A thermal simulation method, system and circuit board based on 5G circuit board

Abstract

The present invention relates to the field of 5G circuit board thermal simulation technology, and in particular to a thermal simulation method, system and circuit board based on 5G circuit boards. The method comprises the following steps: measuring and calculating the material properties of the 5G circuit board, and constructing a material database to obtain an enhanced material property matrix; constructing a simplified model of the 5G circuit board, and defining boundary conditions and heat sources to obtain a simplified CAD model and boundary conditions and heat source definition data; solving the heat transfer equation of the simplified CAD model according to the boundary conditions and heat source definition data, and performing temperature field calculation to obtain temperature field distribution data; performing phase change latent heat influence analysis on the simplified CAD model according to the temperature field distribution data to obtain phase change latent heat data. The present invention achieves higher simulation accuracy, faster simulation speed and stronger design automation capability through 5G circuit board thermal simulation technology.

Description

Technical Field

[0001] The present invention relates to the field of 5G circuit board thermal simulation technology, and in particular to a thermal simulation method, system and circuit board based on a 5G circuit board. Background Art

[0002] The rapid development of 5G technology has placed higher demands on the heat dissipation of electronic devices. 5G circuit boards operate at higher frequencies, have higher integration densities, and consume greater power, leading to significantly increased heat generation. When exposed to high temperatures, components on these boards experience performance degradation or even failure. Heat dissipation has become a key factor limiting the performance and reliability of 5G equipment.

[0003] Existing PCB thermal simulation technologies primarily use simplified models and empirical formulas for estimation. For example, they treat components as simple heat sources and ignore the anisotropy of the PCB's internal structure and materials. This approach offers fast calculation speed but low accuracy, failing to meet the requirements of 5G PCBs. Complex PCB structures require a fine mesh, resulting in a massive computational load, long simulation times, and low efficiency. Summary of the Invention

[0004] Based on this, it is necessary to provide a thermal simulation method, system and circuit board based on a 5G circuit board to solve at least one of the above technical problems.

[0005] To achieve the above objectives, a thermal simulation method based on a 5G circuit board includes the following steps:

[0006] Step S1: Measure and calculate the material properties of the 5G circuit board to obtain thermal conductivity characteristic data and interface thermal resistance data; construct a material database based on the thermal conductivity characteristic data and interface thermal resistance data to obtain an enhanced material property matrix;

[0007] Step S2: construct a simplified model of the 5G circuit board, and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; solve the heat transfer equation of the simplified CAD model according to the boundary conditions and heat source definition data, and perform temperature field calculation to obtain temperature field distribution data; perform phase change latent heat influence analysis on the simplified CAD model according to the temperature field distribution data to obtain phase change latent heat data; perform entropy increase and thermal impedance analysis based on transient heat sources according to the phase change latent heat data to obtain path entropy value data and path thermal impedance data; perform key path extraction according to the path entropy value data and path thermal impedance data, and perform hotspot identification to obtain a key heat dissipation path diagram;

[0008] Step S3: performing initial mesh division according to the key heat dissipation path diagram, and performing adaptive mesh optimization to obtain an adaptive mesh structure; performing mesh iterative optimization based on the adaptive mesh structure to obtain an optimized mesh structure;

[0009] Step S4: Using the optimized grid structure to extract thermal impedance data under different working conditions to obtain a thermal impedance data set; constructing a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; using the thermal impedance prediction model to perform rapid temperature field calculation to obtain a rapid temperature field distribution; performing heat dissipation performance analysis on the rapid temperature field distribution, and performing design iteration and optimization to obtain an optimized design scheme.

[0010] By combining macroscopic measurements and microscopic simulations, this paper establishes a comprehensive material database (EMPM) that includes thermal conductivity characteristics and interface thermal resistance data for commonly used materials in 5G circuit boards. This not only improves the accuracy of material parameters but also provides a reliable data foundation for subsequent simulation analysis, thereby enhancing the accuracy and reliability of the overall simulation. By constructing a simplified model, defining boundary conditions and heat sources, and combining phase change latent heat and transient heat source analysis, a more realistic simulation of the thermal behavior of 5G circuit boards is achieved. In particular, entropy increase and thermal impedance analysis can more accurately identify heat dissipation bottlenecks and critical paths, providing precise guidance for subsequent optimization design. An adaptive mesh optimization strategy based on critical heat dissipation paths significantly improves simulation efficiency while ensuring simulation accuracy. By increasing the mesh density in critical areas and increasing the mesh density in non-critical areas, the number of meshes can be effectively controlled and computing resources can be concentrated on the areas with the greatest impact on the temperature field, thereby obtaining more accurate results in a shorter time. Using a neural network model to predict thermal impedance avoids repeated, time-consuming finite element simulations, greatly accelerating design iterations. The prediction model built through training data can quickly evaluate the heat dissipation performance of different design schemes, thereby efficiently finding the optimal design scheme, shortening the design cycle, and reducing R&D costs. Therefore, the present invention provides a thermal simulation method based on 5G circuit boards, which effectively solves the accuracy and efficiency problems faced by existing methods in 5G circuit board thermal design. Through more accurate material property description, more comprehensive heat dissipation path analysis, more intelligent meshing and faster thermal impedance prediction, higher simulation accuracy, faster simulation speed and stronger design automation capabilities are achieved.

[0011] Preferably, step S1 includes the following steps:

[0012] Step S11: measuring the material properties of the 5G circuit board to obtain a material property table;

[0013] Step S12: performing energy-minimized material NPT ensemble simulation according to the material property table to obtain equilibrium atomic trajectory data;

[0014] Step S13: Calculating heat conduction characteristics based on the equilibrium atomic trajectory data to obtain heat conduction characteristic data;

[0015] Step S14: Calculating the interface thermal resistance based on the equilibrium atomic trajectory data to obtain interface thermal resistance data;

[0016] Step S15: performing nanoscale material property modeling and calibration based on the heat conduction property data and the interface thermal resistance data to obtain a nanoscale material property table;

[0017] Step S16: performing data fusion and calibration according to the material property table and the nanoscale material property table to obtain an enhanced material property matrix.

[0018] This paper obtains real-world material thermal property data, such as thermal conductivity, specific heat capacity, and density, by conducting actual measurements of commonly used materials in 5G circuit boards. This avoids errors caused by using literature or empirical values, provides reliable basic data for subsequent simulation analysis, and ensures the accuracy of simulation results. Using NPT ensemble simulation to minimize the material's energy, atomic trajectory data at equilibrium is obtained. This provides the necessary data foundation for subsequent calculations of nanoscale thermal conductivity and interfacial thermal resistance, enabling a deeper understanding of the influence of a material's microstructure on its thermal properties. By analyzing the equilibrium atomic trajectory data, thermal conductivity data for the material is calculated. This overcomes the shortcomings of macroscopic measurement methods at the nanoscale, enabling a more accurate description of the thermal conduction behavior of materials at the microscale and improving simulation accuracy. Using the equilibrium atomic trajectory data, thermal resistance data at the material interface is calculated. This provides key parameters for accurately simulating heat transfer between different materials, more realistically reflecting the thermal conduction conditions at the interfaces of different materials in 5G circuit boards, and further improving simulation accuracy. By modeling and calibrating nanoscale thermal conductivity and interface thermal resistance data, a more accurate nanoscale material property model was established. This enables a more accurate description of the thermal behavior of materials at the nanoscale and provides more reliable material parameters for macroscale simulations. By fusing and calibrating macroscale measurement data with nanoscale simulation data, an enhanced material property matrix (EMPM) was constructed. This combines material property information at two different scales, providing more comprehensive and accurate material parameters, laying a solid foundation for subsequent thermal simulation analysis and effectively improving overall simulation accuracy.

[0019] Preferably, step S2 includes the following steps:

[0020] Step S21: Obtain a 5G circuit board CAD model and perform model simplification to obtain a simplified CAD model;

[0021] Step S22: defining boundary conditions and heat sources according to the enhanced material property matrix and the simplified CAD model to obtain boundary condition and heat source definition data;

[0022] Step S23: Solving the heat transfer equation for the simplified CAD model according to the boundary conditions and the heat source definition data, and performing temperature field calculation to obtain temperature field distribution data;

[0023] Step S24: performing a phase change latent heat influence analysis on the simplified CAD model based on the temperature field distribution data to obtain phase change latent heat data; performing an entropy increase and thermal impedance analysis based on a transient heat source based on the phase change latent heat data to obtain path entropy increase data and path thermal impedance data;

[0024] Step S25: extracting key paths based on the path entropy value-added data and the path thermal impedance data, and identifying hot spots to obtain a key heat dissipation path diagram.

[0025] This method simplifies the CAD model and removes details that have little impact on the thermal simulation results. While maintaining simulation accuracy, it significantly reduces model complexity and computational complexity, improves simulation efficiency, and makes subsequent meshing and simulation calculations faster and more stable. Accurately defining boundary conditions and heat sources is crucial for thermal simulation. This step defines parameters such as the air velocity and temperature for forced convection, as well as the average chip power consumption and fluctuation amplitude, based on the actual operating environment and chip power consumption data. This provides the necessary input conditions for solving the heat transfer equations and ensures the accuracy and reliability of the simulation results. Solving the heat transfer equations yields temperature field distribution data for the 5G circuit board, which forms the basis for thermal analysis. This step provides a preliminary understanding of the overall thermal performance of the 5G circuit board and provides the necessary data support for subsequent phase change latent heat analysis and entropy increase / thermal impedance analysis. This analysis considers the influence of the latent heat effect of the phase change material on the temperature field and, in combination with the characteristics of the transient heat source, conducts an entropy increase and thermal impedance analysis. This makes simulation results more closely aligned with the actual operating conditions of 5G circuit boards, enabling more accurate identification of heat dissipation bottlenecks and critical paths, providing more precise guidance for subsequent design optimization. By analyzing path entropy value data and path thermal impedance data, critical heat dissipation paths are extracted and hotspots on the 5G circuit board are identified. This provides clear targets for subsequent mesh optimization and design optimization, enabling more targeted improvements to improve heat dissipation efficiency. The critical heat dissipation path diagram intuitively illustrates the main heat transfer paths and heat dissipation bottlenecks.

[0026] Preferably, step S24 includes the following steps:

[0027] Step S241: identifying the phase change material according to the enhanced material property matrix to obtain the phase change material; performing path selection and unit division according to the phase change material and the simplified CAD model to obtain path unit information;

[0028] Step S242: determining the phase change state based on the temperature field distribution data and the phase change material to obtain a phase change state mark; calculating the phase change latent heat based on the temperature field distribution data and the phase change material to obtain phase change latent heat data;

[0029] Step S243: performing material thermal response analysis according to the enhanced material property matrix to obtain material thermal response data;

[0030] Step S244: performing heat flux calculation based on the temperature field distribution data, the phase change latent heat data, and the material thermal response data to obtain heat flux data;

[0031] Step S245: Calculating the unit entropy increase rate based on the heat flux density data to obtain unit entropy increase rate data; performing path entropy value integration based on the unit entropy increase rate data and the path unit information to obtain path entropy value data;

[0032] Step S246: Calculating the thermal impedance of the component based on the temperature field distribution data to obtain component thermal impedance data; calculating the thermal impedance of the heat dissipation path based on the heat flux density data and the component thermal impedance data to obtain path thermal impedance data.

[0033] By identifying phase change materials and performing path selection and unit division, this method lays the foundation for subsequent calculation of phase change latent heat and entropy increase / thermal impedance analysis. Acquiring path unit information enables more detailed analysis of heat transfer along critical paths, thereby more accurately identifying heat dissipation bottlenecks. Accurately determining the state of the phase change material and calculating its latent heat value are crucial for simulating real-world heat transfer processes. This step takes into account the heat absorption and release effects of the phase change material at different temperatures, improving the accuracy of simulation results, especially in the presence of temperature fluctuations. By analyzing the material's thermal response characteristics, such as thermal diffusivity, we can better understand the material's response speed to temperature changes. This helps to more accurately simulate the effects of transient heat sources and correct heat flux calculations, improving the reliability of simulation results. By comprehensively considering factors such as heat conduction, phase change latent heat, and material thermal response, more accurate heat flux data is calculated. This provides a more reliable data foundation for subsequent entropy increase rate and thermal impedance calculations, making the analysis results more representative. By calculating the unit entropy increase rate and path entropy increase, we can quantitatively analyze irreversible losses in the heat transfer process, thereby identifying heat dissipation bottlenecks and optimizing heat dissipation paths. Entropy increase analysis provides a new perspective for thermal design, effectively guiding the optimization of cooling solutions. Calculating component thermal impedance and heat dissipation path thermal impedance allows evaluation of the efficiency of different heat dissipation paths and provides a reference for subsequent design optimization. By analyzing thermal impedance data, the path with the highest thermal resistance can be identified, enabling more targeted improvements to reduce thermal resistance and improve heat dissipation efficiency.

[0034] Preferably, step S244 is specifically as follows:

[0035] Calculate the heat flux density of heat conduction according to the temperature field distribution data to obtain the heat flux density data of unit heat conduction;

[0036] Calculate the heat flux density of convective heat transfer according to the temperature field distribution data to obtain the heat flux density data of unit convective heat transfer;

[0037] Calculate the heat flux density of radiation heat transfer according to the temperature field distribution data to obtain the heat flux density data of unit radiation heat transfer;

[0038] Perform heat flux superposition on the unit heat conduction heat flux density data, the unit convection heat transfer heat flux density data, and the unit radiation heat transfer heat flux density data to obtain heat flux superposition data;

[0039] Perform transient heat source impact analysis based on temperature field distribution data to obtain transient heat source impact data;

[0040] The heat flux density distribution inside the unit is carried out according to the transient heat source influence data to obtain the local heat flux density influence data;

[0041] The heat flux direction is corrected on the heat flux superposition data according to the local heat flux influence data to obtain the heat flux data.

[0042] By calculating the heat flux density of each unit cell, this method accurately describes the heat transfer process within solid materials. This is the basis for thermal analysis and provides an important data foundation for subsequent calculation of the total heat flux density and analysis of heat transfer paths. Calculating the heat flux density of convective heat transfer accurately describes the heat exchange process between the 5G circuit board and the surrounding environment, which is crucial for evaluating the effectiveness of heat sinks and optimizing heat dissipation designs. Considering the influence of radiative heat transfer allows for a more comprehensive simulation of the heat transfer process, especially at high temperatures, where the role of radiative heat transfer cannot be ignored. Calculating the heat flux density of radiative heat transfer improves the accuracy and reliability of the simulation results. Superimposing the heat flux densities of the three heat transfer methods produces more comprehensive heat flux data, which can more accurately reflect the complex heat transfer process in the 5G circuit board. Considering the influence of transient heat sources allows for a more realistic simulation of the thermal behavior of the 5G circuit board in actual operating conditions, as chip power consumption often varies over time. This is crucial for evaluating the effectiveness of cooling solutions and predicting the maximum device temperature. By analyzing the impact of transient heat sources on the heat flux density distribution within the unit, the dynamic thermal behavior of the heat source can be more precisely simulated, leading to more accurate predictions of temperature changes in the heat source area. By correcting the direction of the heat flux density, the flow of heat in 5G circuit boards can be more accurately described. This is crucial for identifying critical heat dissipation paths and optimizing thermal design, enabling more efficient heat transfer from the heat source to the heat sink.

[0043] Preferably, the transient heat source impact analysis is performed based on the temperature field distribution data as follows:

[0044] Identify the heat source location based on the temperature field distribution data and the simplified CAD model to obtain heat source location identification data;

[0045] Determine the geometric shape based on the heat source position identification data to obtain the heat source geometric information;

[0046] According to the heat source geometry information, the heat source power is analyzed over time to obtain the heat source power time function;

[0047] Selecting a transient heat source model according to heat source geometry information to obtain a selected transient heat source model;

[0048] The transient heat source heat flux density is calculated based on the selected transient heat source model and the heat source power-time function to obtain the transient heat source impact data.

[0049] Accurately identifying the location of the heat source is a prerequisite for transient heat source impact analysis. This step, by combining temperature field distribution data with a simplified CAD model, precisely locates the heat source's location and geometric information, providing the necessary spatial information for subsequent analysis. Determining the heat source's geometric shape and dimensions allows for more accurate calculation of its volume and surface area, which is crucial for subsequent calculations of heat flux density and thermal analysis, avoiding errors caused by inaccurate geometric information. By analyzing the temporal variation of heat source power, a heat source power-time function is established, enabling a more realistic simulation of the actual thermal behavior of 5G circuit boards. Chip power consumption is typically not constant but varies over time. Selecting an appropriate transient heat source model, such as a volume heat source model, can more accurately describe the heat source's thermal characteristics, simplify the calculation process, improve simulation efficiency, and ensure the reliability of the simulation results. By calculating the transient heat source's heat flux density, a more accurate description of the heat source's heat output at different times can be achieved. This is crucial for simulating the dynamic thermal behavior of 5G circuit boards and predicting the transient temperature response of devices, enabling more effective evaluation of the performance of cooling solutions.

[0050] Preferably, step S3 includes the following steps:

[0051] Step S31: performing initial mesh division according to the key heat dissipation path diagram and the simplified CAD model to obtain an initial mesh structure;

[0052] Step S32: performing finite element simulation on the initial grid structure according to the enhanced material property matrix, boundary conditions, and heat source definition data, and performing error analysis to obtain preliminary temperature field data;

[0053] Step S33: selecting an encrypted area based on the preliminary temperature field data to obtain an encrypted area;

[0054] Step S34: selecting a sparse region based on the path entropy value-added data and the path thermal impedance data to obtain a sparse region;

[0055] Step S35: applying the grid size control parameter to the dense area and the sparse area to obtain the grid size control parameter;

[0056] Step S36: setting encryption and sparse thresholds according to the grid size control parameters to obtain a grid adjustment strategy;

[0057] Step S37: performing mesh adjustment and iterative optimization on the simplified CAD model according to the mesh adjustment strategy to obtain an adaptive mesh structure;

[0058] Step S38: performing iterative grid optimization according to the enhanced material property matrix and the adaptive grid structure to obtain an optimized grid structure.

[0059] By performing initial meshing based on the key heat dissipation path diagram, the present invention can effectively control the number of meshes while ensuring simulation accuracy, improve simulation efficiency, and lay the foundation for subsequent adaptive mesh optimization. Performing a preliminary simulation and analyzing the errors can understand the accuracy of the initial mesh and provide guidance for subsequent mesh optimization, thereby targetedly improving mesh quality and enhancing the reliability of the simulation results. By analyzing preliminary temperature field data, areas with large temperature gradients, i.e., areas with drastic heat flux changes, are identified and marked as densified areas, providing clear targets for subsequent mesh densification. By analyzing path entropy value and thermal impedance data, areas that contribute less to overall heat dissipation are identified and marked as sparse areas. This can reduce the number of meshes and improve simulation efficiency without affecting simulation accuracy. Applying different mesh size control parameters to densified and sparse areas can achieve local densification and densification of the mesh, thereby improving overall simulation efficiency while ensuring simulation accuracy in key areas. Setting densification and densification thresholds can automatically determine which cells need densification or densification based on indicators such as temperature gradient, entropy value, and thermal impedance value, thereby achieving adaptive adjustment of the mesh. Through iterative optimization, the mesh is gradually adjusted to make the mesh distribution more reasonable and better adapt to changes in the temperature field, thereby improving simulation accuracy and making the simulation results more convergent and stable. The final iterative mesh optimization ensures that the final mesh can accurately capture the details of the temperature field and achieve the preset accuracy requirements for simulation results, providing reliable mesh data for subsequent thermal impedance prediction and design optimization.

[0060] Preferably, step S4 includes the following steps:

[0061] Step S41: performing operating condition design using the optimized grid structure and the enhanced material property matrix to obtain an operating condition parameter table;

[0062] Step S42: extracting thermal impedance data according to the operating parameter table to obtain a thermal impedance data set;

[0063] Step S43: constructing a thermal impedance prediction model based on a neural network according to the thermal impedance data set to obtain a thermal impedance prediction model;

[0064] Step S44: obtaining the 5G circuit board working condition data; inputting the 5G circuit board working condition data into the thermal impedance prediction model, performing a rapid temperature field calculation, and obtaining a rapid temperature field distribution;

[0065] Step S45: performing heat dissipation performance analysis on the rapid temperature field distribution to obtain a heat dissipation performance evaluation report;

[0066] Step S46: performing design iteration and optimization based on the heat dissipation performance evaluation report, the simplified CAD model, and the enhanced material property matrix to obtain an optimized design solution.

[0067] By designing multiple sets of different operating parameters, the present invention can comprehensively cover the various working environments and load conditions that 5G circuit boards may encounter, thereby training a more generalized neural network model, enabling it to accurately predict the thermal impedance under different working conditions. Extracting thermal impedance data under different working conditions and constructing a thermal impedance data set provide the necessary training data for training the neural network model. The diversity and integrity of the data directly affect the prediction accuracy and generalization ability of the model. Constructing a thermal impedance prediction model based on a neural network can quickly and accurately predict the thermal impedance under different working conditions, avoiding repeated time-consuming finite element simulations and significantly improving design efficiency. Using the trained neural network model, the temperature field distribution of the 5G circuit board under actual working conditions can be quickly calculated, providing fast and accurate temperature data for heat dissipation performance analysis and shortening the design cycle. By analyzing the rapid temperature field distribution, it is possible to evaluate whether the heat dissipation performance of the 5G circuit board meets the design requirements, identify potential heat dissipation problems, and provide guidance for design optimization. Based on the thermal performance evaluation report, combined with the simplified CAD model and enhanced material property matrix, the design of the 5G circuit board can be iteratively optimized, such as adjusting the size of the radiator, changing the material, or improving the thermal solution, ultimately obtaining an optimized design that meets the thermal requirements.

[0068] Preferably, the present invention further provides a thermal simulation system based on a 5G circuit board, which is used to execute the thermal simulation method based on a 5G circuit board as described above. The thermal simulation system based on a 5G circuit board includes:

[0069] The material database construction module is used to measure and calculate the material properties of 5G circuit boards to obtain thermal conductivity data and interface thermal resistance data. The material database is constructed based on the thermal conductivity data and interface thermal resistance data to obtain an enhanced material property matrix.

[0070] The topological structure entropy analysis module is used to construct a simplified model of the 5G circuit board and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; the heat transfer equation of the simplified CAD model is solved based on the boundary conditions and heat source definition data, and the temperature field is calculated to obtain temperature field distribution data; the simplified CAD model is analyzed for the influence of phase change latent heat based on the temperature field distribution data to obtain phase change latent heat data; entropy increase and thermal impedance analysis based on transient heat sources are performed based on the phase change latent heat data to obtain path entropy value-added data and path thermal impedance data; key paths are extracted based on the path entropy value-added data and path thermal impedance data, and hot spots are identified to obtain a key heat dissipation path diagram;

[0071] The adaptive grid optimization module is used to perform initial grid division according to the key heat dissipation path diagram, and perform adaptive grid optimization to obtain an adaptive grid structure; and perform grid iterative optimization based on the adaptive grid structure to obtain an optimized grid structure;

[0072] The neural network thermal impedance prediction module is used to extract thermal impedance data under different working conditions using an optimized grid structure to obtain a thermal impedance data set; construct a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; use the thermal impedance prediction model to perform rapid temperature field calculations to obtain a rapid temperature field distribution; perform heat dissipation performance analysis on the rapid temperature field distribution, and perform design iteration and optimization to obtain an optimized design solution.

[0073] Preferably, the circuit board is manufactured by the thermal simulation method based on the 5G circuit board as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic diagram of the steps of a thermal simulation method based on a 5G circuit board;

[0075] Figure 2 Schematic diagram of the detailed implementation steps of step S2 in the present invention.

[0076] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0077] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0078] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0079] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0080] To achieve this, please refer to Figures 1 to 2 , a thermal simulation method based on a 5G circuit board, comprising the following steps:

[0081] Step S1: Measure and calculate the material properties of the 5G circuit board to obtain thermal conductivity characteristic data and interface thermal resistance data; construct a material database based on the thermal conductivity characteristic data and interface thermal resistance data to obtain an enhanced material property matrix;

[0082] Step S2: construct a simplified model of the 5G circuit board, and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; solve the heat transfer equation of the simplified CAD model according to the boundary conditions and heat source definition data, and perform temperature field calculation to obtain temperature field distribution data; perform phase change latent heat influence analysis on the simplified CAD model according to the temperature field distribution data to obtain phase change latent heat data; perform entropy increase and thermal impedance analysis based on transient heat sources according to the phase change latent heat data to obtain path entropy value data and path thermal impedance data; perform key path extraction according to the path entropy value data and path thermal impedance data, and perform hotspot identification to obtain a key heat dissipation path diagram;

[0083] Step S3: performing initial mesh division according to the key heat dissipation path diagram, and performing adaptive mesh optimization to obtain an adaptive mesh structure; performing mesh iterative optimization based on the adaptive mesh structure to obtain an optimized mesh structure;

[0084] Step S4: Using the optimized grid structure to extract thermal impedance data under different working conditions to obtain a thermal impedance data set; constructing a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; using the thermal impedance prediction model to perform rapid temperature field calculation to obtain a rapid temperature field distribution; performing heat dissipation performance analysis on the rapid temperature field distribution, and performing design iteration and optimization to obtain an optimized design scheme.

[0085] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a flow chart showing the steps of a thermal simulation method based on a 5G circuit board according to the present invention. In this example, the thermal simulation method based on a 5G circuit board includes the following steps:

[0086] Step S1: Measure and calculate the material properties of the 5G circuit board to obtain thermal conductivity characteristic data and interface thermal resistance data; construct a material database based on the thermal conductivity characteristic data and interface thermal resistance data to obtain an enhanced material property matrix;

[0087] In this embodiment of the present invention, thermal properties of various materials used in 5G circuit boards, such as FR-4 substrates, copper conductors, chip packaging materials, and phase-change heat dissipation materials, were measured. Thermal diffusivity and specific heat capacity were measured using a flash thermal conductivity meter, while density was measured using a balance, and thermal conductivity was calculated. Simultaneously, molecular dynamics simulations were used to calculate nanoscale thermal conductivity and interfacial thermal resistance. The macroscopic measurement data and nanoscale simulation data were fused and calibrated to construct an enhanced material property matrix (EMPM) encompassing the thermal properties of all materials.

[0088] Step S2: construct a simplified model of the 5G circuit board, and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; solve the heat transfer equation of the simplified CAD model according to the boundary conditions and heat source definition data, and perform temperature field calculation to obtain temperature field distribution data; perform phase change latent heat influence analysis on the simplified CAD model according to the temperature field distribution data to obtain phase change latent heat data; perform entropy increase and thermal impedance analysis based on transient heat sources according to the phase change latent heat data to obtain path entropy value data and path thermal impedance data; perform key path extraction according to the path entropy value data and path thermal impedance data, and perform hotspot identification to obtain a key heat dissipation path diagram;

[0089] In this embodiment of the present invention, a CAD model of a 5G circuit board is imported into SpaceClaim for simplification. Details unimportant to thermal analysis, such as screw holes and chamfers, are removed, and complex components are replaced with simplified geometric shapes to produce a simplified CAD model. Based on the actual operating environment and chip power consumption data, boundary conditions and heat sources are defined in FloTHERM, including the air velocity and temperature for forced convection, as well as the average power consumption and fluctuation amplitude of the chip. A steady-state thermal simulation is performed using FloTHERM to obtain an initial temperature field distribution. Based on this initial temperature field distribution, the state of the phase change material is identified, and the latent heat of the phase change is calculated. This latent heat is added to the model as an additional heat source or heat sink. Considering the transient characteristics of chip power consumption, a sinusoidal function is used to simulate its time-varying variation, and a transient thermal simulation is performed. Based on the transient temperature field distribution, the entropy increase rate of each unit and the thermal impedance of the critical heat dissipation path are calculated. Based on the entropy increase and thermal impedance, critical heat dissipation paths and hotspots are identified, and a critical heat dissipation path map is generated.

[0090] Step S3: performing initial mesh division according to the key heat dissipation path diagram, and performing adaptive mesh optimization to obtain an adaptive mesh structure; performing mesh iterative optimization based on the adaptive mesh structure to obtain an optimized mesh structure;

[0091] In an embodiment of the present invention, a simplified CAD model is imported into Ansys Meshing, and initial meshing is performed based on the critical heat dissipation path diagram. A smaller mesh size is used near the critical heat dissipation paths and important components, while a larger mesh size is used in other areas. A hybrid meshing method is then used. The initial mesh is imported into Ansys Fluent, and material properties are assigned according to EMPM, along with boundary conditions and heat sources. A steady-state thermal simulation is performed, and the temperature field distribution error is analyzed. Based on the temperature gradient, entropy increase, and thermal impedance value, areas requiring densification and sparseness are selected, and the mesh size control parameters are adjusted. The mesh adjustment and simulation process is repeated until the temperature field distribution error meets the preset accuracy requirements, resulting in an optimized mesh structure.

[0092] Step S4: extracting thermal impedance data under different working conditions using the optimized grid structure to obtain a thermal impedance data set; constructing a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; performing rapid temperature field calculation using the thermal impedance prediction model to obtain a rapid temperature field distribution; performing heat dissipation performance analysis on the rapid temperature field distribution, and performing design iteration and optimization to obtain an optimized design solution;

[0093] In an embodiment of the present invention, an optimized grid structure and EMPM are used to design multiple sets of different working conditions, such as changing ambient temperature, wind speed, and heat source power. Each set of working conditions is simulated, the thermal impedance data of the heat source is extracted, and a thermal impedance data set is constructed. TensorFlow is used to build a multi-layer perceptron (MLP) neural network model, which takes ambient temperature, wind speed, and heat source power as input and thermal impedance as output. The neural network model is trained using the thermal impedance data set. The actual working condition data of the 5G circuit board is obtained and input into the trained neural network model to predict the thermal impedance value. The predicted thermal impedance value is used for fast temperature field calculation and heat dissipation performance analysis. If the heat dissipation performance does not meet the design requirements, the design parameters are adjusted according to the analysis results, such as modifying the geometric structure, replacing the material, or optimizing the heat dissipation solution, and the simulation and optimization process is repeated until the design goal is achieved.

[0094] Preferably, step S1 includes the following steps:

[0095] Step S11: measuring the material properties of the 5G circuit board to obtain a material property table;

[0096] Step S12: performing energy-minimized material NPT ensemble simulation according to the material property table to obtain equilibrium atomic trajectory data;

[0097] Step S13: Calculating heat conduction characteristics based on the equilibrium atomic trajectory data to obtain heat conduction characteristic data;

[0098] Step S14: Calculating the interface thermal resistance based on the equilibrium atomic trajectory data to obtain interface thermal resistance data;

[0099] Step S15: performing nanoscale material property modeling and calibration based on the heat conduction property data and the interface thermal resistance data to obtain a nanoscale material property table;

[0100] Step S16: performing data fusion and calibration according to the material property table and the nanoscale material property table to obtain an enhanced material property matrix.

[0101] In an embodiment of the present invention, for a 5G circuit board, the key materials requiring thermal simulation are first identified, such as the PCB substrate material (FR-4, Rogers, etc.), chip packaging material, heat sink material (copper, aluminum, graphene, etc.), and thermal interface material. The thermal diffusivity and specific heat capacity of the material are measured using a Netzsch LFA 467 NanoFlash flash thermal conductivity meter. The material to be tested is cut into circular samples with a diameter of 12.7 mm and a thickness of 1-3 mm. Before measurement, a layer of graphite is sprayed on the sample surface to increase the absorptivity. The measurement temperature range is set to 25°C to 125°C, with measurements taken at 10°C intervals. Each temperature point is measured three times, and the average value is taken as the final result. A Mettler Toledo XS205 balance is used to measure the mass and dimensions of the sample and calculate the sample density. Substitute the measured thermal diffusivity, specific heat, and density into the formula λ = α·ρ·c (λ is the thermal conductivity, α is the thermal diffusivity, ρ is the density, and c is the specific heat) to calculate the thermal conductivity of the material. Record all measured data in a material property table, including the material name, thermal diffusivity, specific heat, density, and calculated thermal conductivity, and note the measurement error.

[0102] Select a key material from the material properties table, such as FR-4, and construct its atomic structure model using Materials Studio software. FR-4 is primarily composed of epoxy resin and glass fiber. In this simplified model, cross-linked epoxy resin chains represent the epoxy resin matrix, with amorphous silica molecules embedded to represent the glass fiber reinforcement. Energy minimization was performed using the Forcite module, and interatomic interactions were calculated using the COMPASS II force field. The convergence criteria for energy minimization were set to an energy change of less than 1.0×10^-5 kcal / mol and a force of less than 0.001 kcal / mol / Å. After energy minimization, molecular dynamics simulations were performed using the Discover module in the NPT ensemble. The simulation temperature was set to 298 K, the pressure to 1 atm, the time step to 1 fs, and the total simulation duration to 100 ps. Atomic coordinates and velocities were saved every 100 fs to obtain equilibrium atomic trajectory data.

[0103] The thermal conductivity of the material was calculated using equilibrium atomic trajectory data and the Green-Kubo equation. The atomic trajectory data was imported into LAMMPS software, and the heat flux autocorrelation function was calculated using the `fix heatflux` command. The heat flux autocorrelation function was time-integrated to obtain the thermal conductivity value. The calculation was repeated three times, and the average result was taken as the final result. The calculation error was also recorded.

[0104] Construct an atomic structure model of the interface between two materials, such as the interface between FR-4 and copper. Use the same NPT ensemble simulation method as in step S12 to obtain equilibrium atomic trajectory data. Calculate the interfacial thermal resistance using non-equilibrium molecular dynamics (NEMD). During the simulation, apply a heat flux to one of the materials and measure the temperature difference between the two materials. The interfacial thermal resistance is calculated as R = ΔT / Q, where R is the interfacial thermal resistance, ΔT is the temperature difference, and Q is the heat flux.

[0105] Record the thermal conductivity and interfacial thermal resistance data calculated in steps S13 and S14 in a nanoscale material properties table. Based on this data, a more accurate nanoscale material model can be established, for example, by accounting for factors such as phonon scattering and interface effects. Compare this data with experimental measurements or literature values ​​to calibrate the model and improve its accuracy.

[0106] The macroscale material properties measured in step S11 and the nanoscale material properties calculated in step S15 are fused. For example, for thermal conductivity, a weighted average method can be used to combine the macroscale measured values ​​and the nanoscale calculated values. The weights can be determined based on the material's microstructure and measurement / calculation errors. The fused data is calibrated using existing experimental data or literature data to correct for any deviations. The resulting enhanced material property matrix (EMPM) contains the name of each material, its macroscale and nanoscale thermal parameters, and the fused and calibrated values ​​and error ranges.

[0107] Preferably, step S2 includes the following steps:

[0108] Step S21: Obtain a 5G circuit board CAD model and perform model simplification to obtain a simplified CAD model;

[0109] Step S22: defining boundary conditions and heat sources according to the enhanced material property matrix and the simplified CAD model to obtain boundary condition and heat source definition data;

[0110] Step S23: Solving the heat transfer equation for the simplified CAD model according to the boundary conditions and the heat source definition data, and performing temperature field calculation to obtain temperature field distribution data;

[0111] Step S24: performing a phase change latent heat influence analysis on the simplified CAD model based on the temperature field distribution data to obtain phase change latent heat data; performing an entropy increase and thermal impedance analysis based on a transient heat source based on the phase change latent heat data to obtain path entropy increase data and path thermal impedance data;

[0112] Step S25: extracting key paths based on the path entropy value-added data and the path thermal impedance data, and identifying hot spots to obtain a key heat dissipation path diagram.

[0113] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0114] Step S21: Obtain a 5G circuit board CAD model and perform model simplification to obtain a simplified CAD model;

[0115] In an embodiment of the present invention, a STEP format CAD model of a 5G circuit board is exported from Altium Designer. The STEP model is imported into SpaceClaim. The model integrity is checked to ensure that there are no missing or incorrect geometric elements. Components that have little impact on the thermal simulation results, such as screw holes, test points, and decorative features on the housing, are removed. Components with complex shapes, such as BGA packaged chips, are simplified into rectangular blocks, retaining their external dimensions and position information. For vias and through-hole structures on the PCB board, they are replaced with solid models of equivalent thermal conductivity based on their density and size. The copper trace network on the PCB is simplified into several representative copper-clad areas, and assigned corresponding thickness and conductivity properties. The simplified model retains all key components related to heat dissipation, such as chips, heat sinks, PCB substrates, etc., and ensures that their geometric shapes, sizes, and relative positions are accurate. The simplified model is saved as a new STEP file, i.e., the simplified CAD model.

[0116] Step S22: defining boundary conditions and heat sources according to the enhanced material property matrix and the simplified CAD model to obtain boundary condition and heat source definition data;

[0117] In an embodiment of the present invention, the simplified CAD model is imported into FloTHERM. Boundary conditions are defined based on the actual working conditions and environmental conditions of the 5G circuit board. Assume that the circuit board operates in a forced convection environment, set the wind speed to 3m / s, the wind direction is parallel to the surface of the circuit board, and the ambient temperature is 25°C. Enter these parameters into the corresponding boundary condition setting interface in FloTHERM. Define the heat source based on the power consumption data of the chip. For example, set the power consumption of the CPU chip to 3W, and apply the power consumption evenly to the simplified model surface corresponding to the CPU chip. Record the power, position, and effective area information of all heat sources in a table. Save the defined boundary conditions and heat source information as a file, i.e., the boundary conditions and heat source definition data.

[0118] Step S23: Solving the heat transfer equation for the simplified CAD model according to the boundary conditions and the heat source definition data, and performing temperature field calculation to obtain temperature field distribution data;

[0119] In an embodiment of the present invention, in FloTHERM, the simplified CAD model is meshed. Hexahedral meshes are mainly used, and local tetrahedral meshes are used for transition in geometrically complex areas. The mesh size is set to ensure that the mesh density in key areas is high enough. The material properties in the enhanced material property matrix (EMPM) obtained in step S16 are assigned to the corresponding model components. Based on the defined boundary conditions and heat sources, FloTHERM automatically establishes heat transfer control equations, including three heat transfer modes: heat conduction, convection, and radiation. Select the steady-state solver and set the convergence accuracy to 10^-4. Run the solver to calculate the temperature field distribution. The calculated temperature field data, including the temperature value of each grid node, is exported as a data file, namely the temperature field distribution data.

[0120] Step S24: performing a phase change latent heat influence analysis on the simplified CAD model based on the temperature field distribution data to obtain phase change latent heat data; performing an entropy increase and thermal impedance analysis based on a transient heat source based on the phase change latent heat data to obtain path entropy increase data and path thermal impedance data;

[0121] In this embodiment of the present invention, it is assumed that a phase-change heat dissipation material is used in a 5G circuit board. Based on the temperature field distribution data, the state of the phase-change material (solid or liquid) is determined. If the phase-change material is within the phase-change temperature range, the latent heat absorbed or released is calculated. The latent heat calculation formula is Q = m·L, where Q is the latent heat, m is the mass of the phase-change material, and L is the phase-change latent heat coefficient. The calculated latent heat data is added to the thermal simulation model as an additional heat source or heat sink. Assuming that the power consumption of the CPU chip varies over time, a sinusoidal function is used to simulate its transient characteristics, for example, P(t) = P_avg + P_amp * sin(ωt), where P_avg is the average power consumption, P_amp is the power consumption fluctuation amplitude, ω is the angular frequency, and t is time. The transient heat source is added to the FloTHERM model. The temperature field distribution is recalculated, and based on the new temperature field distribution, the entropy increase rate of each unit and the thermal impedance of the critical heat dissipation path are calculated. The calculated path entropy increase data and path thermal impedance data are saved separately.

[0122] Step S25: extracting key paths based on the path entropy value-added data and the path thermal impedance data, and identifying hot spots to obtain a key heat dissipation path diagram;

[0123] In this embodiment of the present invention, the paths with the highest entropy increase, representing heat dissipation bottlenecks, are identified based on the path entropy value data. The paths with the highest thermal impedance, representing paths with low heat dissipation efficiency, are identified based on the path thermal impedance data. These paths are marked as critical heat dissipation paths. The temperature field distribution data is used to identify the areas with the highest temperatures, representing hotspots. The critical heat dissipation paths and hotspot locations are marked on a simplified CAD model to generate a critical heat dissipation path map.

[0124] Preferably, step S24 includes the following steps:

[0125] Step S241: identifying the phase change material according to the enhanced material property matrix to obtain the phase change material; performing path selection and unit division according to the phase change material and the simplified CAD model to obtain path unit information;

[0126] Step S242: determining the phase change state based on the temperature field distribution data and the phase change material to obtain a phase change state mark; calculating the phase change latent heat based on the temperature field distribution data and the phase change material to obtain phase change latent heat data;

[0127] Step S243: performing material thermal response analysis according to the enhanced material property matrix to obtain material thermal response data;

[0128] Step S244: performing heat flux calculation based on the temperature field distribution data, the phase change latent heat data, and the material thermal response data to obtain heat flux data;

[0129] Step S245: Calculating the unit entropy increase rate based on the heat flux density data to obtain unit entropy increase rate data; performing path entropy value integration based on the unit entropy increase rate data and the path unit information to obtain path entropy value data;

[0130] Step S246: Calculating the thermal impedance of the component based on the temperature field distribution data to obtain component thermal impedance data; calculating the thermal impedance of the heat dissipation path based on the heat flux density data and the component thermal impedance data to obtain path thermal impedance data.

[0131] In an embodiment of the present invention, each material record in the enhanced material property matrix (EMPM) is traversed. If the name of the material contains the word "phase change" or its latent heat value is greater than 100 J / g, the material is identified as a phase change material, and its name, phase change temperature range, and latent heat value are recorded. In the simplified CAD model, the area containing the phase change material is identified. Taking the CPU chip as the heat source and the radiator as the heat sink as an example, a straight line path from the center of the CPU chip to the bottom surface of the radiator is defined. The path is discretized, and a series of cubic units are constructed with an interval of 0.1 mm and the points on the path as the unit center. The side length of each unit is 0.1 mm. The center coordinates, material type, and path number of each unit are recorded, and this information is stored as path unit information.

[0132] For each path unit, the temperature value of the unit is extracted from the temperature field distribution data based on its center coordinates. This temperature value is compared with the phase change temperature range of the phase change material. If the unit temperature is lower than the lower limit of the phase change temperature range, the unit is marked as solid; if the unit temperature is higher than the upper limit of the phase change temperature range, the unit is marked as liquid; if the unit temperature is within the phase change temperature range, the unit is marked as a phase change state. For units in a phase change state, their latent heat value is calculated based on their temperature value and the latent heat curve of the phase change material. The latent heat calculation formula can use a linear interpolation method to calculate the corresponding latent heat value based on the position of the unit temperature in its phase change temperature range. The phase change state mark and latent heat value of each unit are stored as phase change latent heat data.

[0133] For each material in the EMPM, calculate its thermal diffusivity. The formula for calculating thermal diffusivity is α = λ / (ρ·c), where α is the thermal diffusivity, λ is the thermal conductivity, ρ is the density, and c is the specific heat capacity. Thermal diffusivity reflects the material's response to temperature changes. Record the thermal diffusivity of each material as the material thermal response data.

[0134] For each path element, Fourier's law is used to calculate the heat flux density due to thermal conduction based on its temperature and the temperature values ​​of adjacent elements. The heat flux density due to thermal conduction is calculated as q = -λ·∇T, where q is the heat flux density, λ is the thermal conductivity, and ∇T is the temperature gradient. For elements containing phase change materials, the latent heat value calculated in step S242 is converted to an equivalent heat flux density and superimposed on the heat flux density due to thermal conduction. Based on the material thermal response data, the heat flux density is corrected to account for the thermal inertia effect of the material. The heat flux density vector for each element is recorded as the heat flux density data.

[0135] For each path unit, calculate its entropy increase rate based on its heat flux and temperature. The entropy increase rate is calculated as ds / dt = q / T, where ds / dt is the entropy increase rate, q is the heat flux, and T is the temperature. Record the entropy increase rate of each unit as the unit entropy increase rate data. For each path, integrate the entropy increase rates of all units along the path to obtain the path entropy increase. This integration method can use numerical integration methods, such as the trapezoidal formula or the Simpson formula. Record the entropy increase of each path as the path entropy increase data.

[0136] Based on the temperature field distribution data, the average temperature of the CPU chip and the ambient temperature are extracted. The component thermal impedance is calculated as R = (T_chip - T_amb) / P, where R is the component thermal impedance, T_chip is the average chip temperature, T_amb is the ambient temperature, and P is the chip power consumption. The calculated component thermal impedance is recorded as the component thermal impedance data. For each path, the path thermal impedance is calculated based on the heat flux density along the path and the component thermal impedance. The path thermal impedance is calculated as R_path = ∫(1 / λ)dx + R_interface + R_component, where R_path is the path thermal impedance, λ is the thermal conductivity of the path material, x is the path length, R_interface is the interface thermal resistance, and R_component is the component thermal impedance. The thermal impedance of each path is recorded as the path thermal impedance data.

[0137] Preferably, step S244 is specifically as follows:

[0138] Calculate the heat flux density of heat conduction according to the temperature field distribution data to obtain the heat flux density data of unit heat conduction;

[0139] Calculate the heat flux density of convective heat transfer according to the temperature field distribution data to obtain the heat flux density data of unit convective heat transfer;

[0140] Calculate the heat flux density of radiation heat transfer according to the temperature field distribution data to obtain the heat flux density data of unit radiation heat transfer;

[0141] Perform heat flux superposition on the unit heat conduction heat flux density data, the unit convection heat transfer heat flux density data, and the unit radiation heat transfer heat flux density data to obtain heat flux superposition data;

[0142] Perform transient heat source impact analysis based on temperature field distribution data to obtain transient heat source impact data;

[0143] The heat flux density distribution inside the unit is carried out according to the transient heat source influence data to obtain the local heat flux density influence data;

[0144] The heat flux direction is corrected on the heat flux superposition data according to the local heat flux influence data to obtain the heat flux data.

[0145] In this embodiment of the present invention, for each grid cell, the central node temperature and the central node temperatures of the six adjacent cells (assuming a structured hexahedral grid) are obtained. The temperature gradient of each cell in the x, y, and z directions is calculated using the central difference method. For example, in the x direction, the temperature gradient calculation formula is ∇T_x = (T_x+Δx- T_x-Δx) / (2Δx), where T_x+Δx and T_x-Δx are the temperatures of the adjacent cells in the x direction, and Δx is the length of the cell in the x direction. According to Fourier's law, the heat conduction heat flux density components of each cell in the x, y, and z directions are calculated: q_x = -λ·∇T_x, q_y = -λ·∇T_y, q_z = -λ·∇T_z, where λ is the thermal conductivity of the material in which the cell is located, taken from the enhanced material property matrix (EMPM). The three heat flux components (q_x, q_y, q_z) of each unit are combined into a vector as the heat conduction heat flux data of the unit and stored in the unit heat conduction heat flux data set.

[0146] For each cell on the surface of a 5G circuit board, the convective heat flux is calculated based on the pre-set convective heat transfer coefficient h, the ambient temperature T_amb, and the cell surface temperature T_surf (using the cell center node temperature as an approximation). The convective heat flux is calculated using the formula q_conv = h·(T_surf - T_amb). The calculated heat flux value is used as the cell's normal vector, pointing away from the board surface. For internal cells, the convective heat flux is zero. The convective heat flux data for each cell is stored in the cell convective heat flux dataset.

[0147] For each cell on the surface of a 5G circuit board, the radiation heat flux is calculated according to the Stefan-Boltzmann law. The radiation heat flux is calculated as q_rad = ε·σ·(T_surf^4 - T_amb^4), where ε is the emissivity of the cell's material, σ is the Stefan-Boltzmann constant, T_surf is the cell surface temperature, and T_amb is the ambient temperature. The calculated heat flux value is used as the cell's normal vector, pointing away from the board surface. For internal cells, the radiation heat flux is zero. The radiation heat flux data for each cell is stored in the cell radiation heat flux dataset.

[0148] For each element, the three vectors of heat flux density due to conduction, heat flux density due to convection, and heat flux density due to radiation are superimposed to obtain the total heat flux density for that element. The total heat flux density data for each element is stored in the heat flux superposition dataset.

[0149] Based on the CPU chip's transient power consumption curve, for example, P(t) = P_avg + P_amp * sin(ωt), calculate the power consumption value at each moment. Divide the power consumption value at each moment by the chip's surface area to obtain the transient heat flux density value. Store the heat flux density value at each moment in the transient heat source impact dataset.

[0150] For the cells corresponding to the CPU chip, based on the transient heat source impact data, the transient heat flux values ​​are evenly distributed within the chip unit. Assuming the chip unit is further subdivided into several subunits, the transient heat flux values ​​are evenly distributed to each subunit. The heat flux data for each subunit is stored in the local heat flux impact dataset.

[0151] For the unit corresponding to the CPU chip, the total heat flux direction of that unit in the heat flux overlay data is corrected based on the heat flux direction of each subunit in the local heat flux impact data. For example, the average vector of the heat flux directions of all subunits can be calculated and used as the corrected total heat flux direction. The corrected heat flux data is stored in the final heat flux dataset.

[0152] Preferably, the transient heat source impact analysis is performed based on the temperature field distribution data as follows:

[0153] Identify the heat source location based on the temperature field distribution data and the simplified CAD model to obtain heat source location identification data;

[0154] Determine the geometric shape based on the heat source position identification data to obtain the heat source geometric information;

[0155] According to the heat source geometry information, the heat source power is analyzed over time to obtain the heat source power time function;

[0156] Selecting a transient heat source model according to heat source geometry information to obtain a selected transient heat source model;

[0157] The transient heat source heat flux density is calculated based on the selected transient heat source model and the heat source power-time function to obtain the transient heat source impact data.

[0158] In an embodiment of the present invention, the simplified CAD model is imported into a program library capable of geometric analysis, such as OpenCASCADE. All geometric entities in the simplified CAD model are traversed. For each entity, its predefined attributes in the model are checked, such as name or label. If the name of the entity contains keywords such as "CPU" or "GPU", or its label is set to "heat source", the entity is identified as a heat source. The geometric center coordinates (x, y, z) and bounding box information of the heat source entity are extracted, including the minimum vertex coordinates (x_min, y_min, z_min) and maximum vertex coordinates (x_max, y_max, z_max) of the bounding box. The geometric center coordinates and bounding box information of the heat source are stored in the heat source position identification data structure.

[0159] Based on the bounding box information in the heat source location identification data, calculate the heat source's dimensions, such as length L = x_max - x_min, width W = y_max - y_min, and height H = z_max - z_min. Determine the heat source's shape type based on the actual geometry of the heat source entity in the simplified CAD model. If the heat source entity is a rectangular parallelepiped, record its shape type as "rectangular parallelepiped" along with its length, width, and height. If the heat source entity is a cylinder, record its shape type as "cylinder" along with its radius and height. Store the heat source's shape type and dimensions in the heat source geometry information data structure.

[0160] Assume that the power of the heat source varies sinusoidally over time. Define the heat source power-time function as P(t) = P_avg + P_amp * sin(2πft + φ), where P_avg is the average power consumption, P_amp is the power consumption fluctuation amplitude, f is the frequency, φ is the phase, and t is time. Based on the actual operating conditions of a 5G circuit board, set P_avg = 3W, P_amp = 1W, f = 1Hz, and φ = 0. Substituting these parameters into the heat source power-time function, we obtain P(t) = 3 + 1 * sin(2πt).

[0161] Since the heat source is a rectangular parallelepiped or cylindrical geometry and its power varies sinusoidally with time, the body heat source model is selected as the transient heat source model. The body heat source model assumes that the heat is uniformly distributed within the volume of the heat source.

[0162] At each time step Δt, for example, Δt = 0.01s, calculate the instantaneous power P(t_i) of the heat source based on the heat source power-time function P(t), where t_i = i * Δt, and i is the time step index. Divide the instantaneous power P(t_i) by the heat source's volume V to obtain the transient heat flux q(t_i) = P(t_i) / V. Calculate the heat source's volume V based on the heat source's geometry. For example, if the heat source is a rectangular parallelepiped, V = L * W * H. If the heat source is a cylinder, V = πr^2h, where r is the radius and h is the height. Store the transient heat flux q(t_i) at each time step in the transient heat source influence data structure, which contains the time step index i and the corresponding heat flux value q(t_i).

[0163] Preferably, step S3 includes the following steps:

[0164] Step S31: performing initial mesh division according to the key heat dissipation path diagram and the simplified CAD model to obtain an initial mesh structure;

[0165] Step S32: performing finite element simulation on the initial grid structure according to the enhanced material property matrix, boundary conditions, and heat source definition data, and performing error analysis to obtain preliminary temperature field data;

[0166] Step S33: selecting an encrypted area based on the preliminary temperature field data to obtain an encrypted area;

[0167] Step S34: selecting a sparse region based on the path entropy value-added data and the path thermal impedance data to obtain a sparse region;

[0168] Step S35: applying the grid size control parameter to the dense area and the sparse area to obtain the grid size control parameter;

[0169] Step S36: setting encryption and sparse thresholds according to the grid size control parameters to obtain a grid adjustment strategy;

[0170] Step S37: performing mesh adjustment and iterative optimization on the simplified CAD model according to the mesh adjustment strategy to obtain an adaptive mesh structure;

[0171] Step S38: performing iterative grid optimization according to the enhanced material property matrix and the adaptive grid structure to obtain an optimized grid structure.

[0172] In an embodiment of the present invention, the simplified CAD model is imported into the Ansys Meshing module. Based on the key heat dissipation path diagram, a smaller mesh size, such as 0.2 mm, is set near the key heat dissipation paths and important components such as heat sources and radiators. In other areas, a larger mesh size, such as 0.5 mm, is set. A hybrid meshing method is adopted, using hexahedral meshes in key areas and near important components to improve accuracy, and using tetrahedral meshes in other areas to simplify the mesh generation process. An initial mesh is generated, and the mesh quality is checked to ensure that the mesh's aspect ratio, distortion, and skewness meet the simulation accuracy requirements. The generated initial mesh and information such as node coordinates and unit connection relationships are saved as an initial mesh structure data file.

[0173] Import the initial mesh structure into the Ansys Fluent module. Based on the Enhanced Material Property Matrix (EMPM), assign corresponding material properties to each mesh element, including thermal conductivity, density, and specific heat capacity. Set the boundary conditions and heat source according to the boundary condition and heat source definition data. Select the steady-state heat conduction solver and set the convergence residual to 1e-6. Run the simulation calculation to obtain the preliminary temperature field distribution. Calculate the temperature gradient of each unit and compare the temperature difference between adjacent units. If the maximum temperature difference exceeds the preset threshold, such as 0.1°C, it is considered that the error is large and the mesh needs to be refined. Save the calculated temperature field distribution data and error analysis results as a preliminary temperature field data file.

[0174] Analyze the preliminary temperature field data and calculate the temperature gradient for each cell. Mark cells whose temperature gradient exceeds a preset threshold, such as 10°C / mm, as cells requiring infill. Define the area containing these cells as the infill region, and record the boundary information of the infill region, such as a bounding box or geometry.

[0175] Analyze the path entropy value and path thermal impedance data. Mark cells on the path where both entropy value and thermal impedance values ​​are below a preset threshold as cells that can be sparsed. Define the area where these cells are located as a sparse region, and record the boundary information of the sparse region.

[0176] For dense areas, set a smaller grid size control parameter, such as 0.1 mm. For sparse areas, set a larger grid size control parameter, such as 1 mm. Store the grid size control parameter for each area in the grid size control parameter data structure.

[0177] Based on the mesh size control parameters, set the density and thinning thresholds. For example, set the density threshold to a temperature gradient of 5°C / mm and the thinning threshold to a temperature gradient of 1°C / mm. If the temperature gradient of a cell exceeds the density threshold, the mesh is denser; if the temperature gradient of a cell is below the thinning threshold, the mesh is thinned. These thresholds and the corresponding operation (density or thinning) are stored in the mesh adjustment strategy data structure.

[0178] Re-import the simplified CAD model into the Ansys Meshing module. Based on the mesh adjustment strategy, perform mesh adjustments on the dense and sparse areas. For the dense areas, reduce the mesh size to the set control parameter value. For the sparse areas, increase the mesh size to the set control parameter value. Regenerate the mesh and check the mesh quality. Repeat steps S32 to S36 for iterative optimization until the temperature field distribution error meets the preset accuracy requirements. Save the final mesh, along with information such as node coordinates and element connectivity, as an adaptive mesh structure data file.

[0179] Import the adaptive mesh structure into the Ansys Fluent module. Repeat the simulation process in step S32 and perform error analysis. If the error is still large, further adjust the mesh size control parameters and the density / sparseness threshold based on the error distribution, and repeat step S37 until the temperature field distribution error meets the preset accuracy requirements. Save the resulting mesh, along with information such as node coordinates and element connectivity, as an optimized mesh structure data file.

[0180] Preferably, step S4 includes the following steps:

[0181] Step S41: performing operating condition design using the optimized grid structure and the enhanced material property matrix to obtain an operating condition parameter table;

[0182] Step S42: extracting thermal impedance data according to the operating parameter table to obtain a thermal impedance data set;

[0183] Step S43: constructing a thermal impedance prediction model based on a neural network according to the thermal impedance data set to obtain a thermal impedance prediction model;

[0184] Step S44: obtaining the 5G circuit board working condition data; inputting the 5G circuit board working condition data into the thermal impedance prediction model, performing a rapid temperature field calculation, and obtaining a rapid temperature field distribution;

[0185] Step S45: performing heat dissipation performance analysis on the rapid temperature field distribution to obtain a heat dissipation performance evaluation report;

[0186] Step S46: performing design iteration and optimization based on the heat dissipation performance evaluation report, the simplified CAD model, and the enhanced material property matrix to obtain an optimized design solution.

[0187] In an embodiment of the present invention, key parameters affecting the thermal performance of a 5G circuit board, such as ambient temperature, wind speed, and heat source power, are determined. The variation range of each parameter is set. The ambient temperature variation range is set to 20°C to 80°C, with a step size of 10°C. The wind speed variation range is set to 1 m / s to 5 m / s, with a step size of 1 m / s. The heat source power variation range is set to 1W to 5 W, with a step size of 1 W. Using the Latin hypercube sampling method, 100 sets of different operating parameter combinations are generated. Each set of operating parameter combinations, including ambient temperature, wind speed, and heat source power, and the corresponding enhanced material property matrix (EMPM) data, are stored in the operating parameter table.

[0188] For each set of operating parameters in the operating parameter table, import the optimized mesh structure into Ansys Fluent. Based on the operating parameters, set boundary conditions and heat sources. Run the simulation to obtain the temperature field distribution. Extract the average temperature of the heat source and the ambient temperature to calculate the thermal impedance. The thermal impedance calculation formula is R = (T_source - T_amb) / P, where R is the thermal impedance, T_source is the average temperature of the heat source, T_amb is the ambient temperature, and P is the heat source power. Each set of operating parameters and its corresponding thermal impedance value are stored in a thermal impedance dataset.

[0189] Use TensorFlow to build a multi-layer perceptron (MLP) neural network model. The input layer contains three neurons, one for ambient temperature, wind speed, and heat source power. The output layer contains one neuron, corresponding to thermal impedance. Set up two hidden layers, each with 10 neurons, and use the Reluctant Unit (ReLU) activation function. Use the mean squared error (MSE) as the loss function and the Adam optimizer for training. Split the thermal impedance dataset into a training set and a test set with an 8:2 ratio. Use the training set to train the neural network model, and use the test set to evaluate the model's prediction accuracy. Save the trained neural network model as the thermal impedance prediction model.

[0190] Obtain actual operating condition data for the 5G circuit board, including ambient temperature, wind speed, and heat source power. Input this data into a trained thermal impedance prediction model to obtain predicted thermal impedance values. Substitute the predicted thermal impedance values ​​into a simplified thermal resistance network model to calculate the temperatures of key nodes. For example, a one-dimensional thermal resistance network model can be used to connect the heat source, heat sink, and ambient temperature nodes. The temperature of each node can be calculated based on the thermal impedance. The calculated temperature distribution data is stored as fast temperature field distribution data.

[0191] Analyze rapid temperature field distribution data to extract key indicators, such as maximum temperature, average temperature, and temperature gradient. Compare these indicators with pre-set design requirements to assess whether heat dissipation performance meets the requirements. Compile the analysis results into a heat dissipation performance evaluation report.

[0192] If the thermal performance evaluation report indicates that the heat dissipation performance does not meet the design requirements, design iteration and optimization are required. Based on the issues identified in the evaluation report, such as excessively high maximum temperatures, the simplified CAD model can be modified, such as increasing the heat sink size or changing the heat sink material. Based on the Enhanced Material Property Matrix (EMPM), a material with better thermal conductivity can be selected. Repeat steps S41 to S45 until the heat dissipation performance meets the design requirements. The final design solution, including the geometry, material selection, and heat dissipation solution, is stored as the optimized design solution.

[0193] Preferably, the present invention further provides a thermal simulation system based on a 5G circuit board, which is used to execute the thermal simulation method based on a 5G circuit board as described above. The thermal simulation system based on a 5G circuit board includes:

[0194] The material database construction module is used to measure and calculate the material properties of 5G circuit boards to obtain thermal conductivity data and interface thermal resistance data. The material database is constructed based on the thermal conductivity data and interface thermal resistance data to obtain an enhanced material property matrix.

[0195] The topological structure entropy analysis module is used to construct a simplified model of the 5G circuit board and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; the heat transfer equation of the simplified CAD model is solved based on the boundary conditions and heat source definition data, and the temperature field is calculated to obtain temperature field distribution data; the simplified CAD model is analyzed for the influence of phase change latent heat based on the temperature field distribution data to obtain phase change latent heat data; entropy increase and thermal impedance analysis based on transient heat sources are performed based on the phase change latent heat data to obtain path entropy value-added data and path thermal impedance data; key paths are extracted based on the path entropy value-added data and path thermal impedance data, and hot spots are identified to obtain a key heat dissipation path diagram;

[0196] The adaptive grid optimization module is used to perform initial grid division according to the key heat dissipation path diagram, and perform adaptive grid optimization to obtain an adaptive grid structure; and perform grid iterative optimization based on the adaptive grid structure to obtain an optimized grid structure;

[0197] The neural network thermal impedance prediction module is used to extract thermal impedance data under different working conditions using an optimized grid structure to obtain a thermal impedance data set; construct a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; use the thermal impedance prediction model to perform rapid temperature field calculations to obtain a rapid temperature field distribution; perform heat dissipation performance analysis on the rapid temperature field distribution, and perform design iteration and optimization to obtain an optimized design solution.

[0198] Preferably, the circuit board is manufactured by the thermal simulation method based on the 5G circuit board as described above.

[0199] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0200] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A thermal simulation method based on a 5G circuit board, characterized in that: The following steps are involved: Step S1: Measure and calculate the material properties of the 5G circuit board to obtain thermal conductivity data and interface thermal resistance data; A material database is constructed based on the heat conduction characteristic data and the interface thermal resistance data to obtain an enhanced material property matrix, and the enhanced material property matrix is ​​called by traversal; wherein step S1 includes the following steps: Step S11: measuring the material properties of the 5G circuit board to obtain a material property table; Step S12: performing energy-minimized material NPT ensemble simulation according to the material property table to obtain equilibrium atomic trajectory data; Step S13: Calculating heat conduction characteristics based on the equilibrium atomic trajectory data to obtain heat conduction characteristic data; Step S14: Calculating the interface thermal resistance based on the equilibrium atomic trajectory data to obtain interface thermal resistance data; Step S15: performing nanoscale material property modeling and calibration based on the heat conduction property data and the interface thermal resistance data to obtain a nanoscale material property table; Step S16: performing data fusion and calibration based on the material property table and the nanoscale material property table to obtain an enhanced material property matrix; Step S2: construct a simplified model of the 5G circuit board, and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; solve the heat transfer equation of the simplified CAD model according to the boundary conditions and heat source definition data, and perform temperature field calculation to obtain temperature field distribution data; perform phase change latent heat influence analysis on the simplified CAD model according to the temperature field distribution data to obtain phase change latent heat data; perform entropy increase and thermal impedance analysis based on transient heat sources according to the phase change latent heat data to obtain path entropy value data and path thermal impedance data; perform key path extraction according to the path entropy value data and path thermal impedance data, and perform hotspot identification to obtain a key heat dissipation path diagram; Step S3: performing initial mesh division according to the key heat dissipation path diagram, and performing adaptive mesh optimization to obtain an adaptive mesh structure; performing mesh iterative optimization according to the adaptive mesh structure to obtain an optimized mesh structure; wherein step S3 includes the following steps: Step S31: performing initial mesh division according to the key heat dissipation path diagram and the simplified CAD model to obtain an initial mesh structure; Step S32: performing finite element simulation on the initial grid structure according to the enhanced material property matrix, boundary conditions, and heat source definition data, and performing error analysis to obtain preliminary temperature field data; Step S33: selecting an encrypted area based on the preliminary temperature field data to obtain an encrypted area; Step S34: selecting a sparse region based on the path entropy value-added data and the path thermal impedance data to obtain a sparse region; Step S35: applying the grid size control parameter to the dense area and the sparse area to obtain the grid size control parameter; Step S36: setting encryption and sparse thresholds according to the grid size control parameters to obtain a grid adjustment strategy; Step S37: performing mesh adjustment and iterative optimization on the simplified CAD model according to the mesh adjustment strategy to obtain an adaptive mesh structure; Step S38: performing iterative grid optimization according to the enhanced material property matrix and the adaptive grid structure to obtain an optimized grid structure; Step S4: Using the optimized grid structure to extract thermal impedance data under different working conditions to obtain a thermal impedance data set; constructing a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; using the thermal impedance prediction model to perform rapid temperature field calculation to obtain a rapid temperature field distribution; performing heat dissipation performance analysis on the rapid temperature field distribution, and performing design iteration and optimization to obtain an optimized design scheme.

2. The thermal simulation method based on the 5G circuit board according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Obtain a 5G circuit board CAD model and perform model simplification to obtain a simplified CAD model; Step S22: defining boundary conditions and heat sources according to the enhanced material property matrix and the simplified CAD model to obtain boundary condition and heat source definition data; Step S23: Solving the heat transfer equation for the simplified CAD model according to the boundary conditions and the heat source definition data, and performing temperature field calculation to obtain temperature field distribution data; Step S24: performing a phase change latent heat influence analysis on the simplified CAD model based on the temperature field distribution data to obtain phase change latent heat data; performing an entropy increase and thermal impedance analysis based on a transient heat source based on the phase change latent heat data to obtain path entropy increase data and path thermal impedance data; Step S25: extracting key paths based on the path entropy value-added data and the path thermal impedance data, and identifying hot spots to obtain a key heat dissipation path diagram.

3. The thermal simulation method based on the 5G circuit board according to claim 2, characterized in that: Step S24 includes the following steps: Step S241: identifying the phase change material according to the enhanced material property matrix to obtain the phase change material; performing path selection and unit division according to the phase change material and the simplified CAD model to obtain path unit information; Step S242: determining the phase change state based on the temperature field distribution data and the phase change material to obtain a phase change state mark; calculating the phase change latent heat based on the temperature field distribution data and the phase change material to obtain phase change latent heat data; Step S243: performing material thermal response analysis according to the enhanced material property matrix to obtain material thermal response data; Step S244: performing heat flux calculation based on the temperature field distribution data, the phase change latent heat data, and the material thermal response data to obtain heat flux data; Step S245: Calculating the unit entropy increase rate based on the heat flux density data to obtain unit entropy increase rate data; performing path entropy value integration based on the unit entropy increase rate data and the path unit information to obtain path entropy value data; Step S246: Calculating the thermal impedance of the component based on the temperature field distribution data to obtain component thermal impedance data; calculating the thermal impedance of the heat dissipation path based on the heat flux density data and the component thermal impedance data to obtain path thermal impedance data.

4. The thermal simulation method based on the 5G circuit board according to claim 3 is characterized in that: Step S244 is specifically as follows: Calculate the heat flux density of heat conduction according to the temperature field distribution data to obtain the heat flux density data of unit heat conduction; Calculate the heat flux density of convective heat transfer according to the temperature field distribution data to obtain the heat flux density data of unit convective heat transfer; Calculate the heat flux density of radiation heat transfer according to the temperature field distribution data to obtain the heat flux density data of unit radiation heat transfer; Perform heat flux superposition on the unit heat conduction heat flux density data, the unit convection heat transfer heat flux density data, and the unit radiation heat transfer heat flux density data to obtain heat flux superposition data; Perform transient heat source impact analysis based on temperature field distribution data to obtain transient heat source impact data; The heat flux density distribution inside the unit is carried out according to the transient heat source influence data to obtain the local heat flux density influence data; The heat flux direction is corrected on the heat flux superposition data according to the local heat flux influence data to obtain the heat flux data.

5. The thermal simulation method based on the 5G circuit board according to claim 4 is characterized in that: The transient heat source impact analysis based on the temperature field distribution data is as follows: Identify the heat source location based on the temperature field distribution data and the simplified CAD model to obtain heat source location identification data; Determine the geometric shape based on the heat source position identification data to obtain the heat source geometric information; According to the heat source geometry information, the heat source power is analyzed over time to obtain the heat source power time function; Selecting a transient heat source model according to heat source geometry information to obtain a selected transient heat source model; The transient heat source heat flux density is calculated based on the selected transient heat source model and the heat source power-time function to obtain the transient heat source impact data.

6. The thermal simulation method based on a 5G circuit board according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing operating condition design using the optimized grid structure and the enhanced material property matrix to obtain an operating condition parameter table; Step S42: extracting thermal impedance data according to the operating parameter table to obtain a thermal impedance data set; Step S43: constructing a thermal impedance prediction model based on a neural network according to the thermal impedance data set to obtain a thermal impedance prediction model; Step S44: obtaining the 5G circuit board working condition data; inputting the 5G circuit board working condition data into the thermal impedance prediction model, performing a rapid temperature field calculation, and obtaining a rapid temperature field distribution; Step S45: performing heat dissipation performance analysis on the rapid temperature field distribution to obtain a heat dissipation performance evaluation report; Step S46: performing design iteration and optimization based on the heat dissipation performance evaluation report, the simplified CAD model, and the enhanced material property matrix to obtain an optimized design solution.

7. A thermal simulation system based on a 5G circuit board, characterized in that: Used to perform the thermal simulation method based on the 5G circuit board according to claim 1, the thermal simulation system based on the 5G circuit board comprises: The material database construction module is used to measure and calculate the material properties of 5G circuit boards to obtain thermal conductivity data and interface thermal resistance data. The material database is constructed based on the thermal conductivity data and interface thermal resistance data to obtain an enhanced material property matrix. The topological structure entropy analysis module is used to construct a simplified model of the 5G circuit board and define boundary conditions and heat sources to obtain a simplified CAD model and boundary condition and heat source definition data; the heat transfer equation of the simplified CAD model is solved based on the boundary conditions and heat source definition data, and the temperature field is calculated to obtain temperature field distribution data; the simplified CAD model is analyzed for the influence of phase change latent heat based on the temperature field distribution data to obtain phase change latent heat data; entropy increase and thermal impedance analysis based on transient heat sources are performed based on the phase change latent heat data to obtain path entropy value-added data and path thermal impedance data; key paths are extracted based on the path entropy value-added data and path thermal impedance data, and hot spots are identified to obtain a key heat dissipation path diagram; The adaptive grid optimization module is used to perform initial grid division according to the key heat dissipation path diagram, and perform adaptive grid optimization to obtain an adaptive grid structure; and perform grid iterative optimization based on the adaptive grid structure to obtain an optimized grid structure; The neural network thermal impedance prediction module is used to extract thermal impedance data under different working conditions using an optimized grid structure to obtain a thermal impedance data set; construct a thermal impedance prediction model based on the thermal impedance data set to obtain a thermal impedance prediction model; use the thermal impedance prediction model to perform rapid temperature field calculations to obtain a rapid temperature field distribution; perform heat dissipation performance analysis on the rapid temperature field distribution, and perform design iteration and optimization to obtain an optimized design solution.

8. A circuit board, characterized in that: The circuit board is manufactured by the thermal simulation method based on a 5G circuit board as described in any one of claims 1 to 6.

Images