Probe card simulation method and system based on counter-acting force analysis

Abstract

The invention discloses a probe card simulation method and system based on counter-acting force analysis, which are used in the field of wafer simulation test, and the probe card simulation method comprises the following steps: integrating multiple physical field working conditions to construct a probe card geometric model, and importing the probe card geometric model into finite element simulation analysis software; carrying out grid division on the geometric model of the probe card by utilizing the hybrid voxel tree model, adjusting the grid density according to the distribution requirement of the multi-physical field, and setting a simulation boundary condition by utilizing a preset boundary condition template according to the actual simulation requirement; starting a simulation calculation process based on the adjusted grid density and simulated boundary conditions; and providing a visual interface of physical field distribution and a structural stress state in the simulation calculation process, and performing iterative optimization on the performance of the probe card in the simulation process according to a reactive force distribution result. According to the method, the grid density is automatically adjusted to adapt to changes of a physical field, and it is ensured that higher simulation precision is obtained in a key area.

Description

Technical Field

[0001] The present invention relates to the field of wafer simulation testing, and in particular, to a probe card simulation method and system based on reaction force analysis. Background Art

[0002] Probe cards are a key tool widely used in semiconductor manufacturing to test the electrical performance of integrated circuits (ICs) or chips. They typically contain hundreds to thousands of tiny pins (probes) that must precisely contact tiny contact points on the chip to ensure comprehensive functional and performance testing before the chip is packaged.

[0003] Probe cards must meet highly precise physical and electrical performance standards. Simulation can help engineers predict and optimize the performance of probe cards in actual operation, especially how to effectively transmit signals and withstand mechanical stress. Through simulation, engineers can verify the effectiveness of probe card design before physical manufacturing, thereby reducing costly prototype testing. Simulation results can be used to adjust the design of probe cards to avoid potential design defects. Probe cards must maintain performance stability under different test environments (for example, temperature and humidity changes). Simulation helps engineers evaluate and optimize the behavior of probe cards under various environmental conditions. Probe cards will repeatedly contact the chip during use, which may cause the probe to deform or break. By analyzing the structural stress and fatigue life of the probe card through simulation, these problems can be predicted and solved in advance.

[0004] Traditional meshing methods are often unable to adaptively adjust mesh density when dealing with complex geometric structures or multi-physics field environments, resulting in insufficient simulation accuracy in key areas (such as high stress concentration or high temperature areas). Fixed mesh strategies may use fine meshes evenly across the entire model, increasing computational load and time, especially in areas where the physical field does not change much. This method is inefficient and consumes unnecessary computing resources. Design optimization in traditional methods often relies on manual iterations and lacks systematic optimization algorithm support, making the design improvement process slow and difficult to effectively optimize for performance requirements under complex conditions. Previous technologies may only focus on a single physical field (such as only focusing on thermal or structural performance) and lack comprehensive performance evaluation of probe cards under composite loading (thermal, structural, electromagnetic, etc.) conditions. In traditional optimization methods, over-reliance on greedy strategies may lead to local optimal solutions, and there is a lack of effective strategies to balance the relationship between exploring new possibilities and utilizing known optimal solutions.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In order to overcome the above problems, the present invention aims to propose a probe card simulation method and system based on reaction force analysis, with the aim of solving the problem that design optimization in traditional methods often relies on manual iteration and lacks systematic optimization algorithm support, which makes the design improvement process slow and difficult to effectively optimize performance requirements under complex conditions. Previous technologies may only focus on a single physical field (such as only focusing on thermal performance or structural performance), and lack comprehensive performance evaluation of the probe card under composite loading (thermal, structural, electromagnetic, etc.) conditions.

[0007] To this end, the specific technical solution adopted by the present invention is as follows: According to one aspect of the present invention, a probe card simulation method based on reaction force analysis is provided, and the probe card simulation method comprises the following steps: S1. Integrate multiple physical field conditions to build a probe card geometry model, and import the probe card geometry model into finite element simulation analysis software; S2. Meshing the probe card geometry model using a hybrid voxel tree model, adjusting the mesh density according to the distribution requirements of the multi-physical fields, and setting the boundary conditions of the simulation using a preset boundary condition template according to the actual simulation requirements; starting the simulation calculation process based on the adjusted mesh density and the boundary conditions of the simulation; S3. Provide a visual interface for physical field distribution and structural stress state during simulation calculation, and iteratively optimize the performance of the probe card during simulation based on the reaction force distribution results; The hybrid voxel tree model is used to mesh the probe card geometry model, and the mesh density is adjusted according to the distribution requirements of the multi-physical field. At the same time, the boundary conditions of the simulation are set using the preset boundary condition template according to the actual simulation requirements. The simulation calculation process is started based on the adjusted mesh density and the boundary conditions of the simulation, including the following steps: S21, obtaining material properties and performance parameters of each part in the probe card geometric model; S22, meshing the loaded probe card geometry model based on the hybrid voxel tree model; S23. In the finite element simulation analysis software, setting the simulation boundary conditions according to the working environment and actual use conditions of the probe card; S24, using the established hybrid voxel tree model and the set boundary conditions, starting the multi-physics simulation to calculate the physical response and chemical response of the probe card under the set loading conditions; S25, analyzing the performance of the probe card according to the simulation calculation results to determine whether it meets the predetermined performance standards and safety requirements; and adjusting the geometric structure, material properties and performance parameters of the probe card according to the determination results; S26. If the geometry, material properties and performance parameters of the probe card do not meet the preset targets, repeat steps S21-S25, and adjust the meshing, boundary conditions and physical loading conditions according to the preset targets. After each iteration, re-evaluate the performance of the probe card during the simulation. If the geometry, material properties and performance parameters of the probe card do not meet the preset targets, output the geometry, material properties and performance parameters of the probe card.

[0008] Optionally, integrating multiple physical field conditions to construct a probe card geometric model, and importing the probe card geometric model into a finite element simulation analysis software includes the following steps: S11, obtaining functional requirements of multi-physics fields in the probe card, and using mapping software to construct a probe card geometric model according to the functional requirements of the multi-physics fields; S12, assigning corresponding material properties and performance parameters to various parts of the probe card geometric model, and configuring various physical loading conditions encountered by the probe card in actual operation; S13, exporting the configured geometric model and converting it into a file in a format that matches the finite element simulation analysis software; S14. Load the probe card geometric model file into the finite element simulation analysis software.

[0009] Optionally, various physical loading conditions include setting mechanical loads, heat conduction equations, and chemical reaction rates.

[0010] Optionally, based on the hybrid voxel tree model, fine meshing of the loaded probe card geometry model comprises the following steps: S221, loading the three-dimensional geometric model of the probe card into finite element simulation analysis software; S222. In the finite element simulation analysis software, the three-dimensional geometric model of the probe card is meshed using a hybrid voxel tree model, and regions where the physical field changes exceed a preset range are marked, while automatically adjusting the mesh size and distribution; S223, after completing the mesh division, checking the quality of the generated mesh and ensuring that the mesh evenly covers the entire probe card model; S224. Based on the quality inspection result of the mesh, if the mesh meets the requirements, the mesh data is exported; if the mesh does not meet the requirements, the process returns to step S222 and the mesh is optimized.

[0011] Optionally, in the finite element simulation analysis software, the three-dimensional geometric model of the probe card is meshed using a hybrid voxel tree model, and regions where the physical field changes exceed a preset range are marked, and the mesh size and distribution are automatically adjusted, including the following steps: S2221. Setting parameters of the hybrid voxel tree model in finite element simulation analysis software; S2222, converting the hybrid voxel tree model into a polyhedron boundary model through Boolean set operations; S2223, meshing the three-dimensional geometric model of the probe card using a polyhedron boundary model, and performing Boolean operations on the meshes after the meshing in the polyhedron boundary model, automatically detecting areas where the physical field changes exceed a preset threshold, and marking the areas that exceed the preset threshold; S2224. Evaluate the divided grid, identify the area of ​​the grid, and automatically adjust the size and distribution of the grid based on the evaluation results.

[0012] Optionally, meshing the three-dimensional geometric model of the probe card using a polyhedron boundary model, performing Boolean operations on the meshed grids in the polyhedron boundary model, automatically detecting areas where the physical field changes exceed a preset threshold, and marking the areas that exceed the preset threshold include the following steps: S22231. Divide the three-dimensional geometric model of the probe card using a polyhedron boundary model, and set the size and density of the grid; S22232. Perform Boolean operations on the divided meshes to merge, cut or reconstruct the meshes to meet the geometric and physical requirements of the model; S22233. In the polyhedral boundary model, automatically detect the changes in the physical field in each grid cell and compare the difference between the physical field value of each grid cell and the surrounding grid or the global average value; S22234. If the physical field change in the current area is detected to exceed a preset threshold, the current area is marked.

[0013] Optionally, the loading conditions include mechanical load, heat conduction equation and chemical reaction rate equation; The calculation formula of mechanical load is: ; The heat conduction equation is calculated as: ; The chemical reaction rate equation is calculated as: ; In the formula, Indicates mechanical load; represents Young's modulus; Indicates displacement; Indicates location; represents the thermal diffusivity; Indicates the heat source per unit volume; Indicates density; represents specific heat capacity; Indicates temperature; t Indicates time; and Indicates the concentration of the product; represents the concentration of reactants; k represents the reaction rate constant; m and n Indicates the chemical reaction order of the reactants; represents partial differential; d Represents a small increment of a variable; represents the Laplace operator.

[0014] Optionally, providing a visualization interface of the physical field distribution and the structural stress state during the simulation calculation process, and iteratively optimizing the performance of the probe card during the simulation process according to the reaction force distribution result includes the following steps: S31. Use the visualization tools provided in the finite element simulation analysis software to display the physical field distribution and structural stress state; S32. Using a performance optimization reinforcement learning algorithm, by setting a reward mechanism, the performance of the probe card is evaluated based on the distribution of stress and reaction force; S33, adjusting the probe card design parameters according to the results output by the performance optimization reinforcement learning algorithm, and re-performing the simulation to evaluate the performance change after the adjustment; S34. Monitor and analyze the frequency and efficiency of action selection in the performance optimization reinforcement learning algorithm, and adjust the exploration parameters appropriately based on real-time feedback until the preset performance standards are met or the upper limit of the number of simulations is reached.

[0015] Optionally, using a performance optimization reinforcement learning algorithm, by setting a reward mechanism, evaluating the performance of the probe card based on the distribution of stress and reaction force includes the following steps: S321, determining a reward function according to an optimization target of the probe card performance; S322, setting a learning rate of a performance optimization reinforcement learning algorithm, and analyzing the distribution of stress and reaction force to determine the load-bearing capacity and possible failure points of the probe card structure in actual use; S323, based on the judgment result, initializing a Q value table, wherein each state and action in the Q value table corresponds to an initially estimated Q value; S324, integrate the Q learning algorithm with the finite element simulation software to ensure that in each simulation iteration, actions can be selected based on the current state and feedback can be obtained from the simulation results; S325, selecting an execution action according to the current Q value table, and performing simulation; S326, calculating the reward value according to the simulation results and the reward mechanism, and updating the Q value table using the Q learning update rule; S327, executing new action selection and simulation according to the updated Q value table, and repeating this process until the performance standard is reached or the set number of iterations is completed; The calculation formula of the reward value is: ; In the formula, x Indicates the current state; y Indicates the action taken; R ( s , a ) indicates the reward received; s ′ represents the new state after the action; a ' indicates the action in the new state; Indicates that in a given state x Take action y Expected total return; represents the learning rate; represents the discount factor; Indicates new status x Take action y Expected total return; Indicates that in a specific state x Take action y The expected total return.

[0016] According to another aspect of the present invention, there is also provided a probe card simulation system based on reaction force analysis, the system comprising: a geometric model building and integration module, a meshing and simulation setting module and a simulation result performance optimization module; The geometric model building and integration module, the meshing and simulation setting module, and the simulation result performance optimization module are connected in sequence; The geometric model building and integration module is used to integrate multiple physical field conditions to build the probe card geometric model and import the probe card geometric model into the finite element simulation analysis software; A meshing and simulation setting module is used to mesh the probe card geometry model using a hybrid voxel tree model, and adjust the mesh density according to the distribution requirements of the multi-physics field, and set the boundary conditions of the simulation using a preset boundary condition template according to the actual simulation requirements; and start the simulation calculation process based on the adjusted mesh density and the boundary conditions of the simulation; The simulation result performance optimization module is used to provide a visual interface for the physical field distribution and structural stress state during the simulation calculation process, and iteratively optimize the performance of the probe card during the simulation process based on the reaction force distribution results.

[0017] Compared with the prior art, this application has the following beneficial effects: 1. The present invention integrates multiple physical field requirements to design the probe card, making the simulation closer to the actual usage environment, and can simulate and analyze the behavior of the probe card under various extreme conditions, thereby improving the practicality and reliability of the design. Finite element analysis software is used to analyze the structural strength, thermal stability, and electromagnetic compatibility of the probe card, helping engineers to comprehensively evaluate all aspects of product design performance, discover problems through simulation analysis and solve them, which can significantly reduce production risks and costs.

[0018] 2. The hybrid voxel tree model in the present invention can provide efficient meshing in complex geometry and multi-physics environment. By automatically adjusting the mesh density to adapt to the changes in the physical field, higher simulation accuracy is ensured in key areas such as high stress or high temperature areas. The adaptive meshing strategy allows the use of coarser meshes in areas where the physical field does not change much, and fine meshes in key areas. This not only reduces the consumption of computing resources, but also speeds up the simulation process, making the entire simulation more efficient. By marking the areas where the physical field changes exceed the preset threshold, engineers can easily identify areas where problems may exist. Boolean operations provide powerful tools in the meshing process, allowing engineers to adapt to complex model geometries by merging, cutting or reconstructing meshes. By performing a comprehensive mesh quality check before simulation, it is ensured that all meshes meet the requirements of numerical analysis. After each simulation, the model is adjusted and optimized according to the results, so that the design gradually approaches the expected goal. Multi-physics simulation allows multiple loading conditions to be considered simultaneously, such as thermal, structural, electromagnetic, etc., providing a comprehensive evaluation of the comprehensive performance of the probe card.

[0019] 3. Through visualization tools, the present invention allows engineers to intuitively observe the performance of the probe card under different physical field conditions, including temperature distribution, stress state, etc. This visualization support helps engineers better understand complex data and make more accurate decisions. The performance optimization reinforcement learning algorithm is used to continuously learn and adjust during the simulation process. By continuously iteratively updating the Q value table, the algorithm can optimize the design of the probe card to adapt to various test conditions and expected performance. It adopts a piecewise asymptotic greedy strategy to effectively balance exploration and utilization, avoid the problem of local optimal solutions, and allow the algorithm to search for global optimal solutions more effectively, bringing higher adaptability and optimization potential to the probe card design. Through reward mechanisms and performance evaluation, the reinforcement learning algorithm can be based on the set performance goals. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above characteristics, features and advantages of the present invention and their implementation methods and methods will become more clearly understood in conjunction with the following description of the embodiments, which are described in detail in conjunction with the accompanying drawings. Herein, a schematic diagram is shown: Figure 1 is a flow chart of a probe card simulation method based on reaction force analysis according to an embodiment of the present invention; Figure 2 It is a principle block diagram of a probe card simulation system based on reaction force analysis according to an embodiment of the present invention.

[0021] In the figure: 1. Geometric model construction and integration module; 2. Meshing and simulation setting module; 3. Simulation result performance optimization module. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] According to an embodiment of the present invention, a probe card simulation method and system based on reaction force analysis are provided.

[0024] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, a probe card simulation method based on reaction force analysis is provided, and the probe card simulation method comprises the following steps: S1. Integrate multiple physical field conditions to build a probe card geometry model, and import the probe card geometry model into the finite element simulation analysis software.

[0025] Preferably, integrating multiple physical field conditions to construct a probe card geometric model, and importing the probe card geometric model into a finite element simulation analysis software comprises the following steps: S11. Obtain the functional requirements of the multi-physics field in the probe card, and use mapping software to construct a probe card geometric model according to the functional requirements of the multi-physics field; wherein the multi-physics field functional requirements usually involve different operating environments that the probe card needs to adapt to, such as temperature changes, electromagnetic field influences, mechanical stress, etc. The functional requirements define the performance standards and stability that the probe card should maintain under these environments; S12. Assign corresponding material properties and performance parameters to each part of the probe card geometry model, and configure various physical loading conditions encountered by the probe card in actual operation; material properties and performance parameters include elastic modulus, density, thermal expansion coefficient, conductivity, etc. of the material; material properties directly affect the behavior and stability of the probe card in various physical fields; physical loading conditions include temperature range (such as -40°C to 85°C), mechanical stress (such as pressure, tension, bending), and voltage load (such as voltage peak during signal transmission); S13, exporting the configured geometric model and converting it into a file in a format that matches the finite element simulation analysis software; S14. Load the probe card geometric model file into the finite element simulation analysis software.

[0026] Preferably, various physical loading conditions include setting mechanical load, heat conduction equation and chemical reaction rate.

[0027] It should be explained that the use conditions and performance indicators of the probe card are collected and analyzed, and then the geometric model of the probe card is designed according to these requirements using CAD or other drawing software to ensure that the design meets all expected functional requirements. The correct material properties are assigned to each part in the geometric model of the probe card, and the physical loading conditions that may be encountered in the expected operation are set. The physical loading conditions will be used in the simulation analysis to verify the performance of the probe card. The probe card model created in the drawing software is exported to a format suitable for finite element analysis, such as STEP or IGES files, to ensure that the model can be correctly loaded and analyzed in the finite element software. Finally, the exported file is loaded into the finite element analysis software (such as ANSYS, ABAQUS, etc.), and further settings are made in the software, such as defining the mesh size, selecting the analysis type (static, dynamic, thermal analysis, etc.), and preparing for detailed simulation analysis.

[0028] S2. Use the hybrid voxel tree model to mesh the probe card geometry model, adjust the mesh density according to the distribution requirements of the multi-physical fields, and set the simulation boundary conditions using the preset boundary condition template according to the actual simulation requirements; start the simulation calculation process based on the adjusted mesh density and simulation boundary conditions.

[0029] The hybrid voxel tree model is used to mesh the probe card geometry model, and the mesh density is adjusted according to the distribution requirements of the multi-physical field. At the same time, the boundary conditions of the simulation are set using the preset boundary condition template according to the actual simulation requirements. The simulation calculation process is started based on the adjusted mesh density and the simulation boundary conditions, including the following steps: S21. Obtain material properties and performance parameters of each part in the probe card geometric model.

[0030] S22. Based on the hybrid voxel tree model, mesh the loaded probe card geometry model.

[0031] Preferably, based on the hybrid voxel tree model, detailed meshing of the loaded probe card geometric model comprises the following steps: S221, loading the three-dimensional geometric model of the probe card into finite element simulation analysis software; S222. In the finite element simulation analysis software, the three-dimensional geometric model of the probe card is meshed using a hybrid voxel tree model, and the areas where the physical field changes exceed the preset range value are marked, and the mesh size and distribution are automatically adjusted.

[0032] Preferably, in the finite element simulation analysis software, the three-dimensional geometric model of the probe card is meshed using a hybrid voxel tree model, and the area where the physical field changes exceed the preset range value is marked, and the mesh size and distribution are automatically adjusted, including the following steps: S2221. Setting parameters of the hybrid voxel tree model in finite element simulation analysis software; S2222, converting the hybrid voxel tree model into a polyhedron boundary model through Boolean set operations; S2223, meshing the three-dimensional geometric model of the probe card using a polyhedron boundary model, and performing Boolean operations on the meshes after the meshing in the polyhedron boundary model, automatically detecting areas where the physical field changes exceed a preset threshold, and marking the areas that exceed the preset threshold; S2224. Evaluate the divided grid, identify the area of ​​the grid, and automatically adjust the size and distribution of the grid based on the evaluation results.

[0033] It should be explained that data are obtained through literature research, experimental data or communication with material suppliers, and these data such as thermal conductivity, elastic modulus, resistivity, etc. of the material are recorded in the database for calling in the simulation software. The hybrid voxel tree model method is used to achieve adaptive meshing, and the size and distribution of the mesh are dynamically adjusted according to the geometric complexity of the model and the requirements of the physical field, so as to improve the accuracy and efficiency of the calculation. CAD software or specific conversion tools are used to convert the geometric model into a format supported by the simulation software, and then these files are loaded in the simulation software.

[0034] According to the specific application scenarios and simulation requirements of the probe card, appropriate voxel size, depth and splitting criteria are set to ensure a balance between the model's detail and computational efficiency. Boolean operation technology is used to convert the voxel model into a polyhedral mesh model that can be used for finite element analysis, merge small voxels, delete invalid voxels, and so on. After the meshing is completed, according to the results of the physical field analysis, the areas where the physical field changes exceed the predetermined threshold are automatically detected and marked, and the mesh areas that require special attention are identified. According to the mesh quality assessment results, the mesh is optimized and adjusted, such as adjusting the mesh density and improving the mesh shape, to ensure the accuracy and efficiency of the simulation.

[0035] The hybrid voxel tree model is an advanced technology for meshing and geometric modeling, especially suitable for complex geometry and multi-physics simulation environments. This model combines the characteristics of voxels and octrees, providing an effective way to process and simplify the representation of 3D geometric data while ensuring the accuracy and efficiency of the simulation.

[0036] The key features of the hybrid voxel tree model are as follows: 1) Voxel representation: A voxel is a small cube in three-dimensional space. Each voxel stores physical properties about its location (such as material type, density, thermal conductivity, etc.). The hybrid voxel tree simplifies model processing by decomposing complex three-dimensional models into voxel grids; 2) Octree structure: Octree is a tree-shaped data structure used for three-dimensional space division. Each node represents a spatial region, which can be further subdivided into eight sub-regions. In the hybrid voxel tree model, the octree is used to dynamically manage the distribution of voxels and adjust the size and distribution of voxels according to the complexity of geometric or physical requirements. 3) Adaptive meshing: The model automatically adjusts the mesh density based on changes in the physical field or the complexity of the geometry, using finer meshes for areas with high details or high changes, and coarser meshes in areas with less changes.

[0037] Boolean operations are a technique for combining or modifying geometric entities, commonly used in computer graphics, CAD (computer-aided design), and geometric modeling. These operations can be performed between two or more geometric entities and include operations such as union, intersection, and difference, which are used to create complex shapes and models. The following are the three basic types of Boolean operations: 1) Union: Operation: Merge two geometries into a new geometry that contains all the space of the two original geometries.

[0038] Application: Used to merge two independent parts or shapes, commonly used in assembly or model integration.

[0039] 2) Intersection: Operation: Create a new geometry that has the common parts of the two geometries.

[0040] Application: Used to find the area occupied by two shapes. Often used in the analysis of overlapping and matching parts in design.

[0041] 3) Difference: Operation: Cut out the part of one geometry that overlaps with another geometry.

[0042] Application: Used to remove parts of a complex block, such as openings, cuts or shape trimming.

[0043] Preferably, the three-dimensional geometric model of the probe card is meshed using a polyhedron boundary model, and Boolean operations are performed on the meshes after meshing in the polyhedron boundary model to automatically detect areas where the physical field changes exceed a preset threshold, and mark the areas that exceed the preset threshold, including the following steps: S22231. Divide the three-dimensional geometric model of the probe card using a polyhedron boundary model, and set the size and density of the grid; S22232. Perform Boolean operations on the divided meshes to merge, cut or reconstruct the meshes to meet the geometric and physical requirements of the model; S22233. In the polyhedral boundary model, automatically detect the changes in the physical field in each grid cell and compare the difference between the physical field value of each grid cell and the surrounding grid or the global average value; S22234. If the physical field change in the current area is detected to exceed a preset threshold, the current area is marked.

[0044] S223, after completing the mesh division, checking the quality of the generated mesh and ensuring that the mesh evenly covers the entire probe card model; S224. Based on the quality inspection result of the mesh, if the mesh meets the requirements, the mesh data is exported; if the mesh does not meet the requirements, the process returns to step S222 and the mesh is optimized.

[0045] Preferably, the loading conditions include mechanical load, heat conduction equation and chemical reaction rate equation; The calculation formula of mechanical load is: ; The heat conduction equation is calculated as: ; The chemical reaction rate equation is calculated as: ; In the formula, Indicates mechanical load; represents Young's modulus; Indicates displacement; Indicates location; represents the thermal diffusivity; Indicates the heat source per unit volume; Indicates density; represents specific heat capacity; Indicates temperature; t Indicates time; and Indicates the concentration of the product; represents the concentration of reactants; k represents the reaction rate constant; m and n Indicates the chemical reaction order of the reactants; represents partial differential; d Represents a small increment of a variable; represents the Laplace operator.

[0046] It should be explained that the geometric model of the probe card is converted into a polyhedral boundary model. This process is usually completed in finite element analysis software. The size and density of the grid are set by selecting an appropriate resolution to ensure that every detail can be fully captured. High-precision areas, such as high stress concentration or high temperature gradient areas, will be set with a denser grid. Boolean operations are used to optimize the grid structure, including merging adjacent grid cells with similar characteristics, or cutting the grid to adapt to complex geometric structures, which helps to reduce the total number of grids while maintaining the level of detail required for analysis. Physical field analysis is performed through the set analysis software to automatically identify areas where physical field changes exceed preset thresholds, including changes in physical quantities such as temperature, stress, and electromagnetic fields. Once these areas are identified, they are immediately marked. , for further analysis or re-meshing, set up the automatic annotation function in the software, and immediately visually annotate these areas when physical field changes exceeding the threshold are detected. Visual annotation helps engineers identify areas that may require special attention or design improvements. Use the mesh quality detection tool of the finite element software to check the uniformity of the mesh, the shape quality and size distribution of the mesh units to ensure that all meshes meet the standards of numerical analysis and there are no mesh units that are too distorted or extremely uneven in size. If the mesh inspection results show that the mesh quality does not meet expectations, it is necessary to adjust the meshing parameters or re-run the Boolean operation. If the mesh meets the requirements, the mesh data can be exported to a format suitable for further simulation analysis, such as a file format supported by FEA or CFD software.

[0047] S23. In the finite element simulation analysis software, set the simulation boundary conditions according to the working environment and actual use conditions of the probe card.

[0048] S24. Using the established hybrid voxel tree model and the set boundary conditions, start multi-physics field simulation to calculate the physical response and chemical response of the probe card under the set loading conditions.

[0049] S25. Analyze the performance of the probe card based on the simulation calculation results to determine whether it meets the predetermined performance standards and safety requirements; and adjust the geometric structure, material properties and performance parameters of the probe card based on the judgment results.

[0050] S26. If the geometry, material properties and performance parameters of the probe card do not meet the preset targets, repeat steps S21-S25, and adjust the meshing, boundary conditions and physical loading conditions according to the preset targets. After each iteration, re-evaluate the performance of the probe card during the simulation. If the geometry, material properties and performance parameters of the probe card do not meet the preset targets, output the geometry, material properties and performance parameters of the probe card.

[0051] It needs to be explained that the working environment of the probe card is defined, including temperature range, voltage or current level, mechanical pressure, etc., and these conditions are set as boundary conditions in the simulation software. For example, if the probe card works at high temperature, the corresponding temperature boundary conditions should be set to ensure that the simulation reflects the actual operating environment. Based on the set boundary conditions and the prepared mesh model, the simulation analysis is started, and the multi-physics simulation module (such as ANSYS Multiphysics) is used to calculate the stress, temperature distribution, electromagnetic effect, etc. of the probe card under various loading conditions. Through these calculations, the physical and chemical stability of the probe card is evaluated, and the simulation results are analyzed, especially focusing on those areas that exceed the safety threshold or potential failure points. According to the analysis results, the design of the probe card is adjusted, such as changing the material to improve heat resistance or adjusting the geometry to reduce stress concentration, to ensure that the probe card meets all performance and safety standards. Simulation should not be a one-time process. Depending on the results of each simulation, it may be necessary to go back to the previous step to adjust the meshing of the model, update the boundary conditions, or optimize the loading conditions. This iterative process helps to gradually improve the design of the probe card until all predetermined performance goals are met.

[0052] In the finite element simulation analysis software, setting the boundary conditions of the simulation according to the working environment and actual use conditions of the probe card includes the following steps: Setting the boundary conditions of the probe card in the finite element simulation analysis software usually involves defining various parameters of its operating environment, such as temperature range, mechanical load (such as pressure or impact), voltage and current levels, and configuring them within the software by specifying corresponding parameter values ​​and properties to ensure that the simulation environment accurately reflects the working state of the probe card in actual use; the key to this process is to accurately input the environment and operating parameters to simulate the behavior of the probe card under various physical influences, so as to predict its performance and identify potential structural problems.

[0053] Using the established hybrid voxel tree model and the set boundary conditions, launching a multiphysics simulation to calculate the physical and chemical responses of the probe card under the set loading conditions includes the following steps: Load the 3D geometry model of the probe card in the finite element simulation software. The model has been meshed using a hybrid voxel tree model to ensure that the distribution of the mesh is adapted to various physical and chemical loading conditions. According to the working environment and actual use conditions of the probe card, set corresponding boundary conditions, including temperature boundary, electromagnetic field strength, mechanical load, etc. These conditions directly affect the accuracy of the simulation and the reliability of the results; Depending on the physical and chemical responses that need to be evaluated, select the appropriate simulation type, such as thermal analysis, structural analysis, electromagnetic analysis, or a combination of them to simulate a multiphysics environment; After all necessary conditions and parameters are set, start the simulation process. The software will calculate the response of the probe card under various loading conditions according to the set model and boundary conditions, including temperature changes, stress distribution, electromagnetic effects and other related physical and chemical responses; Once the simulation is complete, analyze and evaluate the physical and chemical response of the probe card, using the tools provided by the software to visualize the simulation results and identify any unexpected behavior or potential failure points.

[0054] S3. Provide a visual interface for the physical field distribution and structural stress state during the simulation calculation process, and iteratively optimize the performance of the probe card during the simulation process based on the reaction force distribution results.

[0055] Preferably, providing a visual interface of the physical field distribution and the structural stress state during the simulation calculation process, and iteratively optimizing the performance of the probe card during the simulation process according to the reaction force distribution result includes the following steps: S31. Use the visualization tools provided in the finite element simulation analysis software to display the physical field distribution and structural stress state; S32. Using the performance optimization reinforcement learning algorithm, the performance of the probe card is evaluated based on the distribution of stress and reaction force by setting up a reward mechanism.

[0056] Preferably, using a performance optimization reinforcement learning algorithm, by setting a reward mechanism, evaluating the performance of the probe card based on the distribution of stress and reaction force includes the following steps: S321, determining a reward function according to an optimization target of the probe card performance; S322, setting a learning rate of a performance optimization reinforcement learning algorithm, and analyzing the distribution of stress and reaction force to determine the load-bearing capacity and possible failure points of the probe card structure in actual use; S323, based on the judgment result, initializing a Q value table, wherein each state and action in the Q value table corresponds to an initially estimated Q value; S324, integrate the Q learning algorithm with the finite element simulation software to ensure that in each simulation iteration, actions can be selected based on the current state and feedback can be obtained from the simulation results; S325, selecting an execution action according to the current Q value table, and performing simulation; S326, calculating the reward value according to the simulation results and the reward mechanism, and updating the Q value table using the Q learning update rule; S327, executing new action selection and simulation according to the updated Q value table, and repeating this process until the performance standard is reached or the set number of iterations is completed; The calculation formula of the reward value is: ; In the formula, x Indicates the current state; y Indicates the action taken; R ( s , a ) indicates the reward received; s ′ represents the new state after the action; a ' indicates the action in the new state; Indicates that in a given state x Take action y Expected total return; represents the learning rate; represents the discount factor; Indicates new status x Take action y Expected total return; Indicates that in a specific state x Take action y The expected total return.

[0057] It should be explained that after the simulation is completed, the built-in visualization tools of the simulation software are used to display the physical field (such as temperature, electromagnetic field) distribution and structural stress state of the probe card. These tools usually provide various graphical display options, such as contour maps, heat maps and vector field maps, so that every detail of the simulation results can be viewed and analyzed in detail. The reward function is set to reflect the goal of optimizing the performance of the probe card. For example, if the goal is to minimize stress concentration, the reward function can be set to increase the reward value when the stress is below a certain threshold. The key to choosing the reward function is to ensure that it correctly reflects the design goals and the expected performance improvement. An appropriate learning rate is selected to control the speed and stability of the learning process. The learning rate determines the extent to which new information covers old information. A moderate learning rate helps the algorithm converge effectively and find the optimal strategy. Initialize the Q value table for each value in the simulation. The possible state and action combinations provide a starting estimate. These initial values ​​are usually set to zero or preset based on prior knowledge. The reinforcement learning algorithm is interfaced with the simulation software, allowing the algorithm to execute decisions in the simulation environment (such as modifying boundary conditions or grid configurations) and receive feedback based on the simulation results to update the Q value. The next action is selected based on the optimal strategy in the Q value table, and then the action is performed in the simulation environment, such as adjusting load conditions or changing material properties. After the simulation is completed, the reward value is calculated based on the set reward function, and then the reward value is used to update the Q value table according to the Q learning formula to improve future decisions. The simulation, evaluation, and learning process is repeated. Each iteration selects an action based on the updated Q value table to continuously optimize the design of the probe card until the performance reaches the predetermined standard or the set number of iterations is reached.

[0058] The performance-optimized reinforcement learning algorithm is an improved Q-learning algorithm. The piecewise asymptotic greedy strategy adopted by the improved Q-learning algorithm aims to solve the problem that the traditional greedy strategy may fall into the local optimal solution. This strategy dynamically adjusts the balance between exploration and utilization to enable the algorithm to search for the global optimal solution more effectively. The following are the detailed steps of the improved greedy strategy: It mainly includes initialization parameter setting, initial exploration through greedy strategy, re-evaluation of action probability to adapt to the learning stage, and dynamic adjustment of strategy and exploration intensity (H). In the initial stage, the algorithm encourages exploration through a higher exploration probability (ϵ), and then gradually increases the utilization of the optimal action, thereby balancing exploration and utilization and optimizing performance. By comparing the efficiency of continuous iterations and adjusting parameters, the algorithm continues to optimize until the upper limit of the number of learning times is reached or other termination conditions are met, and finally the optimal strategy and Q value are determined. This method is suitable for complex decision-making environments that need to deal with uncertainty and dynamic changes.

[0059] S33. Adjust the probe card design parameters according to the results output by the performance optimization reinforcement learning algorithm, and re-execute the simulation to evaluate the performance changes after the adjustment.

[0060] S34. Monitor and analyze the frequency and efficiency of action selection in the performance optimization reinforcement learning algorithm, and adjust the exploration parameters appropriately based on real-time feedback until the preset performance standards are met or the upper limit of the number of simulations is reached.

[0061] It should be explained that the results of the performance optimization reinforcement learning algorithm output are first analyzed, with special attention paid to which parameter adjustments lead to performance improvement or reduction. According to the analysis results, the design parameters of the probe card, such as geometric structure, material properties or loading conditions, are adjusted. The simulation is re-executed on the adjusted model to evaluate the specific impact of these changes on the performance of the probe card. The effectiveness of the design adjustment is verified by comparing the simulation results before and after the adjustment. The simulation results are analyzed to determine whether the predetermined performance standards are met or whether further adjustments are required. During the reinforcement learning process, the selection frequency and efficiency of each action are monitored in real time, including tracking which actions lead to performance improvements and which may be suboptimal choices, and analyzing the specific contribution of each action to performance to determine the most effective strategy. Use this information to adjust the action selection strategy, optimize the balance between exploration and utilization, and appropriately adjust the exploration parameters (such as exploration intensity H or exploration probability ϵ) based on the analysis results of action selection to more effectively guide the optimal design solution. Continue to perform simulations and adjust the algorithm based on real-time feedback until the simulation results show that the probe card design meets the preset performance standards or reaches the upper limit of the number of simulations.

[0062] A specific example is as follows: An electronic equipment manufacturing company wants to develop a high-performance probe card that can work stably under a variety of environmental conditions. This probe card needs to be able to withstand temperature fluctuations, strong electromagnetic field interference, and varying degrees of mechanical stress.

[0063] 1) Collect requirements and prepare for design: Functional requirements capture: The team first determines the conditions that the probe card must meet, such as temperature range (-40°C to 85°C), electromagnetic field effects (anti-interference capabilities are required), and mechanical stresses (such as pressure and tension).

[0064] Design geometry: Use CAD software to design the preliminary geometry of the probe card based on these multi-physics requirements.

[0065] 2) Material selection and model configuration: Material property configuration: Select a suitable material such as high-performance plastic or metal alloy, and configure the material properties (e.g. elastic modulus, conductivity).

[0066] Multi-physics simulation: Import the designed probe card model into simulation software to perform thermal analysis, electromagnetic compatibility analysis, and structural strength analysis to evaluate the response under different physical loading conditions.

[0067] 3) Performance Optimization Reinforcement Learning: Apply reinforcement learning algorithms: Based on preliminary simulation results, use reinforcement learning algorithms to optimize design parameters, such as adjusting probe placement or material thickness, to improve performance and durability.

[0068] Parameter Tuning and Re-Simulation: Adjust key design parameters based on feedback provided by the algorithm and re-simulate to verify performance improvements.

[0069] 4) Prototype manufacturing and testing: Prototype Fabrication: Probe card prototypes are fabricated based on the final optimized design.

[0070] Real-world testing: Testing the performance of the prototype in a controlled environment, such as through thermal cycling, mechanical strength, and electromagnetic compatibility testing, to ensure that it meets all pre-set standards.

[0071] According to another embodiment of the present invention, Figure 2 As shown, a probe card simulation system based on reaction force analysis is also provided, the system comprising: a geometric model building and integration module 1, a meshing and simulation setting module 2 and a simulation result performance optimization module 3; The geometric model building and integration module 1, the meshing and simulation setting module 2, and the simulation result performance optimization module 3 are connected in sequence; The geometric model building and integration module 1 is used to integrate multiple physical field conditions to build a probe card geometric model and import the probe card geometric model into the finite element simulation analysis software; The meshing and simulation setting module 2 is used to mesh the probe card geometry model using a hybrid voxel tree model, and adjust the mesh density according to the distribution requirements of the multi-physical field, and set the boundary conditions of the simulation using a preset boundary condition template according to the actual simulation requirements; and start the simulation calculation process based on the adjusted mesh density and the boundary conditions of the simulation; The simulation result performance optimization module 3 is used to provide a visual interface for the physical field distribution and structural stress state during the simulation calculation process, and iteratively optimize the performance of the probe card during the simulation process based on the reaction force distribution results.

[0072] In summary, with the help of the above-mentioned technical scheme of the present invention, the present invention designs the probe card by integrating multi-physical field requirements, so that the simulation is closer to the actual use environment, and can simulate and analyze the behavior of the probe card under various extreme conditions, improve the practicality and reliability of the design, and use finite element analysis software to analyze the structural strength, thermal stability and electromagnetic compatibility of the probe card, etc., to help engineers comprehensively evaluate all aspects of product design performance, and find and solve problems through simulation analysis, which can significantly reduce production risks and costs; the hybrid voxel tree model in the present invention can provide efficient meshing in complex geometry and multi-physical field environments, and automatically adjust the mesh density to adapt to changes in the physical field to ensure higher simulation accuracy in key areas such as high stress or high temperature areas. The adaptive meshing strategy allows the use of coarser meshes in areas where the physical field changes little, and the use of fine meshes in key areas, which not only reduces the consumption of computing resources, but also speeds up the simulation process, making the entire simulation more efficient. By marking the areas where the physical field changes exceed the preset threshold, engineers can easily identify areas where problems may exist. Boolean operations provide powerful tools in the meshing process, allowing Engineers can adapt to complex model geometry by merging, cutting or reconstructing grids, and perform a comprehensive grid quality check before simulation to ensure that all grids meet the requirements of numerical analysis. After each simulation, the model is adjusted and optimized according to the results to gradually approach the expected goal. Multi-physics field simulation allows multiple loading conditions, such as thermal, structural, electromagnetic, etc., to be considered simultaneously, providing a comprehensive evaluation of the comprehensive performance of the probe card. Through visualization tools, engineers can intuitively observe the performance of the probe card under different physical field conditions, including temperature distribution, stress state, etc. This visualization support helps engineers better understand complex data and make more accurate decisions. The performance optimization reinforcement learning algorithm is used to continuously learn and adjust during the simulation process. By continuously iteratively updating the Q value table, the algorithm can optimize the design of the probe card to adapt to various test conditions and expected performance. A piecewise asymptotic greedy strategy is adopted to effectively balance exploration and utilization, avoid the problem of local optimal solutions, and allow the algorithm to more effectively search for global optimal solutions, bringing higher adaptability and optimization potential to the probe card design. Through reward mechanisms and performance evaluation, the reinforcement learning algorithm can be based on the set performance goals.

[0073] Although the present invention has been disclosed as above with preferred embodiments, the embodiments are merely examples for the convenience of description and are not intended to limit the present invention. Those skilled in the art may make several changes and modifications without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention shall be based on the claims.

Claims

1. A probe card simulation method based on reaction force analysis, characterized in that: The probe card simulation method comprises the following steps: S1. Integrate multiple physical field conditions to build a probe card geometry model, and import the probe card geometry model into finite element simulation analysis software; S2. Meshing the probe card geometry model using a hybrid voxel tree model, adjusting the mesh density according to the distribution requirements of the multi-physical fields, and setting the boundary conditions of the simulation using a preset boundary condition template according to the actual simulation requirements; starting the simulation calculation process based on the adjusted mesh density and the boundary conditions of the simulation; S3. Provide a visual interface for physical field distribution and structural stress state during simulation calculation, and iteratively optimize the performance of the probe card during simulation based on the reaction force distribution results; The method of using a hybrid voxel tree model to mesh the probe card geometry model, adjusting the mesh density according to the distribution requirements of the multi-physical field, and setting the boundary conditions of the simulation using a preset boundary condition template according to the actual simulation requirements; starting the simulation calculation process based on the adjusted mesh density and the boundary conditions of the simulation includes the following steps: S21, obtaining material properties and performance parameters of each part in the probe card geometric model; S22, meshing the loaded probe card geometry model based on the hybrid voxel tree model; S23. In the finite element simulation analysis software, setting the simulation boundary conditions according to the working environment and actual use conditions of the probe card; S24, using the established hybrid voxel tree model and the set boundary conditions, starting the multi-physics simulation to calculate the physical response and chemical response of the probe card under the set loading conditions; S25, analyzing the performance of the probe card according to the simulation calculation results to determine whether it meets the predetermined performance standards and safety requirements; and adjusting the geometric structure, material properties and performance parameters of the probe card according to the determination results; S26. If the geometry, material properties and performance parameters of the probe card do not meet the preset targets, repeat steps S21-S25, and adjust the meshing, boundary conditions and physical loading conditions according to the preset targets. After each iteration, re-evaluate the performance of the probe card during the simulation. If the geometry, material properties and performance parameters of the probe card do not meet the preset targets, output the geometry, material properties and performance parameters of the probe card.

2. A probe card simulation method based on reaction force analysis according to claim 1, characterized in that: The method of integrating multiple physical field conditions to construct a probe card geometric model and importing the probe card geometric model into the finite element simulation analysis software includes the following steps: S11, obtaining functional requirements of multi-physics fields in the probe card, and using mapping software to construct a probe card geometric model according to the functional requirements of the multi-physics fields; S12, assigning corresponding material properties and performance parameters to various parts of the probe card geometric model, and configuring various physical loading conditions encountered by the probe card in actual operation; S13, exporting the configured geometric model and converting it into a file in a format that matches the finite element simulation analysis software; S14. Load the probe card geometric model file into the finite element simulation analysis software.

3. The probe card simulation method based on reaction force analysis according to claim 2, characterized in that: The various physical loading conditions include setting mechanical load, heat conduction equation and chemical reaction rate.

4. The probe card simulation method based on reaction force analysis according to claim 1, characterized in that: The detailed meshing of the loaded probe card geometric model based on the hybrid voxel tree model includes the following steps: S221, loading the three-dimensional geometric model of the probe card into finite element simulation analysis software; S222. In the finite element simulation analysis software, the three-dimensional geometric model of the probe card is meshed using a hybrid voxel tree model, and regions where the physical field changes exceed a preset range are marked, while automatically adjusting the mesh size and distribution; S223, after completing the mesh division, checking the quality of the generated mesh and ensuring that the mesh evenly covers the entire probe card model; S224. Based on the quality inspection result of the mesh, if the mesh meets the requirements, the mesh data is exported; if the mesh does not meet the requirements, the process returns to step S222 and the mesh is optimized.

5. The probe card simulation method based on reaction force analysis according to claim 4, characterized in that: The method of meshing the three-dimensional geometric model of the probe card using a hybrid voxel tree model in the finite element simulation analysis software, marking the area where the physical field changes exceed the preset range value, and automatically adjusting the mesh size and distribution includes the following steps: S2221. Setting parameters of the hybrid voxel tree model in finite element simulation analysis software; S2222, converting the hybrid voxel tree model into a polyhedron boundary model through Boolean set operations; S2223, meshing the three-dimensional geometric model of the probe card using a polyhedron boundary model, and performing Boolean operations on the meshes after the meshing in the polyhedron boundary model, automatically detecting areas where the physical field changes exceed a preset threshold, and marking the areas that exceed the preset threshold; S2224. Evaluate the divided grid, identify the area of ​​the grid, and automatically adjust the size and distribution of the grid based on the evaluation results.

6. The probe card simulation method based on reaction force analysis according to claim 5, characterized in that: The method of meshing the three-dimensional geometric model of the probe card using the polyhedron boundary model, performing Boolean operations on the meshes after meshing in the polyhedron boundary model, automatically detecting areas where the physical field changes exceed a preset threshold, and marking the areas exceeding the preset threshold includes the following steps: S22231. Divide the three-dimensional geometric model of the probe card using a polyhedron boundary model, and set the size and density of the grid; S22232. Perform Boolean operations on the divided meshes to merge, cut or reconstruct the meshes to meet the geometric and physical requirements of the model; S22233. In the polyhedral boundary model, automatically detect the changes in the physical field in each grid cell and compare the difference between the physical field value of each grid cell and the surrounding grid or the global average value; S22234. If the physical field change in the current area is detected to exceed a preset threshold, the current area is marked.

7. The probe card simulation method based on reaction force analysis according to claim 6, characterized in that: The loading conditions include mechanical load, heat conduction equation and chemical reaction rate equation; Wherein, the calculation formula of the mechanical load is: ； The calculation formula of the heat conduction equation is: ； The calculation formula of the chemical reaction rate equation is: ； In the formula, Indicates mechanical load; represents Young's modulus; Indicates displacement; Indicates location; represents the thermal diffusivity; Indicates the heat source per unit volume; Indicates density; represents specific heat capacity; Indicates temperature; t Indicates time; and Indicates the concentration of the product; represents the concentration of reactants; k represents the reaction rate constant; m and n Indicates the chemical reaction order of the reactants; represents partial differential; d Represents a small increment of a variable; represents the Laplace operator.

8. The probe card simulation method based on reaction force analysis according to claim 1, characterized in that: The method of providing a visual interface of the physical field distribution and the structural stress state during the simulation calculation process, and iteratively optimizing the performance of the probe card during the simulation process according to the reaction force distribution result, comprises the following steps: S31. Use the visualization tools provided in the finite element simulation analysis software to display the physical field distribution and structural stress state; S32. Using a performance optimization reinforcement learning algorithm, by setting a reward mechanism, the performance of the probe card is evaluated based on the distribution of stress and reaction force; S33, adjusting the probe card design parameters according to the results output by the performance optimization reinforcement learning algorithm, and re-performing the simulation to evaluate the performance change after the adjustment; S34. Monitor and analyze the frequency and efficiency of action selection in the performance optimization reinforcement learning algorithm, and adjust the exploration parameters appropriately based on real-time feedback until the preset performance standards are met or the upper limit of the number of simulations is reached.

9. The probe card simulation method based on reaction force analysis according to claim 8, characterized in that: The method of using the performance optimization reinforcement learning algorithm to evaluate the performance of the probe card based on the distribution of stress and reaction force by setting a reward mechanism includes the following steps: S321, determining a reward function according to an optimization target of the probe card performance; S322, setting a learning rate of a performance optimization reinforcement learning algorithm, and analyzing the distribution of stress and reaction force to determine the load-bearing capacity and possible failure points of the probe card structure in actual use; S323, based on the judgment result, initializing a Q value table, wherein each state and action in the Q value table corresponds to an initially estimated Q value; S324, integrate the Q learning algorithm with the finite element simulation software to ensure that in each simulation iteration, actions can be selected based on the current state and feedback can be obtained from the simulation results; S325, selecting an execution action according to the current Q value table, and performing simulation; S326, calculating the reward value according to the simulation results and the reward mechanism, and updating the Q value table using the Q learning update rule; S327, executing new action selection and simulation according to the updated Q value table, and repeating this process until the performance standard is reached or the set number of iterations is completed; The calculation formula of the reward value is: ； In the formula, x Indicates the current state; y Indicates the action taken; R ( s , a ) indicates the reward received; s ′ represents the new state after the action; a ' indicates the action in the new state; Indicates that in a given state x Take action y Expected total return; represents the learning rate; represents the discount factor; Indicates new status x Take action y Expected total return; Indicates that in a specific state x Take action y The expected total return.

10. A probe card simulation system based on reaction force analysis, used to implement the probe card simulation method based on reaction force analysis according to any one of claims 1 to 9, characterized in that: The system includes: a geometric model building and integration module, a meshing and simulation setting module, and a simulation result performance optimization module; The geometric model building and integration module, the meshing and simulation setting module, and the simulation result performance optimization module are connected in sequence; The geometric model building and integration module is used to integrate multiple physical field conditions to build the probe card geometric model and import the probe card geometric model into the finite element simulation analysis software; A meshing and simulation setting module is used to mesh the probe card geometry model using a hybrid voxel tree model, and adjust the mesh density according to the distribution requirements of the multi-physics field, and set the boundary conditions of the simulation using a preset boundary condition template according to the actual simulation requirements; and start the simulation calculation process based on the adjusted mesh density and the boundary conditions of the simulation; The simulation result performance optimization module is used to provide a visual interface for the physical field distribution and structural stress state during the simulation calculation process, and iteratively optimize the performance of the probe card during the simulation process based on the reaction force distribution results.

Images