and updated, and deep learning model training continues based on the adjusted and updated deep learning model.

[0071] In one embodiment of the present invention, by using the acquired training dataset and the aforementioned training method to train a deep learning model, the deep learning model can learn the underlying physical laws, ultimately resulting in a deep learning model with strong generalization capabilities. This model can quickly and accurately predict the thermal stress and thermal deformation corresponding to the temperature field of any layout board. In this case, the deep learning model is a proxy model.

[0072] Furthermore, in one embodiment of the present invention, in the method, step S4 further includes:

[0073] Before using the training data set to train the deep learning model, the temperature field in the training data is normalized, and the thermal stress and thermal deformation in the training data are normalized.

[0074] Specifically, the temperature field values ​​are normalized to [0, 1], the thermal stress values ​​are normalized to [0, 1], and the thermal deformation values ​​are normalized to [0, 1].

[0075] Because the output of deep learning models involves thermal stress and thermal deformation, which is a multi-task learning problem, balancing the temperature field, thermal stress, and thermal deformation in the training data is difficult during training if they are not on the same order of magnitude. By normalizing the temperature field, thermal stress, and thermal deformation data to [0, 1], the training speed and accuracy of multiple regression tasks can be adjusted, overcoming the "seesaw" phenomenon that often occurs when training multiple tasks simultaneously.

[0076] Among them, after normalization, during training, the loss function during training can be weighted according to the characteristics of each sub-task.

[0077] Step S5: Use the trained deep learning model to predict the thermal stress and thermal deformation of the layout board.

[0078] Specifically, when predicting thermal stress and thermal deformation of a layout board, if the board's temperature field is known, the board's temperature field can be directly input into the deep learning model to obtain the corresponding thermal stress and thermal deformation output by the deep learning model. If only the board's layout scheme is known, the board's temperature field under the current scheme can be calculated using finite element analysis. The calculated layout board temperature field is then input into the deep learning model to obtain the corresponding thermal stress and thermal deformation output by the deep learning model.

[0079] Furthermore, in one embodiment of the present invention, the method further comprises the following steps:

[0080] Step S6, constructing a layout board layout optimization design problem integrating a deep learning model;

[0081] Step S7 , solving the layout optimization design problem of the layout board, and obtaining a layout solution of the layout board that can minimize the maximum equivalent thermal stress and the maximum equivalent thermal strain in the layout area of ​​the layout board.

[0082] Specifically, based on the mathematical model of layout board optimization design constructed above, the temperature field of the layout board is solved by finite element analysis, and the deep learning model obtained after training is used to replace the finite element method to solve the thermoelastic equation and the thermal balance equation. The layout board optimization design problem integrated with the deep learning model is expressed as:

[0083]

[0084] Where X represents the component layout scheme within the layout area of ​​the layout board, f1(X) and f2(X) represent the objective functions, and σ equivalent represents the equivalent thermal stress, u equivalent represents the equivalent thermal strain, T represents the temperature field of the layout area of ​​the layout board, Indicates the corresponding relationship between the component layout scheme and the temperature field, Represents the predicted thermal stress and predicted thermal deformation corresponding to the temperature field T output by the deep learning model, It represents the corresponding relationship between thermal stress and thermal deformation and equivalent thermal stress and equivalent thermal strain. It can be expressed as

[0085] Furthermore, let’s take the layout optimization problem of the layout board as an input of the layout matrix M1×M2, and the minimum moving distance of the component as the length of a grid. The grid position of the lower left corner vertex of the component is selected as the design variable, and the position coordinate of the i-th component is recorded as (x i ,y i ), then (x i ,y i ) must meet the following requirements:

[0086]

[0087] It can be seen that the above optimization problem is a discrete integer multi-objective optimization problem. By selecting a suitable optimization algorithm to solve the optimization problem, the component layout scheme of the layout board with the optimal structural stability and working stability performance indicators can be obtained within limited computing resources.

[0088] An embodiment of the present invention provides a method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology. The method obtains training data including the temperature field of the layout board and its corresponding thermal stress and thermal deformation according to a mathematical model satisfied by the layout optimization design of the layout board, and uses the training data to train a deep learning model. The method can use the deep learning model to realize fast and accurate calculation of the thermal stress and thermal deformation corresponding to the temperature fields of different layout boards, greatly reducing the amount of calculation required for solving the layout optimization of the layout board when considering the performance requirements of the structural stability of the layout board and the stability of the working accuracy under large temperature difference conditions, reducing the computational cost and time cost required for solving the layout optimization of the layout board, and improving the efficiency of the layout optimization design of the layout board.

[0089] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In addition, "front", "back", "left", "right", "upper" and "lower" in this document are all referenced to the placement states shown in the accompanying drawings.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology, characterized in that: include: According to the structure and layout of the layout board, a mathematical model for the layout optimization design of the layout board is constructed; According to the mathematical model, a training data set is obtained, wherein the training data includes a temperature field of the layout board and its corresponding thermal stress and thermal deformation; Build deep learning models; Using the training dataset, a deep learning model is trained to fit the mapping relationship between the temperature field, thermal stress, and thermal deformation of the layout board. Use the trained deep learning model to predict thermal stress and thermal deformation of the layout board; Assuming that each component in the layout area of ​​the layout board is regarded as a heat source, the mathematical model of the layout optimization design of the layout board is: Where X represents the component layout scheme within the layout area of ​​the layout board, (x i ,y i ) represents the position coordinates of the i-th component in the layout area, N s represents the number of components, f1(X) and f2(X) represent the objective function, σ equivalent represents the equivalent thermal stress, u equivalent represents the equivalent thermal strain, (x, y) represents the position coordinates of a point in the layout area, T represents the temperature field in the layout area of ​​the layout board, φ i represents the strength of the i-th component, T0 represents the temperature value on the layout area boundary, k represents the thermal conductivity in the layout area, n represents the normal vector of the layout area boundary, h represents the surface heat transfer coefficient between the object on the layout area boundary and the surrounding fluid, u x and u y denote the thermal displacement in the x-direction and y-direction, (σ x ,σ y ,σ xy ) represents the thermal stress in three directions, E represents Young's modulus, α represents the linear expansion coefficient, and μ represents Poisson's ratio; Also includes: Build a layout optimization design problem for a layout board that integrates deep learning models; Solve the layout optimization design problem of the layout board and obtain a layout solution of the layout board that can minimize the maximum equivalent thermal stress and the maximum equivalent thermal strain in the layout area of ​​the layout board; The layout optimization design problem of the layout board integrating the deep learning model is: Where X represents the component layout scheme within the layout area of ​​the layout board, f1(X) and f2(X) represent the objective functions, σ equivalent represents the equivalent thermal stress, u equivalent represents the equivalent thermal strain, T represents the temperature field of the layout area of ​​the layout board, Indicates the corresponding relationship between the component layout scheme and the temperature field, Represents the predicted thermal stress and predicted thermal deformation corresponding to the temperature field T output by the deep learning model, It represents the corresponding relationship between thermal stress and thermal deformation and equivalent thermal stress and equivalent thermal strain.

2. The method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology according to claim 1, characterized in that: The training dataset is obtained in the following way: The layout area of ​​the layout board is meshed, and a corresponding number of grids are selected based on the number of components to place the components to obtain a layout scheme. The temperature field of the layout board under the current layout scheme is calculated using the finite element analysis method. Based on the obtained temperature field of the layout board and the constraint equations in the mathematical model, the thermal stress and thermal deformation corresponding to the temperature field of the current layout board are calculated using the finite element analysis method, thereby obtaining training data including the temperature field of the layout board and its corresponding thermal stress and thermal deformation; The training data acquisition process is repeated until a preset amount of training data is obtained, thereby obtaining a training data set including the preset amount of training data.

3. The method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology according to claim 1 or 2, characterized in that: Use the training dataset to train the deep learning model, including: The temperature field of the layout board in the training data is used as input, and the thermal stress and thermal deformation in the training data are used as output to train the deep learning model.

4. The method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology according to claim 3, characterized in that: The temperature field of the layout board in the training data is used as input, and the thermal stress and thermal deformation in the training data are used as output to train the deep learning model, including: Step S401: inputting the temperature fields of the layout boards in the plurality of training data into the deep learning model in sequence to obtain the predicted thermal stress and predicted thermal deformation output by the deep learning model; Step S402: comparing the predicted thermal stress and predicted thermal deformation output by the deep learning model with the thermal stress and thermal deformation in the training data, and calculating the prediction accuracy of the deep learning model; Step S403, determine whether the prediction accuracy obtained for at least two consecutive times is greater than the preset accuracy threshold. If so, use the current deep learning model as the deep learning model to complete the training. If not, adjust the parameters of the deep learning model and return to step S401.

5. The method for predicting thermal stress and thermal deformation of a layout board based on deep learning technology according to claim 1, wherein: Also includes: Before using the training data set to train the deep learning model, the temperature field in the training data is normalized, and the thermal stress and thermal deformation in the training data are normalized.

Images