Prediction method and device for copper plating thickness uniformity, electronic equipment and medium

Abstract

The embodiment of the invention provides a copper plating thickness uniformity prediction method and device, electronic equipment and a medium, and relates to the technical field of display screen manufacturing. The method comprises the following steps: inputting copper plating design parameters of a circuit board to be produced into a pre-trained thickness prediction model; a simulation module in the thickness prediction model simulates the copper plating process flow of the to-be-produced circuit board based on the process design parameters and the structure design parameters to obtain the simulation thickness of each area in the to-be-produced circuit board, and inputs the simulation thickness into a correction network in the thickness prediction model; the correction network determines the corrected copper plating thickness of each area in the circuit board to be produced under the copper plating design parameters based on the copper plating design parameters and the simulation thickness, and outputs the corrected copper plating thickness; and based on the corrected copper plating thickness of each region, determining a thickness uniformity result after copper plating is performed on the circuit board to be produced by adopting the copper plating design parameters. Therefore, the copper plating thickness uniformity of the circuit board to be produced can be quickly and accurately obtained.

Description

Technical Field

[0001] This application relates to the field of display manufacturing technology, and in particular to a method, apparatus, electronic device and medium for predicting the uniformity of copper plating thickness. Background Technology

[0002] Copper plating is a crucial step in the manufacturing process of Mini LEDs (Light-emitting Diodes). In this step, an electrochemical method is used to deposit a copper layer on the surface of a micron-sized electrode to achieve electrical interconnection and heat dissipation.

[0003] The uniformity of copper plating thickness in different areas of a circuit board directly affects the product's conductivity, heat dissipation efficiency, and lifespan. However, due to the high-density structure and miniaturization of Mini LED panels, uneven current distribution and uneven mass transfer of the chemical solution can easily occur in different areas during the copper plating process due to edge effects and shielding effects. This results in differences in the copper plating layer in different areas, affecting the uniformity of the copper plating thickness.

[0004] Before producing Mini LEDs, the uniformity of the copper plating thickness of the Mini LEDs to be produced can be predicted, and the structure of the Mini LEDs can be adjusted according to the prediction results to ensure that the copper plating thickness of the produced Mini LEDs meets the uniformity requirements, thereby improving the quality of the produced Mini LEDs. Summary of the Invention

[0005] The purpose of this application is to provide a method, apparatus, electronic device, and medium for predicting the uniformity of copper plating thickness in Mini LEDs to be manufactured. The specific technical solution is as follows:

[0006] In a first aspect, embodiments of this application provide a method for predicting the uniformity of copper plating thickness, the method comprising:

[0007] The copper plating design parameters of the circuit board to be manufactured are input into a pre-trained thickness prediction model, wherein the thickness prediction model includes a simulation module and a correction network; the copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be manufactured.

[0008] The simulation module simulates the copper plating process of the pre-set circuit board to be produced based on the process design parameters and the structural design parameters, obtains the simulated thickness of each area in the circuit board to be produced, and inputs the simulated thickness into the correction network.

[0009] The correction network determines the corrected copper plating thickness of each area in the circuit board to be manufactured under the copper plating design parameters and the simulated thickness, and outputs the corrected copper plating thickness.

[0010] Based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured, the thickness uniformity result of copper plating on the circuit board to be manufactured using the copper plating design parameters is determined.

[0011] Optionally, the training method for the thickness prediction model includes:

[0012] Multiple training samples and thickness labels corresponding to each training sample are obtained. Each training sample includes copper plating design parameters of the sample circuit board. The copper plating design parameters include sample process design parameters corresponding to the copper plating process and sample structure design parameters of each region of the sample circuit board.

[0013] For each training sample, input the training sample into the initial thickness prediction model;

[0014] The simulation module in the initial thickness prediction model simulates the copper plating process of the sample circuit board based on the sample process design parameters and the sample structure design parameters, obtains the sample simulation thickness of each region in the sample circuit board, and inputs the sample simulation thickness into the initial correction network in the initial thickness prediction model.

[0015] The initial calibration network determines the corrected copper plating thickness of each region in the sample circuit board under the sample copper plating design parameters, based on the sample copper plating design parameters and the sample simulated thickness.

[0016] Based on the difference between the copper plating thickness after sample correction for each training sample and the thickness label corresponding to that training sample, the model parameters of the initial correction network are adjusted until the initial correction network meets the convergence condition, thus obtaining the trained thickness prediction model.

[0017] Optionally, the step of obtaining multiple training samples and the thickness label corresponding to each training sample includes:

[0018] Data expansion was performed on the actual copper plating parameters in the circuit board manufacturing process to obtain multiple sample copper plating design parameters;

[0019] Input the copper plating design parameters of each sample into the initial thickness prediction model, and obtain the sample simulation thickness output by the simulation module in the initial thickness prediction model;

[0020] Based on the predetermined error distribution law satisfied by the output results of the simulation module, the sample error corresponding to the simulation thickness of each sample is determined.

[0021] For each sample copper plating design parameter, based on the sample simulation thickness and sample error corresponding to the sample copper plating design parameter, the true value of the copper plating thickness corresponding to the sample copper plating design parameter is determined.

[0022] Each sample's copper plating design parameters are used as a training sample, and the true value of the copper plating thickness corresponding to the sample's copper plating design parameters is used as the thickness label corresponding to the training sample.

[0023] Optionally, the step of augmenting the actual copper plating parameters during the circuit board manufacturing process to obtain multiple sample copper plating design parameters includes:

[0024] By generating adversarial networks, the joint distribution law of the actual copper plating parameters in the circuit board production process is learned, and the actual copper plating parameters are augmented according to the joint distribution law to obtain multiple sample copper plating design parameters that meet the preset augmentation constraints.

[0025] Optionally, before the step of determining the sample error corresponding to the simulation thickness of each sample based on the predetermined error distribution law satisfied by the output results of the simulation module, the method further includes:

[0026] Each real copper plating parameter is input into the initial thickness prediction model, and the target simulated thickness corresponding to the real copper plating parameter output by the simulation module in the initial thickness prediction model is obtained.

[0027] Calculate the difference between the true value of the copper plating thickness and the target simulated thickness for each actual copper plating parameter;

[0028] The difference between the true value of the copper plating thickness corresponding to each real copper plating parameter and the target simulated thickness is input into the initial error prediction model so that the initial error prediction model can optimize the kernel function parameters to obtain an error prediction model that satisfies the error distribution law of the simulation module.

[0029] The step of determining the sample error corresponding to each sample simulation thickness based on the predetermined error distribution law satisfied by the output results of the simulation module includes:

[0030] Input the simulated thickness of each sample into the error prediction model, and obtain the sample error corresponding to the simulated thickness of that sample output by the error prediction model.

[0031] Optionally, the step of determining the thickness uniformity result of copper plating on the circuit board to be manufactured after applying the copper plating design parameters based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured includes:

[0032] Based on the corrected copper plating thickness of each region of the circuit board to be manufactured, the dispersion of the corrected copper plating thickness of each region is calculated, and the thickness uniformity result of the copper plating design parameters after copper plating of the circuit board to be manufactured is obtained.

[0033] Optionally, after the step of determining the thickness uniformity result of copper plating on the circuit board to be manufactured using the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be manufactured, the method further includes:

[0034] The copper plating design parameters are multiple, and the thickness uniformity result includes the thickness uniformity corresponding to each copper plating design parameter; the copper plating design parameter with the optimal copper plating thickness uniformity is taken as the target copper plating design parameter, and the circuit board is produced based on the structural design parameters included in the target copper plating design parameter, and the produced circuit board is copper plating performed according to the process design parameters included in the target copper plating design parameter; or,

[0035] If the thickness uniformity result corresponding to the copper plating design parameters does not meet the preset uniformity requirement, the copper plating design parameters, the thickness uniformity result, and the preset optimization algorithm are used to determine candidate copper plating design parameters that meet the parameter constraints corresponding to the copper plating design parameters. The copper plating design parameters of the circuit board to be produced are updated to the candidate copper plating design parameters, and the step of inputting the copper plating design parameters of the circuit board to be produced into the pre-trained thickness prediction model is returned until the thickness uniformity result corresponding to the obtained copper plating design parameters meets the preset uniformity requirement. The obtained copper plating design parameters are used as target copper plating design parameters to produce the circuit board based on the structural design parameters included in the target copper plating design parameters, and the produced circuit board is copper plated according to the process design parameters included in the target copper plating design parameters.

[0036] Secondly, embodiments of this application provide a device for predicting the uniformity of copper plating thickness, the device comprising:

[0037] The parameter input module is used to input the copper plating design parameters of the circuit board to be produced into a pre-trained thickness prediction model. The thickness prediction model includes a simulation module and a correction network. The copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be produced.

[0038] The simulation module is used to simulate the copper plating process of the pre-set circuit board to be produced based on the process design parameters and the structural design parameters, to obtain the simulated thickness of each area in the circuit board to be produced, and to input the simulated thickness into the correction network.

[0039] The calibration module is used by the calibration network to determine the corrected copper plating thickness of each area in the circuit board to be produced under the copper plating design parameters based on the copper plating design parameters and the simulation thickness, and to output the corrected copper plating thickness.

[0040] The prediction module is used to determine the thickness uniformity result of copper plating on the circuit board to be manufactured after applying the copper plating design parameters, based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured.

[0041] Thirdly, embodiments of this application provide an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0042] Memory, used to store computer programs;

[0043] When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0044] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the steps described in the first aspect above.

[0045] Fifthly, embodiments of this application also provide a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the steps described in the first aspect above.

[0046] Beneficial effects of the embodiments in this application:

[0047] The technical solution provided in this application constructs a thickness prediction model that integrates a simulation module and a correction network in series. This model combines the copper plating process simulation capability of the simulation module with the error correction capability of the correction network for the output results of the simulation module. Thus, when using this thickness prediction model to predict the uniformity of copper plating thickness on a circuit board to be manufactured, the simulation module can obtain a physically reliable simulated thickness that conforms to the copper plating process. The correction network then corrects the error of this simulated thickness, resulting in a highly accurate corrected copper plating thickness that closely approximates the actual copper plating result. This allows for the rapid and accurate determination of the copper plating thickness in each area of ​​the circuit board to be manufactured, ensuring the accuracy of the predicted uniformity of copper plating thickness. Of course, implementing any product or method of this application does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other embodiments can be obtained based on these drawings.

[0049] Figure 1 A flowchart illustrating a method for predicting the uniformity of copper plating thickness provided in an embodiment of this application;

[0050] Figure 2 A schematic diagram of a circuit board provided in an embodiment of this application;

[0051] Figure 3 A schematic diagram of the thickness prediction model provided in the embodiments of this application;

[0052] Figure 4 A schematic diagram showing the corrected copper plating thickness of each area of ​​the circuit board to be manufactured, provided in an embodiment of this application;

[0053] Figure 5 This is a schematic diagram of the training process of the thickness prediction model provided in the embodiments of this application;

[0054] Figure 6 for Figure 5 A schematic diagram illustrating the specific implementation of step S501;

[0055] Figure 7 This is a schematic diagram illustrating the overall process for achieving uniform copper plating thickness in an embodiment of this application.

[0056] Figure 8 A schematic diagram of a device for predicting the uniformity of copper plating thickness provided in an embodiment of this application;

[0057] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0058] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art based on this application are within the scope of protection of this application.

[0059] To predict the uniformity of copper plating thickness in Mini LEDs to be manufactured, embodiments of this application provide a method, apparatus, electronic device, computer-readable storage medium, and computer program product for predicting copper plating thickness uniformity. The method for predicting copper plating thickness uniformity provided in this application embodiment is described below.

[0060] The method for predicting the uniformity of copper plating thickness provided in this application embodiment can be applied to any electronic device with processing capabilities, such as a PC (Personal Computer), a cloud server, etc., and the electronic device can be an independent electronic device or a cluster of devices composed of multiple electronic devices, without specific limitations.

[0061] like Figure 1 As shown, a method for predicting the uniformity of copper plating thickness includes:

[0062] S101: Input the copper plating design parameters of the circuit board to be manufactured into the pre-trained thickness prediction model.

[0063] The thickness prediction model includes a simulation module and a correction network; the copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be produced.

[0064] S102: The simulation module simulates the copper plating process of the pre-set circuit board to be produced based on the process design parameters and the structural design parameters, obtains the simulated thickness of each area in the circuit board to be produced, and inputs the simulated thickness into the correction network.

[0065] S103: The correction network determines the corrected copper plating thickness of each area in the circuit board to be produced under the copper plating design parameters and the simulated thickness, and outputs the corrected copper plating thickness.

[0066] S104: Based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured, determine the thickness uniformity result of copper plating on the circuit board to be manufactured using the copper plating design parameters.

[0067] The technical solution provided in this application constructs a thickness prediction model that integrates a simulation module and a correction network in series. This model combines the copper plating process simulation capability of the simulation module with the error correction capability of the correction network for the output results of the simulation module. Thus, when using this thickness prediction model to predict the uniformity of copper plating thickness on a circuit board to be manufactured, the simulation module can obtain a physically reliable simulated thickness that conforms to the copper plating process. The correction network then corrects the error of this simulated thickness, resulting in a highly accurate corrected copper plating thickness that closely approximates the actual copper plating result. This allows for the rapid and accurate determination of the copper plating thickness in each area of ​​the circuit board to be manufactured, ensuring the accuracy of the predicted uniformity of copper plating thickness.

[0068] In the production of circuit boards, copper layers need to be deposited on the electrodes of the circuit board using electrochemical methods, i.e., copper plating, to achieve electrical interconnection and heat dissipation. The uniformity of copper plating thickness is affected by the circuit board structure and electroplating parameters. Therefore, in order to improve the uniformity of copper plating thickness, advanced electroplating equipment, special additives, optimized process design parameters, and circuit board structural design parameters can be used. Experiments, simulation models, or AI (Artificial Intelligence) methods can be used to control the copper plating process to ensure the yield rate of the produced circuit boards and improve the reliability of the circuit boards.

[0069] For example, ensuring the uniformity of copper plating thickness on the Mini LED circuit board is a critical and extremely challenging aspect of Mini LED manufacturing. The difficulty stems primarily from the high density and miniaturization of the Mini LED structure itself. Uneven current density distribution, extremely small pattern layout and spacing, limitations in electroplating solution mass transfer efficiency, and issues with seed layer quality and uniformity all affect the uniformity of copper plating thickness. To overcome the problems of uneven current distribution and electrolyte mass transfer caused by the microscale and high-density layout, systematic and coordinated optimization is needed across multiple aspects, including copper plating equipment, copper plating solutions, process design parameters, and circuit board structural design parameters.

[0070] In the copper plating experiment, to achieve uniform copper thickness, experienced engineers need to understand the principle of copper plating electrochemical reaction and determine the circuit board structure with better copper plating thickness uniformity through multiple adjustments to the circuit board structure for mass production. The adjustment is extremely difficult and the experiment is very expensive.

[0071] Before manufacturing, simulation models, such as finite element modeling, can be used to predict the uniformity of copper plating thickness on Mini LED circuit boards. However, this method is extremely complex and requires significant simplification. First, it necessitates constructing a copper plating process flow for tens of thousands of micrometer-scale circuit board structures. This construction process requires multi-physics coupling calculations, demanding extremely high computing power and is very time-consuming. Furthermore, the accuracy of the simulation model heavily depends on the simulation input parameters, especially since the mechanism of action of electroplating additives is extremely complex and difficult to describe with precise mathematical models. This becomes the biggest source of error in the simulation model. In addition, the simulation model ignores side reactions and impurities, requiring repeated calibration of the simulation model parameters through expensive experiments to improve simulation accuracy. Therefore, while simulation models can obtain results on copper plating thickness uniformity through simulation and qualitatively guide the optimization design trend of circuit boards, they cannot achieve high-precision quantitative predictions and still need to be closely integrated with experiments.

[0072] Predicting copper plating thickness using AI methods, such as building deep learning models, requires a large amount of real copper plating experimental data or simulation data. However, the high cost of copper plating experiments results in a very limited amount of real copper plating experimental data, and the simulation data is limited by the simulation accuracy of the simulation model, resulting in insufficient accuracy of the simulation data. Therefore, the application effect of AI methods to predict the copper plating thickness of circuit boards is not ideal.

[0073] As can be seen above, the design parameters, physical properties, and electrochemical reactions required for simulation modeling cannot be accurately obtained or calculated. Limited by human experience, the prediction accuracy of simulation modeling is limited. The method of predicting the uniformity of copper plating thickness on circuit boards based on pure AI large model methods relies on massive copper plating experimental data or simulation data. The design and implementation of copper plating thickness experiments are expensive, and the experimental data is very limited, resulting in insufficient accuracy of the trained large model. Therefore, it is impossible to use pure AI large model methods to predict copper thickness with high accuracy and thus achieve circuit board design optimization.

[0074] Therefore, how to provide a feasible, accurate and low-cost copper plating thickness uniformity prediction solution is a technical problem that needs to be solved. In order to solve this technical problem, the embodiments of this application pre-train a thickness prediction model, which includes a series simulation module and a correction network.

[0075] The simulation module describes the physical field and electrochemical reaction in the copper plating process using a predetermined formula. By describing the physical field and electrochemical reaction in the copper plating process, the copper plating design parameters to be tested are input into the simulation module. The simulation module can then use these copper plating design parameters to simulate the corresponding copper plating process and obtain the simulated thickness corresponding to these copper plating design parameters.

[0076] This simulation module is based on Faraday's law and the theory of secondary current distribution followed by the copper plating process. It establishes the control equations describing the relationship between current density and electrode geometry and bath parameters, and defines boundary conditions. Then, it uses numerical calculation methods such as the finite element method to solve the equations. The resulting mathematical model can simulate and calculate the copper plating thickness of each area on the circuit board based on the process design parameters corresponding to the input copper plating process and the structural design parameters of each area of ​​the circuit board.

[0077] For example, the governing equation is the Laplace equation. ,in, It is the Laplace operator, and in a three-dimensional Cartesian coordinate system (x, y, z), its expansion is: ; Represents electric potential, measured in volts (V).

[0078] For example, boundary conditions include anodic potential conditions, cathodic potential conditions, and polarization behavior conditions.

[0079] The correction network is a type of deep learning model, such as a convolutional neural network model or a recurrent neural network model. It can perform physical prior correction on the output results of the simulation module and output the corrected high-precision copper plating thickness. Thus, by cascading and fusing the simulation module and the correction network, it is possible to achieve high-precision prediction of the uniformity of copper plating thickness on circuit boards with strong spatial distribution characteristics.

[0080] For clarity, the training method of the thickness prediction model will be explained below.

[0081] When using a pre-trained thickness prediction model to predict the uniformity of copper plating thickness on a circuit board to be manufactured, the copper plating design parameters of the circuit board to be manufactured can be determined first. These copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be manufactured.

[0082] The structural design drawing of the circuit board (Panel) to be manufactured, such as... Figure 2 As shown, different regions of the circuit board to be manufactured have different structures, material properties, distances from the edges, and other parameters. The structural design drawings of the circuit board to be manufactured can be described using process design parameters so that the model can understand the structure of the circuit board to be manufactured and predict the uniformity of copper plating thickness for the circuit board to be manufactured with this structure.

[0083] The process design parameters corresponding to the above copper plating process are the electroplating parameters and chemical parameters used when plating copper on the circuit board to be produced. Among them, the electroplating parameters include current density, voltage, electroplating time, bath temperature, and pulse electroplating parameters; the chemical parameters include copper ion concentration, sulfuric acid concentration, solution ratio, and solution volume.

[0084] Multiple prediction points can be determined on the circuit board to be manufactured according to the preset prediction point selection conditions. The area within a preset range, such as the area within a 1mm×1mm range centered on the prediction point, is taken as the area corresponding to the prediction point, thereby determining multiple areas of the circuit board to be manufactured.

[0085] The structural design parameters of each region of the circuit board to be manufactured include the local density of each region of the circuit board to be manufactured, such as the metal coverage, the line width and line spacing of the conductors in the region, the distance between the region and the boundary of the circuit board, and the region type. The region type can be a dense wiring region, an isolated pad, or an empty region, and the region type can be quantitatively determined by the plating density.

[0086] Furthermore, the copper plating process exhibits spatial correlation. To facilitate the thickness prediction model's understanding of the circuit board's spatial structure, the structural design parameters for each region of the circuit board to be produced can also include the statistical characteristics of the surrounding regions, such as the average density of neighboring areas. Moreover, to accelerate model convergence and improve accuracy, parameters with different dimensions in the copper plating design parameters can be standardized or normalized, and the processed copper plating design parameters can be used as input copper plating design parameters for the thickness prediction model.

[0087] The electronic device can input the copper plating design parameters of the circuit board to be manufactured into a pre-trained thickness prediction model, i.e., execute step S101.

[0088] The thickness prediction model can input copper plating design parameters into the simulation module and the correction network.

[0089] Since the simulation module has a pre-set copper plating process for the circuit board to be produced, after obtaining the process design parameters and structural design parameters, the simulation module can simulate the copper plating process of the circuit board to be produced based on the process design parameters and structural design parameters, and obtain the simulated thickness of each area in the circuit board to be produced.

[0090] After obtaining the simulated thickness of each region in the circuit board to be manufactured, the simulation module can input the simulated thickness into the correction network in the thickness prediction model for correction, i.e., execute step S102.

[0091] The calibration network can determine the corrected copper plating thickness of each area in the circuit board to be produced under the copper plating design parameters and the simulated thickness, and output the corrected copper plating thickness, i.e., execute step S103.

[0092] The thickness prediction model integrates a simulation module and a correction network in series. The simulation module simulates the copper plating process of the circuit board to obtain the simulated thickness; the correction network corrects the simulated thickness, adjusting for errors in the simulation module's output, and outputting a high-precision corrected copper plating thickness that closely approximates the actual experimental data. In other words, the corrected copper plating thickness output by the correction network is actually the sum of the simulated thickness output by the simulation module and the error in that simulated thickness.

[0093] Then the electronic device can execute step S104, which is to determine the thickness uniformity result of the copper plating of the circuit board to be produced after applying copper plating design parameters based on the corrected copper plating thickness of each area of ​​the circuit board to be produced.

[0094] Since the uniformity of copper plating thickness of the circuit board to be manufactured can characterize the uniformity of copper plating thickness in each area of ​​the circuit board to be manufactured, after obtaining the corrected copper plating thickness of each area of ​​the circuit board to be manufactured output by the thickness prediction model, the electronic device can calculate the uniformity of the corrected copper plating thickness of each area and obtain the thickness uniformity result of the circuit board to be manufactured after copper plating using the copper plating design parameters.

[0095] In one implementation, the electronic device can calculate the average value of the corrected copper plating thickness of each area of ​​the circuit board to be manufactured, as the result of the thickness uniformity after copper plating the circuit board to be manufactured using the copper plating design parameters.

[0096] In another implementation, the electronic device can calculate the variance of the corrected copper plating thickness in each area of ​​the circuit board to be manufactured, as a result of the thickness uniformity of the circuit board to be manufactured after copper plating using the copper plating design parameters.

[0097] For example, such as Figure 3 As shown, the simulation module and correction network in the thickness prediction model are connected in series. The electronic device inputs the copper plating design parameters of the circuit board to be manufactured into the thickness prediction model; that is, it inputs the process design parameters and structural design parameters, which are included in the copper plating design parameters, into the simulation module and the correction network, respectively. Based on the process design parameters and structural design parameters, the simulation module simulates the copper plating process of the circuit board to be manufactured, obtaining the simulated thickness of each area in the circuit board, and inputs the simulated thickness into the correction network. The correction network determines the corrected copper plating thickness based on the process design parameters, structural design parameters, and the simulated thickness.

[0098] The copper plating thickness of each area of ​​the circuit board to be produced after calibration is as follows: Figure 4As shown, a coordinate system is established with the center point of the circuit board as the origin. The X-coordinate range is [-172, 172], and the Y-coordinate ranges from [-872, 895]. Since the circuit board is divided into multiple regions, the position of each region can be characterized by the coordinates of its center point. After obtaining the corrected copper plating thickness for each region, a table representing the corrected copper plating thickness of each region can be generated and displayed. This table includes the coordinates of each region and its corresponding corrected copper plating thickness. For example, for the region with coordinates (30, -710), the corrected copper plating thickness is 9.907; for the region with coordinates (-6, 15), the corrected copper plating thickness is 4.083; and for the region with coordinates (440, 700), the corrected copper plating thickness is 22.554. These are not listed individually. The electronic device can calculate the average value of the corrected copper plating thickness for each region to obtain the thickness uniformity result after copper plating the circuit board to be produced using the copper plating design parameters.

[0099] As can be seen, the technical solution provided in this application, by constructing a thickness prediction model that integrates a simulation module and a correction network in series, combines the copper plating process simulation capability of the simulation module with the error correction capability of the correction network for the output results of the simulation module. Thus, when using this thickness prediction model to predict the uniformity of copper plating thickness on a circuit board to be produced, the simulation module can obtain a physically reliable simulated thickness that conforms to the copper plating process. The correction network then corrects the error of this simulated thickness, resulting in a high-precision corrected copper plating thickness that closely approximates the actual copper plating result. This allows for the rapid and accurate determination of the copper plating thickness in each area of ​​the circuit board to be produced, ensuring the accuracy of the copper plating thickness uniformity prediction results.

[0100] As one implementation method of this application, such as Figure 5 As shown, the training methods for the above thickness prediction model can include:

[0101] S501: Obtain multiple training samples and the thickness label corresponding to each training sample.

[0102] Each training sample includes sample copper plating design parameters for a sample circuit board, wherein the sample copper plating design parameters include sample process design parameters corresponding to the copper plating process flow and sample structure design parameters for each region of the sample circuit board.

[0103] S502: For each training sample, input the training sample into the initial thickness prediction model.

[0104] S503: The simulation module in the initial thickness prediction model simulates the copper plating process of the sample circuit board based on the sample process design parameters and the sample structure design parameters, obtains the sample simulation thickness of each region in the sample circuit board, and inputs the sample simulation thickness into the initial correction network in the initial thickness prediction model.

[0105] To train a thickness prediction model, you can obtain the training samples for the thickness prediction model and the thickness label corresponding to each training sample.

[0106] After acquiring multiple training samples, the electronic device can input each training sample into the initial thickness prediction model. The initial thickness prediction model includes a simulation module and an initial calibration network.

[0107] The electronic device can input training samples into the simulation module and the initial correction network of the initial thickness prediction model, respectively.

[0108] The simulation module simulates the copper plating process of the pre-set sample circuit board based on the sample process design parameters and sample structure design parameters, obtains the sample simulation thickness of each area in the sample circuit board, and inputs the sample simulation thickness into the initial correction network.

[0109] S504: The initial calibration network determines the copper plating thickness after calibration in each region of the sample circuit board under the sample copper plating design parameters, based on the sample copper plating design parameters and the sample simulation thickness.

[0110] S505: Based on the difference between the copper plating thickness after sample correction corresponding to each training sample and the thickness label corresponding to that training sample, adjust the model parameters of the initial correction network until the initial correction network meets the convergence condition, and obtain the trained thickness prediction model.

[0111] The initial calibration network determines the calibrated copper plating thickness of each region in the sample circuit board under the sample copper plating design parameters and the sample simulation thickness, thus obtaining the calibrated copper plating thickness corresponding to the training sample.

[0112] After obtaining the corrected copper plating thickness for each training sample, the difference between the corrected copper plating thickness for that training sample and the thickness label for that training sample can be determined. The model parameters of the initial correction network are then adjusted based on the differences for each sample until the initial correction network meets the convergence condition, thus obtaining the trained correction network and the trained thickness prediction model.

[0113] In one implementation, the electronic device can adjust the parameters of the initial calibration network based on the difference between the copper plating thickness after sample calibration and the thickness label through backpropagation, so as to reduce the difference between the copper plating thickness after sample calibration and the thickness label, until the initial calibration network meets the convergence condition, thereby obtaining the trained calibration network and the trained thickness prediction model.

[0114] In this way, by training the thickness prediction model using multiple training samples and the corresponding thickness labels for each training sample, the correction network in the thickness prediction model can be trained, enabling the correction network to correct the simulated thickness output by the simulation module under the sample copper plating design parameters. The resulting thickness prediction model can then simulate the copper plating process of a circuit board, predict the copper plating thickness in various regions of the circuit board, and further correct the predicted copper plating thickness in each region of the circuit board to obtain high-precision copper plating thickness prediction results.

[0115] In the embodiments of this application, the electronic device used to train the thickness prediction model and the electronic device used to perform the prediction method for copper plating thickness uniformity using the thickness prediction model can be the same electronic device or different electronic devices, and no specific limitation is made here.

[0116] It does not rely on complex multiphysics models and massive amounts of data, but can directly learn from limited historical data (such as design graphics, process parameters and measured thickness). It is not limited by the simulation accuracy of simulation software, and can predict local copper thickness with high accuracy, achieve consistent prediction and optimization of thickness across the entire domain. It can break through the simulation limits of human experience and finite element simulation software, achieve high-precision prediction with a small amount of data, effectively capture the influence of factors that are difficult to model, and achieve efficient design iteration and online optimization. It has a stronger predictive generalization ability for complex graphic layouts.

[0117] As can be seen, in this embodiment, multiple training samples and their corresponding thickness labels are obtained. For each training sample, the training sample is input into the initial thickness prediction model. The simulation module in the initial thickness prediction model can predict the simulated thickness of each region in the sample circuit board and input the simulated thickness into the initial correction network in the thickness prediction model. Based on the sample copper plating design parameters and the simulated thickness, the initial correction network determines the corrected copper plating thickness of each region in the sample circuit board under the sample copper plating design parameters. According to the difference between the corrected copper plating thickness corresponding to each training sample and the thickness label corresponding to that training sample, the model parameters of the initial correction network are adjusted until the initial correction network meets the convergence condition, thus obtaining the trained thickness prediction model. In this way, through the above settings, the correction network can be trained using the training samples and the prediction results obtained by the simulation module predicting the training samples, based on the data processing relationship between the cascaded simulation module and the correction network in the thickness prediction model, thereby improving the overall accuracy of the thickness prediction model.

[0118] As one implementation method of this application, such as Figure 6 As shown, step S501 above, namely the step of obtaining multiple training samples and the thickness label corresponding to each training sample, may include:

[0119] S601: Expand the data of the actual copper plating parameters in the circuit board production process to obtain multiple sample copper plating design parameters.

[0120] To train the thickness prediction model, real copper plating parameters from the circuit board manufacturing process can be obtained as training samples. These real copper plating parameters include the real process parameters corresponding to the copper plating process and the real structural parameters of each area of ​​the produced circuit board.

[0121] However, as mentioned earlier, the actual copper plating parameters in the circuit board manufacturing process are limited. Using a limited number of training samples to train the thickness prediction model will result in training results that are difficult to meet the actual accuracy expectations.

[0122] Therefore, in order to obtain a sufficient number of training samples without consuming a large amount of experimental resources, the actual copper plating parameters in the circuit board manufacturing process can be augmented to obtain multiple sample copper plating design parameters. These multiple sample copper plating design parameters can include both the augmented copper plating parameters and the actual copper plating parameters; alternatively, they can include only the augmented copper plating parameters—both are reasonable and not specifically limited here.

[0123] In this way, without relying on complex and massive production experimental data, multiple sets of sample copper plating design parameters can be generated by conducting large-scale, designed sampling within a parameter space that covers real copper plating parameters and is appropriately expanded.

[0124] S602: Input the copper plating design parameters of each sample into the initial thickness prediction model, and obtain the sample simulation thickness output by the simulation module in the initial thickness prediction model.

[0125] However, since the true value of the copper plating thickness corresponding to the copper plating design parameters of the sample obtained by data augmentation is unknown, and the true value of the copper plating thickness is an important data in the model training process as a label, in order to obtain the true value of the copper plating thickness corresponding to the copper plating design parameters of the sample, the copper plating design parameters of each sample can be input into the initial thickness prediction model.

[0126] The initial thickness prediction model can input the sample copper plating design parameters into the simulation module. The simulation module can perform copper plating simulation based on the sample copper plating design parameters and the pre-set copper plating process to obtain the sample simulation thickness corresponding to the sample copper plating design parameters.

[0127] S603: Based on the predetermined error distribution law satisfied by the output results of the simulation module, determine the sample error corresponding to the simulation thickness of each sample.

[0128] Because the simulation accuracy of the simulation module is limited, there is an error between the simulated thickness of the sample copper plating design parameters output by the simulation module and the actual copper plating thickness. However, the output of the simulation module meets a certain error distribution law. Therefore, the error distribution law that the output of the simulation module meets can be determined in advance, and the sample error corresponding to the simulated thickness of the sample output by the simulation module can be determined by using the error distribution law, thus obtaining the sample error corresponding to the sample copper plating design parameters.

[0129] S604: For each sample copper plating design parameter, based on the sample simulation thickness and sample error corresponding to the sample copper plating design parameter, determine the true value of the copper plating thickness corresponding to the sample copper plating design parameter.

[0130] For each sample copper plating design parameter, after determining the sample error corresponding to the sample copper plating design parameter, the sum of the sample simulated thickness corresponding to the sample copper plating design parameter and the sample error corresponding to the sample copper plating design parameter can be calculated to obtain the true value of the copper plating thickness corresponding to the sample copper plating design parameter.

[0131] S605: Use the copper plating design parameters of each sample as a training sample, and use the true value of the copper plating thickness corresponding to the copper plating design parameters of that sample as the thickness label of that training sample.

[0132] Furthermore, once the true value of the copper plating thickness corresponding to the copper plating design parameters of each sample is obtained, the copper plating design parameters of that sample can be used as a training sample, and the true value of the copper plating thickness corresponding to the copper plating design parameters of that sample can be used as the thickness label of that training sample.

[0133] Traditional data augmentation modules directly expand the input and output data of the model. However, in this embodiment, the model input is expanded, and the simulation module in the thickness prediction model is used to perform simulation prediction to obtain the predicted thickness and determine the labels corresponding to the input data. Through the above process, large-scale and designed sampling is carried out in a parameter space that covers the real copper plating parameters and is appropriately expanded. Multiple sets of sample copper plating design parameters and the true value pairs of copper plating thickness corresponding to the sample copper plating design parameters are generated, resulting in an "augmented dataset" for training the correction network. This expands the input data and training labels of the correction network. Therefore, the expanded data is essentially the input data of the simulation module, as well as the input data and labels required by the correction network.

[0134] As can be seen, in this embodiment, the electronic device can augment the data of the actual copper plating parameters in the circuit board manufacturing process to obtain multiple sample copper plating design parameters; each sample copper plating design parameter is input into the initial thickness prediction model, and the sample simulated thickness output by the simulation module in the initial thickness prediction model is obtained; based on the error distribution law satisfied by the output results of the simulation module in advance, the sample error corresponding to each sample simulated thickness is determined; for each sample copper plating design parameter, based on the sample simulated thickness and sample error corresponding to the sample copper plating design parameter, the true value of the copper plating thickness corresponding to the sample copper plating design parameter is determined; each sample copper plating design parameter is used as a training sample, and the true value of the copper plating thickness corresponding to the sample copper plating design parameter is used as the thickness label corresponding to the training sample. In this way, through the above settings, without obtaining massive amounts of experimental or production data, a large number of training samples and corresponding labels can be obtained using limited actual copper plating parameters in the circuit board manufacturing process, thereby ensuring the data of the training samples, and thus ensuring the accuracy of the thickness prediction model obtained by training using the training samples and their corresponding labels, realizing efficient iteration and online optimization of model parameters, and improving the model's prediction generalization ability for complex copper plating design parameters.

[0135] As one embodiment of this application, step S601, namely the step of augmenting the actual copper plating parameters in the circuit board manufacturing process to obtain multiple sample copper plating design parameters, may include:

[0136] By generating adversarial networks, the joint distribution law of the actual copper plating parameters in the circuit board production process is learned, and the actual copper plating parameters are augmented according to the joint distribution law to obtain multiple sample copper plating design parameters that meet the preset augmentation constraints.

[0137] Since the actual copper plating parameters in the circuit board manufacturing process are limited, in order to obtain better model training results, a Production Adversarial Network (Conditional Tabular GAN, CTGAN) can be used to augment the actual copper plating parameters. This way, a sufficient number of training samples can be obtained without additional copper plating experiments, thus achieving a cold start for model training.

[0138] Generative Adversarial Networks (GANs) are deep learning frameworks specifically designed to generate high-quality synthetic tabular data. These GANs learn the distribution patterns of real-world data, generating virtual datasets that are highly similar in distribution to the original real data while preserving the correlations between data columns.

[0139] Electronic devices can utilize generative adversarial networks (GANs) to learn the joint distribution patterns of actual copper plating parameters during circuit board manufacturing. Then, based on these joint distribution patterns, and within a reasonable range, the actual copper plating parameters can be sampled or have random perturbations added to generate a batch of new sample copper plating design parameters that meet preset augmentation constraints, thereby achieving data augmentation.

[0140] The aforementioned preset expansion constraints can be set according to actual conditions. They may include the sample copper plating design parameters obtained from data expansion satisfying the joint distribution law, and may also include the parameter constraints to be expanded in the pre-set real copper plating parameters, the data quantity constraints after data expansion, and the step size constraints of data expansion. No specific limitations are made here.

[0141] For example, generative adversarial networks have a conditional vector mechanism, which can construct conditional vectors to force the generative adversarial network to follow the joint distribution law of real copper plating parameters and other conditional constraints when data is augmented, so as to ensure the rationality of the copper plating design parameters of the generated samples.

[0142] As can be seen, in this embodiment, the electronic device can learn the joint distribution pattern of the actual copper plating parameters in the circuit board manufacturing process through a generative adversarial network (GAN). Following this joint distribution pattern, the actual copper plating parameters are augmented to obtain multiple sample copper plating design parameters that satisfy preset augmentation constraints. In this way, by augmenting data through a GAN, a large number of sample copper plating design parameters that are highly similar to the actual copper plating parameters, meet design requirements, and possess diversity can be generated. This fills the training sample gap in the cold start phase of model training. Furthermore, the model can be trained using the sample copper plating design parameters obtained from the data augmentation, allowing the model to learn sufficient features and thus possess high predictive ability.

[0143] As one embodiment of this application, before step S603 above, i.e., before determining the sample error corresponding to each sample simulation thickness based on the error distribution law satisfied by the output result of the simulation module, the copper plating thickness uniformity prediction method provided in this application embodiment may further include:

[0144] Each real copper plating parameter is input into the initial thickness prediction model, and the target simulated thickness corresponding to the real copper plating parameter output by the simulation module in the initial thickness prediction model is obtained; the difference between the true value of the copper plating thickness corresponding to each real copper plating parameter and the target simulated thickness is calculated; the difference between the true value of the copper plating thickness corresponding to each real copper plating parameter and the target simulated thickness is input into the initial error prediction model, so that the initial error prediction model optimizes the kernel function parameters to obtain an error prediction model that satisfies the error distribution law of the simulation module.

[0145] Since the simulation module cannot perfectly and accurately reproduce the actual copper plating process, there will be an error between the simulation module's output and the actual copper plating result. Assuming that the error of the simulation module varies smoothly within a certain range, the error of the simulation module's output may follow a certain distribution pattern.

[0146] Therefore, electronic devices can use the actual copper plating parameters in the circuit board manufacturing process and the true value of the copper plating thickness corresponding to the actual copper plating parameters to determine the error distribution law satisfied by the output results of the simulation module.

[0147] The electronic device can input each real copper plating parameter into the initial thickness prediction model. The simulation module in the initial thickness prediction model can use the real copper plating parameters and the pre-set copper plating process to perform simulation and output the target simulated thickness corresponding to the real copper plating parameters.

[0148] Then, the electronic device can obtain the target simulated thickness corresponding to the real copper plating parameters output by the simulation module, and calculate the difference between the target simulated thickness and the true value of the copper plating thickness corresponding to the real copper plating parameters, thereby determining the error between the target simulated thickness and the true value of the copper plating thickness.

[0149] Given the errors corresponding to each actual copper plating parameter, the electronic device can use the error between the true value of the copper plating thickness and the target simulated thickness to train an error prediction model. This error prediction model can then capture the error distribution law satisfied by the output results of the simulation module. This error prediction model can be a Gaussian regression model.

[0150] Specifically, the electronic device can input the difference between the true value of the copper plating thickness corresponding to each actual copper plating parameter and the target simulated thickness into the initial error prediction model. The initial error prediction model can then solve for the optimal kernel parameter combination by maximizing the marginal likelihood function based on the difference between the true value of the copper plating thickness corresponding to each actual copper plating parameter and the target simulated thickness. This kernel parameter combination is then used to update the kernel function parameters, resulting in a trained error prediction model. This trained error prediction model possesses the ability to predict errors in the input simulated thickness, thus obtaining an error prediction model that satisfies the error distribution law of the simulation module.

[0151] Step S603 above, namely the step of determining the sample error corresponding to the simulation thickness of each sample based on the error distribution law satisfied by the output results of the simulation module, may include:

[0152] Input the simulated thickness of each sample into the error prediction model, and obtain the sample error corresponding to the simulated thickness of that sample output by the error prediction model.

[0153] After obtaining the error prediction model, the electronic device can input the simulated thickness of each sample into the error prediction model, so that the error prediction model can predict the error of each simulated thickness and output the sample error corresponding to the simulated thickness. The electronic device can obtain the sample error corresponding to the simulated thickness of the sample output by the error prediction model, and then use the error and the simulated thickness of the sample to determine the thickness label of the training sample.

[0154] In this way, by using the difference between the true value of the copper plating thickness corresponding to the actual copper plating parameters and the target simulated thickness as input, a simple error prediction model, such as a Gaussian regression model, is trained to learn the difference between the true value of the copper plating thickness and the target simulated thickness. This error prediction model can predict the deviation Delta_sim of the prediction results output by the simulation module. The target simulated thickness T_sim corresponding to the sample copper plating design parameters predicted by multiple simulation modules is added to the deviation Delta_sim constructed by the error prediction model to calculate the true value of the copper plating thickness corresponding to the sample copper plating design parameters, T_enhanced = T_sim + Delta_sim. In this way, the enhanced training set used to train the correction network can be obtained.

[0155] As can be seen, in this embodiment, the electronic device can use the difference between the target simulated thickness corresponding to the real copper plating parameters output by the simulation module and the true value of the copper plating thickness corresponding to the real copper plating parameters to construct an error prediction model that satisfies the error distribution law of the simulation module, and use the error prediction model to predict the error of the target simulated thickness output by the simulation module, thereby expanding the training data of the correction network.

[0156] As one embodiment of this application, the step of determining the thickness uniformity result of copper plating on the circuit board to be manufactured after applying the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be manufactured includes:

[0157] Based on the corrected copper plating thickness of each region of the circuit board to be manufactured, the dispersion of the corrected copper plating thickness of each region is calculated, and the thickness uniformity result of the copper plating design parameters after copper plating of the circuit board to be manufactured is obtained.

[0158] By using a thickness prediction model to predict the copper plating thickness of the circuit board to be produced, the electronic device can obtain the corrected copper plating thickness of each area of ​​the circuit board to be produced. After obtaining the corrected copper plating thickness of each area, the electronic device can calculate the dispersion of the corrected copper plating thickness of each area.

[0159] The dispersion of the corrected copper plating thickness in each region can be represented by the mean, variance, and standard deviation.

[0160] For example, the circuit board to be manufactured includes n regions, and the corrected copper plating thickness of each region... Where i = 1, 2, ..., n, the average corrected copper plating thickness of each region can be calculated, and the average corrected copper plating thickness of each region can be used. Calculate the variance of the corrected copper plating thickness in each region. :

[0161] ;

[0162] The variance is used as a copper plating design parameter to determine the uniformity of copper plating thickness on the circuit board to be produced. The smaller the variance, the more uniform the copper plating thickness in each area of ​​the circuit board to be produced, and the better the uniformity of copper plating thickness on the circuit board to be produced.

[0163] As can be seen, in this embodiment, the electronic device can calculate the dispersion of the corrected copper plating thickness in each region, and obtain the copper plating design parameters and the thickness uniformity result after copper plating on the circuit board to be produced. Through the above settings, the electronic device can quickly and accurately obtain the thickness uniformity result of the circuit board to be produced, and use the thickness uniformity result to optimize the copper plating design parameters of the circuit board to be produced, so as to produce a circuit board that meets the requirements of copper plating thickness uniformity.

[0164] As one embodiment of this application, the copper plating design parameters are multiple, and the thickness uniformity result includes the thickness uniformity corresponding to each copper plating design parameter.

[0165] After determining the copper plating thickness uniformity result of the copper plating design parameters on the circuit board to be manufactured based on the corrected copper plating thickness of each region, the copper plating thickness uniformity prediction method provided in this application embodiment may further include:

[0166] The copper plating design parameters with the best uniformity of the corresponding copper plating thickness are taken as the target copper plating design parameters. The circuit board is produced based on the structural design parameters included in the target copper plating design parameters, and the produced circuit board is copper plating according to the process design parameters included in the target copper plating design parameters.

[0167] To ensure good copper plating thickness uniformity in the manufactured circuit boards, multiple copper plating design parameters can be designed for the circuit board to be manufactured. A thickness prediction model can be used to predict the copper plating thickness uniformity for each copper plating design parameter, thereby obtaining the thickness uniformity corresponding to each copper plating design parameter.

[0168] In this process, the parameters in the process design parameters corresponding to the copper plating process or the structural design parameters of each area of ​​the circuit board to be produced can be adjusted. For example, the local density, line width and line spacing of the conductors, and the distance between the area and the boundary of the circuit board can be adjusted to obtain multiple copper plating design parameters.

[0169] After obtaining the thickness uniformity corresponding to each copper plating design parameter, the copper plating design parameter with the best copper plating thickness uniformity is taken as the target copper plating design parameter. The optimal copper plating thickness uniformity means that the dispersion of the corrected copper plating thickness in each region is the smallest.

[0170] This allows workers to use the structural design parameters included in the target copper plating design parameters as the structural design parameters for the circuit board, manufacture the circuit board according to these structural design parameters, and then plate the circuit board with copper according to the process design parameters included in the target copper plating design parameters. In this way, the resulting circuit board has better uniformity in copper plating thickness.

[0171] Furthermore, when the thickness uniformity corresponding to each copper plating design parameter does not meet the uniformity requirement, the electronic device analyzes the influence trend of the changing parameters in each copper plating design parameter on the thickness uniformity based on the thickness uniformity of each copper plating design parameter. Based on this influence trend, the copper plating design parameters are further optimized, and the thickness prediction model is used to continue predicting the copper plating thickness uniformity of the optimized copper plating design parameters until the copper plating thickness uniformity of the optimized copper plating design parameters meets the uniformity requirement. The circuit board is then manufactured using the structural design parameters included in the optimized copper plating design parameters, and copper is plated on the circuit board according to the process design parameters included in the optimized copper plating design parameters.

[0172] As can be seen, in this embodiment, when there are multiple copper plating design parameters, the electronic device can use the copper plating design parameter with the optimal copper plating thickness uniformity as the target copper plating design parameter. The circuit board is then manufactured using the structural design parameters included in the target copper plating design parameter, and copper plating is performed on the circuit board using the process design parameters included in the target copper plating design parameter. Thus, by setting multiple copper plating design parameters before circuit board production and predicting the copper plating thickness uniformity for each parameter, the circuit board can be manufactured and copper plating performed using the copper plating design parameter with the best copper plating thickness uniformity, resulting in a final circuit board with excellent performance.

[0173] As one embodiment of this application, after the step of determining the thickness uniformity result of copper plating on the circuit board to be manufactured using the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be manufactured, the copper plating thickness uniformity prediction method provided in this application embodiment may further include:

[0174] If the thickness uniformity result corresponding to the copper plating design parameters does not meet the preset uniformity requirement, the copper plating design parameters, the thickness uniformity result, and the preset optimization algorithm are used to determine candidate copper plating design parameters that meet the parameter constraints corresponding to the copper plating design parameters. The copper plating design parameters of the circuit board to be produced are updated to the candidate copper plating design parameters, and the step of inputting the copper plating design parameters of the circuit board to be produced into the pre-trained thickness prediction model is returned until the thickness uniformity result corresponding to the obtained copper plating design parameters meets the preset uniformity requirement. The obtained copper plating design parameters are used as target copper plating design parameters to produce the circuit board based on the structural design parameters included in the target copper plating design parameters, and the produced circuit board is copper plated according to the process design parameters included in the target copper plating design parameters.

[0175] To find suitable target copper plating design parameters for production, preset uniformity requirements can be set. If the thickness uniformity result corresponding to a certain set of copper plating design parameters meets the preset uniformity requirements, the copper plating design parameters can be used as the target copper plating design parameters.

[0176] As mentioned earlier, copper plating design parameters include circuit board structure design parameters and process design parameters, and each parameter includes a certain range of variation. The method of designing a large number of copper plating design parameters by controlling variables and solving the thickness uniformity result of each copper plating design parameter one by one is wasteful of a lot of computing power and extremely inefficient.

[0177] Therefore, in order to improve the efficiency of determining the target copper plating design parameters, parameter constraints can be used as constraints, and copper plating design parameters, thickness uniformity results and preset optimization algorithms can be used to optimize the parameters, thereby determining the candidate copper plating design parameters that meet the parameter constraints.

[0178] The preset optimization algorithm can be any type of algorithm, such as particle swarm optimization, ant colony optimization, sparrow search, etc. For example, the copper plating design parameters and thickness uniformity results can be input into the algorithm solver that uses particle swarm optimization as the optimization algorithm to obtain the optimized candidate copper plating design parameters.

[0179] The copper plating design parameters of the circuit board to be manufactured are updated to candidate copper plating design parameters. The updated copper plating design parameters of the circuit board to be manufactured are then input into a pre-trained thickness prediction model to obtain the thickness uniformity result corresponding to the updated copper plating design parameters of the circuit board to be manufactured. The model then continues to determine whether the thickness uniformity result meets the preset uniformity requirements.

[0180] If the thickness uniformity result corresponding to the updated copper plating design parameters of the circuit board to be produced still does not meet the preset uniformity requirement, the updated copper plating design parameters, thickness uniformity result and preset optimization algorithm are used to optimize the parameters. This process is repeated until the copper plating design parameters whose thickness uniformity result meets the preset uniformity requirement are obtained, which are then used as the target copper plating design parameters.

[0181] This allows workers to use the structural design parameters included in the target copper plating design parameters as the structural design parameters for the circuit board, manufacture the circuit board according to these structural design parameters, and then plate the circuit board with copper according to the process design parameters included in the target copper plating design parameters. In this way, the resulting circuit board has better uniformity in copper plating thickness.

[0182] As can be seen, in this embodiment, by using a preset optimization algorithm for rapid iteration, the efficiency of obtaining copper plating design parameters that meet the preset uniformity requirements for thickness uniformity is improved, the process design verification speed is accelerated, and manpower is saved.

[0183] To facilitate understanding of the method for predicting copper plating thickness uniformity provided in the embodiments of this application, the following is combined with... Figure 7 The overall process will be introduced.

[0184] like Figure 7 As shown, in order to predict the uniformity of copper plating thickness on a circuit board, a simulation module can be built first. This simulation module is a mechanism model. By building the mechanism model, the copper plating thickness in each area can be predicted based on process design parameters and structural design parameters.

[0185] Subsequently, a data augmentation scheme for electroplating based on conditional table generative adversarial networks (GANs) is implemented. The GANs are used to augment the actual copper plating parameters during the circuit board manufacturing process, increasing the data volume to obtain multiple sample copper plating design parameters. Each sample copper plating design parameter is then input into the simulation module, and the simulated sample thickness output by the simulation module is obtained. Based on a pre-defined error prediction model satisfied by the output results of the simulation module, the sample error corresponding to each sample simulated thickness is determined.

[0186] For each sample copper plating design parameter, based on the sample simulation thickness and sample error corresponding to the sample copper plating design parameter, the true value of the copper plating thickness corresponding to the sample copper plating design parameter is determined; each sample copper plating design parameter is used as a training sample, and the true value of the copper plating thickness corresponding to the sample copper plating design parameter is used as the thickness label corresponding to the training sample to train the correction network, so that the correction network has the ability to correct the error of the output result of the simulation module.

[0187] The simulation module and the correction network are fused together to obtain the fused model, which is the thickness prediction model in this application. The fused model realizes high-precision prediction of the copper plating thickness in each region.

[0188] Multiple copper plating design parameters are designed for the circuit board to be produced. The parameters are adjusted and optimized based on the thickness uniformity prediction results of the thickness prediction model. The copper plating design parameters of the circuit board are adjusted with the copper plating thickness uniformity as the target to obtain the copper plating thickness uniformity optimal copper plating design parameters. The circuit board is then produced and copper plating is performed.

[0189] By cascading and integrating the simulation module and the correction network, the thickness prediction model combines the physical reliability of the simulation module with the high-precision data of the correction network, giving full play to the advantages of both. This achieves an order-of-magnitude increase in prediction speed while maintaining near-high-fidelity simulation accuracy, and greatly reduces computational costs.

[0190] Furthermore, without requiring massive amounts of experimental and simulation data, the simulation module provides physically meaningful simulated thicknesses using only real copper plating parameters from the limited circuit board manufacturing process. This guides the calibration network to learn the correct patterns, enhancing the model's generalization ability in data-scarce regions and the physical rationality of the prediction results. By using the simulation module to simulate real physical reactions and output simulated thicknesses, and then using the calibration network to analyze the residual between the simulated thickness output by the simulation module and the actual copper thickness value, the accuracy of copper thickness prediction is improved. This allows for efficient matching of the copper plating design parameters with optimal uniformity through uniformity testing of multiple copper plating design parameters.

[0191] Corresponding to the above-mentioned method for predicting the uniformity of copper plating thickness, this application also provides a device for predicting the uniformity of copper plating thickness. The following is a description of the device for predicting the uniformity of copper plating thickness provided in this application.

[0192] like Figure 8 As shown, a device for predicting the uniformity of copper plating thickness includes:

[0193] The parameter input module 801 is used to input the copper plating design parameters of the circuit board to be produced into a pre-trained thickness prediction model, wherein the thickness prediction model includes a simulation module and a correction network; the copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be produced.

[0194] Simulation module 802 is used in the thickness prediction model to simulate the copper plating process of the pre-set circuit board to be produced based on the process design parameters and the structural design parameters, to obtain the simulated thickness of each region in the circuit board to be produced, and to input the simulated thickness into the correction network in the thickness prediction model.

[0195] The calibration module 803 is used to determine the corrected copper plating thickness of each area in the circuit board to be produced under the copper plating design parameters based on the copper plating design parameters and the simulation thickness, and output the corrected copper plating thickness.

[0196] The prediction module 804 is used to determine the thickness uniformity result of copper plating on the circuit board to be manufactured after applying the copper plating design parameters, based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured.

[0197] The technical solution provided in this application constructs a thickness prediction model that integrates a simulation module and a correction network in series. This model combines the copper plating process simulation capability of the simulation module with the error correction capability of the correction network for the output results of the simulation module. Thus, when using this thickness prediction model to predict the uniformity of copper plating thickness on a circuit board to be manufactured, the simulation module can obtain a physically reliable simulated thickness that conforms to the copper plating process. The correction network then corrects the error of this simulated thickness, resulting in a highly accurate corrected copper plating thickness that closely approximates the actual copper plating result. This allows for the rapid and accurate determination of the copper plating thickness in each area of ​​the circuit board to be manufactured, ensuring the accuracy of the predicted uniformity of copper plating thickness.

[0198] As one embodiment of this application, the apparatus further includes a training module; the training module includes:

[0199] The sample acquisition submodule is used to acquire multiple training samples and the thickness label corresponding to each training sample. Each training sample includes the sample copper plating design parameters of the sample circuit board. The sample copper plating design parameters include the sample process design parameters corresponding to the copper plating process and the sample structure design parameters of each region of the sample circuit board.

[0200] The first input submodule is used to input the training sample into the initial thickness prediction model for each training sample.

[0201] The second input submodule is used by the simulation module in the initial thickness prediction model to simulate the copper plating process of the sample circuit board based on the sample process design parameters and the sample structure design parameters, to obtain the sample simulation thickness of each region in the sample circuit board, and to input the sample simulation thickness into the initial correction network in the initial thickness prediction model.

[0202] The thickness determination submodule is used by the initial calibration network to determine the corrected copper plating thickness of each region in the sample circuit board under the sample copper plating design parameters, based on the sample copper plating design parameters and the sample simulated thickness.

[0203] The parameter adjustment submodule is used to adjust the model parameters of the initial calibration network based on the difference between the copper plating thickness after sample correction and the thickness label corresponding to each training sample, until the initial calibration network meets the convergence condition and the trained thickness prediction model is obtained.

[0204] As one embodiment of this application, the sample acquisition submodule includes:

[0205] The expansion unit is used to expand the data of the actual copper plating parameters in the circuit board production process to obtain multiple sample copper plating design parameters.

[0206] The input unit is used to input the copper plating design parameters of each sample into the initial thickness prediction model and obtain the sample simulation thickness output by the simulation module in the initial thickness prediction model.

[0207] An error determination unit is used to determine the sample error corresponding to the simulation thickness of each sample based on the error distribution law satisfied by the output results of the simulation module in a pre-determined manner.

[0208] The truth value determination unit is used to determine the true value of the copper plating thickness corresponding to the copper plating design parameters of each sample, based on the sample simulation thickness and sample error corresponding to the copper plating design parameters of that sample.

[0209] The sample determination unit is used to take the copper plating design parameters of each sample as a training sample and take the true value of the copper plating thickness corresponding to the copper plating design parameters of the sample as the thickness label of the training sample.

[0210] As one embodiment of this application, the expansion unit includes:

[0211] The augmentation subunit is used to learn the joint distribution law of the actual copper plating parameters in the circuit board production process through a generative adversarial network, and to augment the actual copper plating parameters according to the joint distribution law to obtain multiple sample copper plating design parameters that meet the preset augmentation constraints.

[0212] As one embodiment of this application, the apparatus further includes:

[0213] The first input module is used to input each real copper plating parameter into the initial thickness prediction model before determining the sample error corresponding to each sample simulation thickness based on the error distribution law satisfied by the output result of the simulation module in the pre-determined simulation module, and to obtain the target simulation thickness corresponding to the real copper plating parameter output by the simulation module in the initial thickness prediction model.

[0214] The difference calculation module is used to calculate the difference between the true value of the copper plating thickness corresponding to each real copper plating parameter and the target simulated thickness;

[0215] The second input module is used to input the difference between the true value of the copper plating thickness corresponding to each real copper plating parameter and the target simulated thickness into the initial error prediction model, so that the initial error prediction model can optimize the kernel function parameters to obtain an error prediction model that satisfies the error distribution law of the simulation module.

[0216] The error determination unit includes:

[0217] The error determination subunit is used to input the simulated thickness of each sample into the error prediction model and obtain the sample error corresponding to the simulated thickness of the sample output by the error prediction model.

[0218] As one embodiment of this application, the prediction module 804 includes:

[0219] The prediction submodule is used to calculate the dispersion of the corrected copper plating thickness of each region of the circuit board to be manufactured based on the corrected copper plating thickness of each region, and to obtain the thickness uniformity result of the copper plating design parameters after copper plating of the circuit board to be manufactured.

[0220] As one embodiment of this application, the apparatus further includes:

[0221] The parameter optimization module is used after the step of determining the thickness uniformity result of copper plating on the circuit board to be produced using the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be produced. The copper plating design parameters are multiple, and the thickness uniformity result includes the thickness uniformity corresponding to each copper plating design parameter. After predicting the thickness uniformity of copper plating on the circuit board to be produced using the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be produced, the module selects the copper plating design parameter with the optimal copper plating thickness uniformity as the target copper plating design parameter. The circuit board is then produced based on the structural design parameters included in the target copper plating design parameter, and copper plating is performed on the produced circuit board according to the process design parameters included in the target copper plating design parameter. Alternatively, in the... If the thickness uniformity result corresponding to the copper plating design parameters does not meet the preset uniformity requirement, the copper plating design parameters, the thickness uniformity result, and the preset optimization algorithm are used to determine candidate copper plating design parameters that meet the parameter constraints corresponding to the copper plating design parameters; the copper plating design parameters of the circuit board to be produced are updated to the candidate copper plating design parameters, and the step of inputting the copper plating design parameters of the circuit board to be produced into the pre-trained thickness prediction model is returned until the thickness uniformity result corresponding to the obtained copper plating design parameters meets the preset uniformity requirement. The obtained copper plating design parameters are then used as target copper plating design parameters to produce the circuit board based on the structural design parameters included in the target copper plating design parameters, and the produced circuit board is copper plated according to the process design parameters included in the target copper plating design parameters.

[0222] This application also provides an electronic device, such as... Figure 9 As shown, it includes a processor 901, a communication interface 902, a memory 903, and a communication bus 904, wherein the processor 901, the communication interface 902, and the memory 903 communicate with each other through the communication bus 904.

[0223] Memory 903 is used to store computer programs;

[0224] The processor 901, when executing the program stored in the memory 903, implements the method for predicting the uniformity of copper plating thickness as described in any of the above embodiments.

[0225] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0226] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0227] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0228] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0229] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor, implements the steps of any of the above-described methods for predicting the uniformity of copper plating thickness.

[0230] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the copper plating thickness uniformity prediction methods described in the above embodiments.

[0231] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0232] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0233] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, electronic devices, computer-readable storage media, and computer program products are basically similar to the method embodiments, and therefore the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0234] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application are included within the scope of protection of this application.

Claims

1. A method for predicting the uniformity of copper plating thickness, characterized in that, The method includes: The copper plating design parameters of the circuit board to be manufactured are input into a pre-trained thickness prediction model, wherein the thickness prediction model includes a simulation module and a correction network; the copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be manufactured. The simulation module simulates the copper plating process of the pre-set circuit board to be produced based on the process design parameters and the structural design parameters, obtains the simulated thickness of each area in the circuit board to be produced, and inputs the simulated thickness into the correction network. The correction network determines the corrected copper plating thickness of each area in the circuit board to be manufactured under the copper plating design parameters and the simulated thickness, and outputs the corrected copper plating thickness. Based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured, the thickness uniformity result of copper plating on the circuit board to be manufactured using the copper plating design parameters is determined.

2. The method according to claim 1, characterized in that, The training methods for the thickness prediction model include: Multiple training samples and thickness labels corresponding to each training sample are obtained. Each training sample includes copper plating design parameters of the sample circuit board. The copper plating design parameters include sample process design parameters corresponding to the copper plating process and sample structure design parameters of each region of the sample circuit board. For each training sample, input the training sample into the initial thickness prediction model; The simulation module in the initial thickness prediction model simulates the copper plating process of the sample circuit board based on the sample process design parameters and the sample structure design parameters, obtains the sample simulation thickness of each region in the sample circuit board, and inputs the sample simulation thickness into the initial correction network in the initial thickness prediction model. The initial calibration network determines the corrected copper plating thickness of each region in the sample circuit board under the sample copper plating design parameters, based on the sample copper plating design parameters and the sample simulated thickness. Based on the difference between the copper plating thickness after sample correction for each training sample and the thickness label corresponding to that training sample, the model parameters of the initial correction network are adjusted until the initial correction network meets the convergence condition, thus obtaining the trained thickness prediction model.

3. The method according to claim 2, characterized in that, The step of obtaining multiple training samples and the thickness label corresponding to each training sample includes: Data expansion was performed on the actual copper plating parameters in the circuit board manufacturing process to obtain multiple sample copper plating design parameters; Input the copper plating design parameters of each sample into the initial thickness prediction model, and obtain the sample simulation thickness output by the simulation module in the initial thickness prediction model; Based on the predetermined error distribution law satisfied by the output results of the simulation module, the sample error corresponding to the simulation thickness of each sample is determined. For each sample copper plating design parameter, based on the sample simulation thickness and sample error corresponding to the sample copper plating design parameter, the true value of the copper plating thickness corresponding to the sample copper plating design parameter is determined. Each sample's copper plating design parameters are used as a training sample, and the true value of the copper plating thickness corresponding to the sample's copper plating design parameters is used as the thickness label corresponding to the training sample.

4. The method according to claim 3, characterized in that, The step of augmenting the actual copper plating parameters during the circuit board manufacturing process to obtain multiple sample copper plating design parameters includes: By generating adversarial networks, the joint distribution law of the actual copper plating parameters in the circuit board production process is learned, and the actual copper plating parameters are augmented according to the joint distribution law to obtain multiple sample copper plating design parameters that meet the preset augmentation constraints.

5. The method according to claim 3, characterized in that, Before the step of determining the sample error corresponding to the simulation thickness of each sample based on the predetermined error distribution law satisfied by the output results of the simulation module, the method further includes: Each real copper plating parameter is input into the initial thickness prediction model, and the target simulated thickness corresponding to the real copper plating parameter output by the simulation module in the initial thickness prediction model is obtained. Calculate the difference between the true value of the copper plating thickness and the target simulated thickness for each actual copper plating parameter; The difference between the true value of the copper plating thickness corresponding to each real copper plating parameter and the target simulated thickness is input into the initial error prediction model so that the initial error prediction model can optimize the kernel function parameters to obtain an error prediction model that satisfies the error distribution law of the simulation module. The step of determining the sample error corresponding to each sample simulation thickness based on the predetermined error distribution law satisfied by the output results of the simulation module includes: Input the simulated thickness of each sample into the error prediction model, and obtain the sample error corresponding to the simulated thickness of that sample output by the error prediction model.

6. The method according to any one of claims 1-5, characterized in that, The step of determining the thickness uniformity result of copper plating on the circuit board to be manufactured after applying the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be manufactured includes: Based on the corrected copper plating thickness of each region of the circuit board to be manufactured, the dispersion of the corrected copper plating thickness of each region is calculated, and the thickness uniformity result of the copper plating design parameters after copper plating of the circuit board to be manufactured is obtained.

7. The method according to any one of claims 1-5, characterized in that, After determining the thickness uniformity result of copper plating on the circuit board to be manufactured using the copper plating design parameters based on the corrected copper plating thickness of each region of the circuit board to be manufactured, the method further includes: The copper plating design parameters are multiple, and the thickness uniformity result includes the thickness uniformity corresponding to each copper plating design parameter; the copper plating design parameter with the optimal copper plating thickness uniformity is taken as the target copper plating design parameter, and the circuit board is produced based on the structural design parameters included in the target copper plating design parameter, and the produced circuit board is copper plating performed according to the process design parameters included in the target copper plating design parameter; or, If the thickness uniformity result corresponding to the copper plating design parameters does not meet the preset uniformity requirement, the copper plating design parameters, the thickness uniformity result, and the preset optimization algorithm are used to determine candidate copper plating design parameters that meet the parameter constraints corresponding to the copper plating design parameters. The copper plating design parameters of the circuit board to be produced are updated to the candidate copper plating design parameters, and the step of inputting the copper plating design parameters of the circuit board to be produced into the pre-trained thickness prediction model is returned until the thickness uniformity result corresponding to the obtained copper plating design parameters meets the preset uniformity requirement. The obtained copper plating design parameters are used as target copper plating design parameters to produce the circuit board based on the structural design parameters included in the target copper plating design parameters, and the produced circuit board is copper plated according to the process design parameters included in the target copper plating design parameters.

8. A device for predicting the uniformity of copper plating thickness, characterized in that, The device includes: The parameter input module is used to input the copper plating design parameters of the circuit board to be produced into a pre-trained thickness prediction model. The thickness prediction model includes a simulation module and a correction network. The copper plating design parameters include the process design parameters corresponding to the copper plating process and the structural design parameters of each area of ​​the circuit board to be produced. The simulation module is used to simulate the copper plating process of the pre-set circuit board to be produced based on the process design parameters and the structural design parameters, to obtain the simulated thickness of each area in the circuit board to be produced, and to input the simulated thickness into the correction network. The calibration module is used by the calibration network to determine the corrected copper plating thickness of each area in the circuit board to be produced under the copper plating design parameters based on the copper plating design parameters and the simulation thickness, and to output the corrected copper plating thickness. The prediction module is used to determine the thickness uniformity result of copper plating on the circuit board to be manufactured after applying the copper plating design parameters, based on the corrected copper plating thickness of each area of ​​the circuit board to be manufactured.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images