Electronic device heat dissipation structure level set topological optimization method and system based on physical information neural network

Abstract

The invention discloses an electronic device heat dissipation structure level set topological optimization method and system based on a physical information neural network, and the method comprises the steps: describing a heat exchange behavior between a radiator and a heat source through a heat conduction model, and determining a corresponding boundary condition through a specific heat conduction physical phenomenon; a topological optimization numerical model suitable for a heat dissipation structure is provided according to a level set method, a heat transfer control equation based on an interpolation material is obtained through methods such as interpolation of thermophysical properties of the material through a characteristic function of a level set, and target functions such as the minimum average temperature or the minimum average temperature variance are set, so that the heat transfer control equation of the heat dissipation structure is obtained. And deriving an adjoint equation and sensitivity of the heat transfer equation by adopting a continuous adjoint method, solving the heat transfer equation, the adjoint equation and the reaction diffusion level set equation through a physical information neural network, and further updating a level set function. By adopting the heat dissipation structure optimization method provided by the invention, the design freedom degree is higher, and the temperature of the heat dissipation area can be effectively reduced.

Description

Technical Field

[0001] The invention relates to a level set topology optimization method for a heat dissipation structure of an electronic device based on a physical information neural network, and belongs to the technical field related to topology optimization design of heat dissipation devices. Background Art

[0002] As electronic devices develop towards integration, miniaturization and high power, the performance of electronic devices continues to improve, and their operating frequency and processing power have been significantly improved. However, the improvement in the working performance of electronic devices leads to an increase in internal power consumption and a large amount of heat is generated in a short period of time. When the heat generation power of electronic devices is too large, the temperature of the device will rise rapidly, which may cause internal components to overheat, performance degradation, and even permanent irreversible damage. Therefore, how to perform thermal management on electronic devices has become a key issue in improving the reliability and efficiency of electronic equipment.

[0003] In order to solve the thermal management problem of electronic devices, high-performance heat sinks are needed to effectively dissipate heat from electronic devices. Heat sinks can quickly conduct and diffuse heat, so they can largely ensure that electronic devices are in a reasonable operating temperature range. By arranging heat sinks on the surface of electronic devices, an efficient heat conduction path can be formed to achieve the purpose of rapid heat transfer in a short time. Therefore, the reasonable distribution of heat dissipation structure can greatly improve the heat dissipation efficiency, extend the service life of the device, and ensure the stable operation of electronic products under high power density and miniaturization trends.

[0004] The layout design of traditional heat dissipation devices is mostly based on the experience gained after multiple tests. It has the disadvantages of strong human subjectivity and long design cycle, and it is impossible to obtain the most optimized thermal path design solution. Topology optimization, with the advantages of high-speed computer calculations and corresponding numerical optimization algorithms, can greatly improve the freedom of structural design and provide generative ideas for the architecture of thermal paths. Therefore, it has been initially applied to the structural optimization design of heat dissipation devices. At present, the update and iteration speed of electronic devices is getting faster and faster, which means that the design efficiency of heat dissipation structures needs to be improved simultaneously. Traditional topology optimization design requires a lot of calculations, so it is necessary to optimize the topology optimization design algorithm through neural networks. Summary of the invention

[0005] The technical problem to be solved by the present invention is: how to design a practical topological structure in a short time and with less computing resources, which can significantly reduce the average temperature of the heat dissipation area of ​​the electronic device and has a better heat dissipation effect.

[0006] In order to solve the above technical problems, the present invention provides a level set topology optimization method for the heat dissipation structure of electronic devices based on physical information neural network, comprising the following steps:

[0007] Step 1: Select the topology optimization geometric design domain and non-design domain, determine the location and size of the heat source, the thermal physical parameters of the high thermal conductivity material, and establish a two-dimensional heat dissipation structure model;

[0008] Step 2: Determine the control equation and corresponding boundary conditions based on the physical phenomenon of heat conduction, and select appropriate optimization objective functions and constraints based on the performance improvement requirements of the radiator;

[0009] Step 3: Construct the level set function and characteristic function, use the material interpolation model to interpolate the thermal properties of the design domain material according to the design variables, and establish a mathematical model for heat transfer topology optimization;

[0010] Step 4: According to the mathematical model of heat transfer topology optimization, the corresponding adjoint equation and the sensitivity of the objective function are obtained, and the level set function is solved using the reaction-diffusion level set equation;

[0011] Step 5: Construct a deep neural network model, construct corresponding loss functions according to the partial differential equation of heat conduction, the adjoint partial differential equation of heat conduction and the reaction-diffusion level set equation, respectively, use a machine learning optimization algorithm to iteratively train the neural network model, and obtain the corresponding temperature field, adjoint temperature field and updated level set function distribution through iterative training;

[0012] Step 6: When the topology optimization design reaches the convergence condition, determine whether the temperature index of the heat dissipation structure meets the set requirements. If so, the heat dissipation structure design is completed. If not, modify the topology optimization parameters and repeat steps 1 to 5.

[0013] In the aforementioned level set topology optimization method for the heat dissipation structure of an electronic device based on a physical information neural network, in step 2, the control equation is a steady-state or transient heat conduction equation, and the boundary conditions are a given temperature, a given heat flux density, or a given convection heat transfer coefficient and a fluid temperature; the optimization objective function includes minimizing the average temperature, minimizing the maximum temperature, and minimizing the average temperature variance, and the constraints include volume constraints.

[0014] In the aforementioned level set topology optimization method for the heat dissipation structure of electronic devices based on physical information neural network, in step 2, the steady-state heat conduction equation is selected for numerical simulation, which is expressed as:

[0015]

[0016] In the formula, represents the design area, is the temperature, is the interpolated thermal conductivity, Q is the heat generation power of the heat generating device, and the boundary condition is a given temperature;

[0017] The optimization objective function is to minimize the average temperature of the heat dissipation area, and the constraint function is selected as the material volume constraint, which can be expressed as:

[0018] Min

[0019]

[0020] In the formula, To optimize the objective function, is the level set characteristic function, is a given volume constraint.

[0021] In the aforementioned level set topology optimization method for the heat dissipation structure of electronic devices based on physical information neural network, in step 3, the level set can be constructed by a step function, a sign function or a radial basis function; the interpolation model includes a linear model, a SIMP model, a RAMP model or a Darcy model.

[0022] In the aforementioned level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network, in step 3, the SIMP model is used to interpolate the thermal physical property parameters of high thermal conductivity materials:

[0023]

[0024]

[0025]

[0026] In the formula, , , are the thermal conductivity of high thermal conductivity materials, , , are the thermal conductivity, density and specific heat capacity of the area without thermal conductive material, , and They are penalty factor one, penalty factor two, and penalty factor three. , , is the interpolated thermal conductivity, density and specific heat capacity, is the characteristic function of the level set function. The relationship between the characteristic function and the level set function is expressed as:

[0027]

[0028] In the formula, is the level set function, is the mapping transformation width of the characteristic function;

[0029] The mathematical model of heat transfer topology optimization is expressed as:

[0030] Min

[0031] st

[0032]

[0033]

[0034] In the aforementioned level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network, in step 4, the adjoint equation and its boundary conditions are expressed as:

[0035]

[0036] (given boundary temperature)

[0037] (given boundary heat flux density)

[0038] In the formula, is the accompanying temperature, is the gradient operator, is the gradient with temperature, represents the objective function in the design domain;

[0039] The sensitivity of the topology optimization It is expressed as:

[0040] ;

[0041] In the formula, is the objective function of topology optimization design, is the accompanying temperature, is the temperature gradient;

[0042] The reaction-diffusion level set equation is expressed as:

[0043]

[0044] In the formula, is pseudo time, is the relaxation factor, is the diffusion coefficient.

[0045] In the aforementioned level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network, in step 5, the deep neural network model includes an input layer, multiple hidden layers and an output layer, and the input layer is the coordinates of the grid nodes;

[0046] When solving the reaction-diffusion equation, the input layer also includes time, discretizing the two-dimensional design area into 102*102=10404 grid nodes;

[0047] When solving the heat transfer equation, the adjoint heat transfer equation and the reaction-diffusion level set equation, the output layer corresponds to the temperature value, the adjoint temperature value and the updated level set function value corresponding to each grid node;

[0048] The hidden layer includes a fully connected layer and its corresponding activation function, and the activation function includes one or more of a Sigmoid function, a ReLU function, and a Tanh function.

[0049] In the aforementioned level set topology optimization method for the heat dissipation structure of electronic devices based on physical information neural network, in step 5, the mean square error loss is selected to construct the loss function. For the physical information neural network, the loss function is expressed as:

[0050]

[0051]

[0052]

[0053] In the formula, is the loss function, and are partial differential equation loss and boundary loss respectively, and To adjust the weight coefficients of partial differential equation loss and boundary loss; and are the number of computational domain grids and the number of boundary grids, respectively. Indicates grid cells, Represents the boundary The temperature of a grid cell.

[0054] A computer device / equipment / system, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-mentioned vehicle network multi-parameter monitoring method based on 5G multi-access edge computing.

[0055] A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that when the computer program / instruction is executed by a processor, the steps of the above-mentioned vehicle network multi-parameter monitoring method based on 5G multi-access edge computing are implemented.

[0056] Beneficial effects achieved by the present invention: The present invention determines the partial differential equation of heat conduction and the corresponding boundary conditions through actual engineering application scenarios. The level set function and its characteristic function are constructed to interpolate the thermal properties such as thermal conductivity and specific heat capacity of the material, so as to minimize the average temperature of the heat dissipation area, the maximum temperature and one or more of the temperature equations to construct the topology optimization objective function, impose design constraints according to actual conditions, and finally establish a numerical model for topology optimization of the heat dissipation structure. The adjoint equation, adjoint boundary conditions and sensitivity of the topology optimization problem are derived by the Lagrange multiplier method, and a physical information neural network model is constructed to solve the heat transfer equation, the adjoint heat transfer equation and the reaction diffusion level set equation, so as to finally realize the evolution of the level set function in the direction of reducing the objective function and obtain the optimal heat dissipation structure. Compared with the traditional heat sink, the heat sink structure of the topology optimization design of the present invention adopts the numerical optimization algorithm of the computer, has higher design efficiency and design freedom, so as to significantly improve the heat dissipation effect and temperature uniformity performance of the heat dissipation structure. The topology optimization result is related to the position and size of the heat source and the thermal properties of the heat-conducting material. Therefore, it is possible to optimize the design of heat-conducting materials with different thermal conductivities according to the working conditions of the actual heat-generating device, thereby effectively reducing the working temperature of the heat dissipation device. Compared with the existing heat dissipation structure topology optimization method, the present invention adopts the proposed heat dissipation structure level set topology optimization design method based on physical information neural network, adopts the level set function to describe the topological configuration of the heat dissipation structure, and combines the neural network to accelerate the iteration of the topology optimization process, so that a smooth and manufacturable heat dissipation structure can be obtained more quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of a method for level set topology optimization of a heat dissipation structure of an electronic device based on a physical information neural network provided in Example 1 of the present invention;

[0058] Figure 2 It is a schematic diagram of topological optimization design of a two-dimensional heat dissipation model in Example 1 of the present invention;

[0059] Figure 3 is a schematic diagram of a physical information neural network in Example 1 of the present invention;

[0060] Figure 4 Schematic diagram of the initial level set hole distribution and optimized thermal conductive material distribution in Example 1 of the present invention. DETAILED DESCRIPTION

[0061] The following describes several preferred embodiments of the present invention with reference to the drawings in the specification, so that the technical content is clearer and easier to understand. The present invention can be embodied through different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0062] Example 1

[0063] like Figure 1 As shown, this embodiment provides a level set topology optimization method for the heat dissipation structure of an electronic device based on a physical information neural network, comprising the following steps:

[0064] Step 1: Select the topology optimization geometric design domain and non-design domain, determine the location and size of the heat source, the thermal physical parameters of the high thermal conductivity material, and establish a two-dimensional heat dissipation structure model;

[0065] The design domain refers to the specific dimensions such as length and width and relative position of the two-dimensional heat dissipation structure design area determined according to the actual geometric dimensions of the heat generating device.

[0066] like Figure 2 As shown, in this embodiment, the two-dimensional heat dissipation structure model established has a length and a width of 1m. The thermal physical parameters of the high thermal conductivity material include the thermal conductivity, density and specific heat of the material. The density, thermal conductivity and specific heat capacity are 1 kg / m 3 , 1 W / (m·K), 1 (J / kg·K), the density, thermal conductivity and specific heat capacity of the area without thermal conductive material should be close to 0 in theory. Considering the numerical stability, the values ​​can be taken as 0.001 kg / m 3 , 0.001 W / (m·K), 0.001 (J / kg·K). In other embodiments, the physical property parameters may adopt other values.

[0067] Step 2: Determine the control equation and corresponding boundary conditions based on the physical phenomenon of heat conduction, and select appropriate optimization objective functions and constraints based on the performance improvement requirements of the radiator;

[0068] The control equation is a steady-state or transient heat conduction equation, and the boundary conditions are a given temperature, a given heat flux density, or a given convection heat transfer coefficient and a fluid temperature; the optimization objective function includes one or more of minimizing the average temperature, minimizing the maximum temperature, and minimizing the average temperature variance; the constraints include volume constraints, etc.

[0069] In this embodiment, the steady-state heat conduction equation is used for numerical simulation, which is expressed as:

[0070]

[0071] In the formula, represents the design area, is the temperature, is the thermal conductivity after interpolation, The heat source is the heat generating device, and the heat generating power Q is 1W / m 2 , the boundary condition is given by temperature T0 = 0℃.

[0072] In this embodiment, the optimization objective function is to minimize the average temperature of the heat dissipation area, and the constraint function is selected as the material volume constraint, which can be expressed as:

[0073] Min

[0074]

[0075] In the formula, To optimize the objective function, is the level set characteristic function, is a given volume constraint.

[0076] Step 3: Construct the level set function and characteristic function, use the material interpolation model to interpolate the thermal properties of the design domain material according to the design variables, and establish a mathematical model for heat transfer topology optimization;

[0077] The level set can be constructed by a step function, a sign function, a radial basis function, etc.

[0078] In this embodiment, the SIMP model is used to interpolate the thermal physical parameters of the high thermal conductivity material:

[0079]

[0080]

[0081]

[0082] In the formula, , , are the thermal conductivity of high thermal conductivity materials, , , are the thermal conductivity, density and specific heat capacity of the area without thermal conductive material, , and They are penalty factor one, penalty factor two, and penalty factor three. , , is the interpolated thermal conductivity, density and specific heat capacity, is the characteristic function of the level set function. The relationship between the characteristic function and the level set function can be expressed as:

[0083]

[0084] In the formula, is the level set function, is the mapping transformation width of the characteristic function.

[0085] In this embodiment, the heat transfer topology optimization mathematical model can be expressed as:

[0086] Min

[0087] st

[0088]

[0089]

[0090] Step 4: According to the mathematical model of heat transfer topology optimization, the corresponding adjoint equation and the sensitivity of the objective function are obtained, and the reaction-diffusion level set equation is used to solve the level set function.

[0091] The adjoint equation and its boundary conditions can be expressed as:

[0092]

[0093] (given boundary temperature)

[0094] (given boundary heat flux density)

[0095] In the formula, is the accompanying temperature, is the gradient operator, is the gradient with temperature, represents the objective function in the design domain.

[0096] The sensitivity of the topology optimization It is expressed as:

[0097]

[0098] In the formula, is the objective function of topology optimization design, is the accompanying temperature, is the temperature gradient. The reaction-diffusion level set equation is expressed as:

[0099]

[0100] In the formula, is pseudo time, is the relaxation factor, is the diffusion coefficient.

[0101] Step 5: Construct a deep neural network model, construct corresponding loss functions according to the partial differential equation of heat conduction, the adjoint partial differential equation of heat conduction and the reaction-diffusion level set equation, respectively, use a machine learning optimization algorithm to iteratively train the neural network model, and obtain the corresponding temperature field, adjoint temperature field and updated level set function distribution through iterative training;

[0102] The deep neural network model includes an input layer, multiple hidden layers and an output layer. The input layer is the coordinates of the grid nodes. When solving the reaction-diffusion equation, the input layer also includes time, and the two-dimensional design area is discretized into 102*102=10404 grid nodes;

[0103] When solving the heat transfer equation, the concomitant heat transfer equation and the reaction-diffusion level set equation, the output layer corresponds to the temperature value, the concomitant temperature value and the updated level set function value corresponding to each grid node.

[0104] The hidden layer includes a fully connected layer and its corresponding activation function, and the activation function includes one or more of a Sigmoid function, a ReLU function, and a Tanh function.

[0105] In this embodiment, the physical information neural network is set to have 6 hidden layers, each hidden layer has 64 neurons, and the activation function is Tanh function. The neural network optimization algorithm includes Adam, AdamW, SGD, etc.

[0106] In this embodiment, the AdamW optimization algorithm is selected to create a neural network optimizer, and the loss function for solving the partial differential equation includes but is not limited to mean square error loss (MSE Loss), cross entropy loss (Cross Entropy Loss), L1 or L2 regularization, etc.

[0107] In this embodiment, the mean square error loss is selected to construct the loss function. For the physical information neural network, the loss function consists of two parts, namely, the boundary loss and the partial differential equation loss. Taking the heat conduction equation as an example, the loss function can be expressed as:

[0108]

[0109]

[0110]

[0111] In the formula, is the loss function, and are partial differential equation loss and boundary loss respectively, and To adjust the weight coefficients of partial differential equation loss and boundary loss; and are the number of computational domain grids and the number of boundary grids, respectively. Indicates grid cells, Represents the boundary The temperature of a grid cell.

[0112] The process of iterative training of the neural network includes forward propagation, loss calculation, back propagation, gradient descent weight update, etc. When the iteration converges, the solved temperature field, accompanying temperature field and updated level set function are output.

[0113] Step 6: When the topology optimization design reaches the convergence condition, determine whether the temperature index of the heat dissipation structure meets the set requirements. If it does, the heat dissipation structure design is completed. If it does not, modify the topology optimization parameters and repeat steps (1) to (5). Figure 4 Shown is the initial level set hole distribution and the thermally conductive material distribution after convergence conditions are reached.

[0114] Example 2

[0115] A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network includes the following steps:

[0116] Step 1: Select the topology optimization geometric design domain and non-design domain, determine the location and size of the heat source, the thermal physical parameters of the high thermal conductivity material, and establish a two-dimensional heat dissipation structure model;

[0117] Step 2: Determine the control equation and corresponding boundary conditions based on the physical phenomenon of heat conduction, and select appropriate optimization objective functions and constraints based on the performance improvement requirements of the radiator;

[0118] Step 3: Construct the level set function and characteristic function, use the material interpolation model to interpolate the thermal properties of the design domain material according to the design variables, and establish a mathematical model for heat transfer topology optimization;

[0119] Step 4: According to the mathematical model of heat transfer topology optimization, the corresponding adjoint equation and the sensitivity of the objective function are obtained, and the level set function is solved using the reaction-diffusion level set equation;

[0120] Step 5: Construct a deep neural network model, construct corresponding loss functions according to the partial differential equation of heat conduction, the adjoint partial differential equation of heat conduction and the reaction-diffusion level set equation, respectively, use a machine learning optimization algorithm to iteratively train the neural network model, and obtain the corresponding temperature field, adjoint temperature field and updated level set function distribution through iterative training;

[0121] Step 6: When the topology optimization design reaches the convergence condition, determine whether the temperature index of the heat dissipation structure meets the set requirements. If so, the heat dissipation structure design is completed. If not, modify the topology optimization parameters and repeat steps 1 to 5.

[0122] In step 2, the control equation is a transient heat conduction equation.

[0123] Other technical features are the same as those of Example 1.

[0124] Example 3

[0125] A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network includes the following steps:

[0126] Step 1: Select the topology optimization geometric design domain and non-design domain, determine the location and size of the heat source, the thermal physical parameters of the high thermal conductivity material, and establish a two-dimensional heat dissipation structure model;

[0127] Step 2: Determine the control equation and corresponding boundary conditions based on the physical phenomenon of heat conduction, and select appropriate optimization objective functions and constraints based on the performance improvement requirements of the radiator;

[0128] Step 3: Construct the level set function and characteristic function, use the material interpolation model to interpolate the thermal properties of the design domain material according to the design variables, and establish a mathematical model for heat transfer topology optimization;

[0129] Step 4: According to the mathematical model of heat transfer topology optimization, the corresponding adjoint equation and the sensitivity of the objective function are obtained, and the level set function is solved using the reaction-diffusion level set equation;

[0130] Step 5: Construct a deep neural network model, construct corresponding loss functions according to the partial differential equation of heat conduction, the adjoint partial differential equation of heat conduction and the reaction-diffusion level set equation, respectively, use a machine learning optimization algorithm to iteratively train the neural network model, and obtain the corresponding temperature field, adjoint temperature field and updated level set function distribution through iterative training;

[0131] Step 6: When the topology optimization design reaches the convergence condition, determine whether the temperature index of the heat dissipation structure meets the set requirements. If so, the heat dissipation structure design is completed. If not, modify the topology optimization parameters and repeat steps 1 to 5.

[0132] In step 3, the interpolation model may be a linear model, a RAMP model or a Darcy model.

[0133] A computer device / equipment / system, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-mentioned vehicle network multi-parameter monitoring method based on 5G multi-access edge computing.

[0134] A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that when the computer program / instruction is executed by a processor, the steps of the above-mentioned vehicle network multi-parameter monitoring method based on 5G multi-access edge computing are implemented.

[0135] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0136] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0137] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0138] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network, characterized in that: The following steps are involved: Step 1: Select the topology optimization geometric design domain and non-design domain, determine the location and size of the heat source, the thermal properties of the high thermal conductivity material, and establish a two-dimensional heat dissipation structure model; Step 2: Determine the control equation and corresponding boundary conditions based on the physical phenomenon of heat conduction, and select the optimization objective function and constraint conditions based on the performance improvement requirements of the radiator; Step 3: Construct the level set function and characteristic function, use the material interpolation model to interpolate the thermal properties of the design domain material according to the design variables, and establish a mathematical model for heat transfer topology optimization; Step 4: According to the mathematical model of heat transfer topology optimization, the corresponding adjoint equation and the sensitivity of the objective function are obtained, and the level set function is solved using the reaction-diffusion level set equation; Step 5: Construct a deep neural network model, construct corresponding loss functions according to the partial differential equation of heat conduction, the adjoint partial differential equation of heat conduction and the reaction-diffusion level set equation, respectively, use a machine learning optimization algorithm to iteratively train the neural network model, and obtain the corresponding temperature field, adjoint temperature field and updated level set function distribution through iterative training; Step 6: When the topology optimization design reaches the convergence condition, determine whether the temperature index of the heat dissipation structure meets the set requirements. If so, the heat dissipation structure design is completed. If not, modify the topology optimization parameters and repeat steps 1 to 5.

2. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 1, characterized in that: In step 2, the control equation is a steady-state or transient heat conduction equation, and the boundary conditions are a given temperature, a given heat flux density, or a given convection heat transfer coefficient and a fluid temperature; the optimization objective function includes one or more of minimizing the average temperature, minimizing the maximum temperature, and minimizing the average temperature variance, and the constraints include volume constraints.

3. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 2, characterized in that: In step 2, the steady-state heat conduction equation is selected for numerical simulation, which is expressed as: ； In the formula, represents the design area, is the temperature, is the interpolated thermal conductivity, Q is the heat generation power of the heat generating device, and the boundary condition is a given temperature; The optimization objective function is to minimize the average temperature of the heat dissipation area, and the constraint function is the material volume constraint, which is expressed as: Min ; ； In the formula, To optimize the objective function, is the level set characteristic function, is a given volume constraint.

4. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 1, characterized in that: In step 3, the level set is constructed by a step function, a sign function or a radial basis function; the interpolation model includes a linear model, a SIMP model, a RAMP model or a Darcy model.

5. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 4, characterized in that: In step 3, the SIMP model is used to interpolate the thermophysical parameters of high thermal conductivity materials: ； ； ； In the formula, , , are the thermal conductivity of high thermal conductivity materials, , , are the thermal conductivity, density and specific heat capacity of the area without thermal conductive material, , and They are penalty factor one, penalty factor two, and penalty factor three. , , are the interpolated thermal conductivity, density and specific heat, is the characteristic function of the level set function. The relationship between the characteristic function and the level set function is expressed as: ； In the formula, is the level set function, is the mapping transformation width of the characteristic function; The mathematical model of heat transfer topology optimization is expressed as: Min ; st; ； 。 6. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 1, characterized in that: In step 4, the adjoint equation and its boundary conditions are expressed as: ； (given boundary temperature); (given boundary heat flux density); In the formula, is the accompanying temperature, is the gradient operator, is the gradient with temperature, represents the objective function in the design domain; The sensitivity of the topology optimization It is expressed as: ； In the formula, The objective function for topology optimization design is is the temperature gradient; The reaction-diffusion level set equation is expressed as: ； In the formula, is pseudo time, is the relaxation factor, is the diffusion coefficient.

7. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 1, characterized in that: In step 5, the deep neural network model includes an input layer, multiple hidden layers and an output layer, and the input layer is the coordinates of the grid nodes; When solving the reaction-diffusion equation, the input layer also includes time, discretizing the two-dimensional design area into 102*102=10404 grid nodes; When solving the heat transfer equation, the adjoint heat transfer equation and the reaction-diffusion level set equation, the output layer corresponds to the temperature value, the adjoint temperature value and the updated level set function value corresponding to each grid node; The hidden layer includes a fully connected layer and its corresponding activation function, and the activation function includes one or more of a Sigmoid function, a ReLU function, and a Tanh function.

8. A level set topology optimization method for heat dissipation structure of electronic devices based on physical information neural network according to claim 1, characterized in that: In step 5, the mean square error loss is used to construct the loss function. For the physical information neural network, the loss function is expressed as: ； ； ； In the formula, is the loss function, and are partial differential equation loss and boundary loss respectively, and To adjust the weight coefficients of partial differential equation loss and boundary loss; and are the number of computational domain grids and the number of boundary grids, respectively. Indicates grid cells, Represents the boundary The temperature of a grid cell.

9. A computer device / equipment / system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the vehicle network multi-parameter monitoring method based on 5G multi-access edge computing as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by the processor, the steps of the vehicle network multi-parameter monitoring method based on 5G multi-access edge computing as described in any one of claims 1-8 are implemented.

Images