Integrated preparation method of solid-state supercapacitor device considering bearing and energy storage

Abstract

This invention provides an integrated fabrication method for solid-state supercapacitor devices that combine load carrying and energy storage. The method includes the following steps: fabricating a planar supercapacitor; using COMSOL simulation to model the performance of the planar supercapacitor under different shape changes; then, employing machine learning coupled with finite element analysis to obtain the relationship between molds of different feature dimensions and the supercapacitor forming results under different loads, guiding the construction of the supercapacitor forming mold; finally, placing the planar supercapacitor and a metal plate in the forming mold for hydroforming and assembly. This invention's fabrication method is simple to operate, convenient for forming control, optimizes the supercapacitor fabrication process, expands the supercapacitor forming methods, and has promising application prospects.

Description

Technical Field

[0001] This invention relates to the field of supercapacitor technology, specifically to an integrated fabrication method for solid-state supercapacitor devices that combines load carrying and energy storage. The method utilizes hydroforming, machine learning, and finite element-assisted mold design to achieve integrated fabrication of solid-state supercapacitor devices. Background Technology

[0002] Supercapacitors, as a new type of high-efficiency and environmentally friendly energy storage element, possess superior performance in terms of energy density, safety, and charge-discharge cycle life. With sufficient technological potential and market prospects, they will be widely used in new energy vehicles, regenerative braking, and energy recovery and utilization, becoming one of the important technological means for energy conservation, environmental protection, and new energy utilization. To realize the large-scale production of supercapacitors, breakthroughs and innovations in their fabrication and molding processes are needed, moving towards lower costs, higher efficiency, green environmental protection, and simpler processes.

[0003] Compared to planar supercapacitors, supercapacitors with three-dimensional morphology not only fulfill energy storage requirements but also provide structural support, thus improving space utilization. A method for fabricating solid-state supercapacitors with complex morphology is urgently needed. Summary of the Invention

[0004] The purpose of this invention is to propose an integrated fabrication method for solid-state supercapacitor devices that combines load carrying and energy storage, so as to obtain supercapacitors with complex morphological characteristics.

[0005] To address the above problems, the present invention provides the following technical solution:

[0006] An integrated fabrication method for solid-state supercapacitor devices that combines load carrying and energy storage is disclosed. This method is used to manufacture supercapacitors with complex three-dimensional morphology. The method involves combining a prepared solid electrolyte with electrodes to synthesize a planar supercapacitor with good recyclability. The method uses machine learning to guide the molding die and adopts a step-by-step molding process. After the metal plate and the planar supercapacitor are respectively hydraulically molded, a solid-state supercapacitor device that combines load carrying and energy storage is synthesized.

[0007] Furthermore, the preparation method includes the following specific steps:

[0008] 1) Fabricate a planar supercapacitor with good recyclability, ensuring that the areal capacitance of the undeformed solid-state supercapacitor remains essentially unchanged compared to the deformed supercapacitor, with a variation not exceeding 0.03 F / cm². 2 ;

[0009] 2) Establish a finite element simulation model of the mold and the entire flat-plate supercapacitor;

[0010] 3) Create a dataset:

[0011] Using a finite element simulation model of a mold, we conduct finite element forming analysis under different loads and mold features. We extract the ID of the cell and the xyz coordinate information def_x, def_y, def_z of each cell node, as well as the corresponding material thickness STH after forming, from the finite element forming analysis results. This forms a data subset for machine learning under different mold features and loads of the same mold.

[0012] Data sets are formed by combining subsets of data from different types of molds; the type of mold is related to the three-dimensional shape of the solid-state supercapacitor device required by the target; the mold features include inflection points, chamfers, depth, and radius.

[0013] 4) Establish a machine learning model, train the machine learning model using the dataset, and obtain an AI model that reflects the thickness distribution of the supercapacitor molding. The input of the AI ​​model is the xyz coordinates of each cell node and the external load, and the output is the thickness of the supercapacitor. The mapping relationship between the mold features under the action of external load and the thickness distribution of the supercapacitor molding is obtained through training.

[0014] 5) Determine and model the initial mold based on the 3D shape of the target solid-state supercapacitor device. Obtain the ID of each cell in the modeled mold and the xyz coordinate information (def_x, def_y, def_z) of each cell node, as well as the external load data. Input these into the AI ​​model to obtain the thickness distribution of the solid-state supercapacitor device under different loads in the current mold. Compare this with the actual thickness distribution of the target solid-state supercapacitor device. If the requirements are met, the mold is confirmed to be qualified, and the external load situation at this time is recorded. If the requirements are not met and local damage or excessive thinness occurs, record the location of the local damage or excessive thinness and adjust the mold features until the requirements are met.

[0015] The requirements are determined based on the actual capacitor demand, ensuring that the capacitors are free from damage, breakage, and excessively thin areas. Excessively thin areas refer to areas no more than 50% of the average thickness, or areas with a thickness no more than 0.1 mm.

[0016] 6) Use the mold obtained in step 5) to perform step-by-step hydraulic forming: stack two metal plates and place them between the mold and the pressure ring for forming. For molds without draft, an auxiliary plate needs to be added between the two metal plates for forming at the same time, so that there is a certain gap between the two metal plates in the vertical direction.

[0017] After the metal plate is formed, the upper metal plate and auxiliary plate are removed, leaving the lower metal plate in the mold. A flat supercapacitor is placed in the mold, and the hydraulic forming process is carried out according to the external load conditions recorded in step 5). The supercapacitor after hydraulic forming is assembled with the metal plate to obtain the target solid-state supercapacitor device.

[0018] The fabrication of the planar supercapacitor requires the preparation of a PVA / NaCl solid electrolyte and a substrate-free CNT / PTFE electrode.

[0019] The PVA / NaCl solid electrolyte is prepared by mixing NaCl solution with a mixture of polyvinyl alcohol (PVA) and deionized water. Deionized water is added to the polyvinyl alcohol (PVA) while stirring, and the mixture is stirred at a constant temperature magnetic stirrer at 90°C for 1 hour. After the stirring is completed, the mixture is cooled to room temperature. The NaCl solution is slowly added dropwise to the mixture of deionized water and polyvinyl alcohol, and the mixture is placed in a constant temperature magnetic stirrer and stirred at 90°C for another hour to form a semi-transparent gel electrolyte.

[0020] The substrate-free CNT / PTFE electrode is composed of a mixture of carbon nanotubes, acetylene, and polytetrafluoroethylene in a ratio of 4:1:1. The three materials are measured, mixed, and ground with an appropriate amount of deionized water to obtain a slurry. The slurry is then blown until semi-dry, heated, and rolled for 15 minutes. Weighing paper is then applied to the rolled electrode, a small amount of deionized water is dripped onto it, transferred to the weighing paper, cut, and dried in a 60°C oven for 12 hours.

[0021] The fabrication process of the planar capacitor is as follows: CNT / PTFE electrodes are cut, leaving the tabs. The prepared electrolyte is evenly coated onto the graphite electrode plates, placed in a drying oven and dried until semi-dry. The electrolyte-coated sides of the two plates are then placed together and placed in the drying oven to continue drying until the solid electrolyte is completely solidified.

[0022] Preferably, a double-layer supercapacitor model of "porous carbon-solid electrolyte-porous carbon" was established, and COMSOL simulation was used to analyze the changes in electrochemical performance after deformation of the solid supercapacitor blank. In this invention, after preparing the planar supercapacitor, COMSOL simulation was used to analyze the changes in electrochemical performance of the planar supercapacitor before and after deformation during the hydraulic forming process. After confirming that it has good recyclability, the subsequent steps were carried out.

[0023] The dataset is generated from simulation data produced by finite element analysis. The AI ​​model can predict the thickness distribution of the part after hydroforming by analyzing the shape features of the mold and the magnitude of the load during the forming process. The digital model (AI model) is a mold model that has undergone feature-based processing, that is, a model obtained by converting the physical information of the mold, such as its shape, structure, and features, into digital information that can be processed by a computer. The AI ​​model can be loaded into any modeling software to form a new digital model of the mold. The modeling software here can be SolidWorks, UG, finite element software, etc.

[0024] In step 6), if a liquid medium is used for hydroforming, an insulating film is placed over the surface of the flat-plate supercapacitor to prevent the liquid medium from leaking out; if a gaseous medium is used for hydroforming, blow molding is employed. Since flexible electrode and electrolyte materials are easily deformable, blow molding technology can be used during hydroforming.

[0025] In step 6), if hydraulic forming is performed using a liquid medium, an insulating film is placed over the surface of the flat-plate supercapacitor to prevent leakage of the liquid medium; if hydraulic forming is performed using a gaseous medium, blow molding is used. Since flexible electrodes and electrolyte materials are easily deformable, blow molding technology can be used during hydraulic forming. Preferably, applying a lubricant or lubricating oil to the workpiece surface can reduce the coefficient of friction between it and the tool to a certain extent.

[0026] The machine learning model is at least one of linear regression, resilient network regression, random forest, SVM, and gradient boosting decision tree; preferably, it is a random forest model. More preferably, a random forest model is used for training.

[0027] After step 5), the mold determined in step 5) is finally verified using finite element simulation software to obtain a more accurate mold model for step 6) to perform step-by-step hydraulic forming.

[0028] The construction process of the mold finite element simulation model described in step 2 is as follows: the metal plate, the blank holder, and the mold all adopt a conical shell model, and the thickness of the blank is given by creating a material interface; the deformation of the mold and the blank holder is not considered, and both the mold and the blank holder are defined as rigid bodies; the mold is fixed in the initial step and the dynamic analysis step, the blank holder is fixed in the initial step, and it is subjected to a blank holder force of 150N in the dynamic analysis step; according to the characteristics of hydraulic forming, the load of the liquid medium on the plate is distributed on the inner side of the blank holder and perpendicular to the surface of the plate, and the load application value is set to linear, rising to 40MPa within 1s; the assembled flat plate supercapacitor is regarded as a single-layer membrane structure that is integrally bound, the thickness of the single-layer membrane structure is set, and the overall material parameters are given according to the comprehensive performance of the two electrode materials; the contact conditions and friction coefficient are set, and a high-precision three-dimensional nonlinear dynamic finite element model is established using ABAQUS finite element software as the mold finite element simulation model to analyze the material deformation and stress distribution after hydraulic forming.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] This invention provides a method for fabricating a non-planar supercapacitor, which offers significant advantages compared to conventional supercapacitor manufacturing processes. It utilizes hydroforming to obtain a solid-state supercapacitor device that combines energy storage and load-bearing functions. Furthermore, compared to traditional stamping and welding techniques, hydroforming effectively improves forming accuracy and reduces costs. Simultaneously, the mold structure is simple and flexible in design, allowing for the fabrication of supercapacitors with various complex morphologies, providing a new approach to supercapacitor fabrication. The organic combination of finite element simulation and machine learning leverages their combined advantages to achieve a closed-loop iteration from data to model to prediction, improving R&D efficiency and product performance, reducing R&D risks, and providing a novel solution for supercapacitor product innovation.

[0031] The present invention utilizes a distributed molding method to obtain a supercapacitor with complex three-dimensional morphological features. The lower metal plate provides clamping and support for the supercapacitor, ensuring structural stability while achieving the desired shape. In this invention, hydroforming can utilize liquid or gas as the force transmission medium, uniformly applying the load to the surface of the part to be formed. This effectively improves the forming accuracy of the sheet metal and allows for precise control of the material's shape and dimensions. Attached Figure Description

[0032] Figure 1 This is a physical image of the planar solid-state supercapacitor in this invention;

[0033] Figure 2 The above are the COMSOL simulation results of the supercapacitor model in this invention;

[0034] Figure 3 This is a schematic diagram illustrating the machine learning-assisted molding process in the present invention.

[0035] Figure 4 This is a diagram of the ABAQUS finite element model of an embodiment of the present invention;

[0036] Figure 5 This is a curve showing the fitting between the AI ​​model's prediction results and the actual results in this invention.

[0037] Figure 6 This is a schematic diagram of the step-by-step molding process of the solid-state supercapacitor of the present invention. Detailed Implementation

[0038] The present invention will be further described below with reference to specific embodiments.

[0039] Figure 3 To facilitate machine learning in the construction of the forming mold, the initial mold is first determined, and feature modeling is performed. Relevant data is input into an AI model trained using finite element analysis results as a database to predict the forming process and determine if the capacitor is damaged or too thin. If damage or thinning is found, the model is modified; otherwise, further finite element simulations are performed to confirm the accuracy of the predictions. If successful, the mold construction is complete. A dataset is built using extensive ABAQUS simulation data, and an AI model is trained that directly reflects the thickness distribution of the supercapacitor during forming. Based on factors such as the mold's thickness, shape, and chamfer data, the modeling of any step directly reflects the thickness distribution of the supercapacitor after forming, avoiding issues such as battery damage, stress concentration, and poor performance at the tip. This allows for fine-tuning of the mold features, assisting in the construction of the hydroforming mold.

[0040] Example 1:

[0041] This invention provides an integrated fabrication method for a solid-state supercapacitor device that combines load carrying and energy storage, comprising the following specific steps: fabricating a planar supercapacitor; using COMSOL simulation to analyze the deformation of the planar supercapacitor during the hydroforming process and the changes in its electrochemical performance after deformation; establishing a dataset of applied external loads using ABAQUS simulation data and actual production data, training a machine learning model to obtain an AI model that directly reflects the thickness distribution of the supercapacitor during forming to assist in constructing the hydroforming mold for the supercapacitor; placing the planar supercapacitor and a metal plate respectively in the hydroforming mold for hydroforming; and finally assembling the formed metal plate and the formed supercapacitor to synthesize a supercapacitor with a complex morphology.

[0042] 1) Fabrication of planar supercapacitors:

[0043] a) Preparation of PVA / NaCl solid electrolyte: Weigh 0.6g of polyvinyl alcohol (PVA), add 50mL of deionized water while stirring with a glass rod, seal the beaker with plastic wrap, place it in a thermostatic magnetic stirrer, and stir at a constant temperature for 1h at 90℃ until the white granules (PVA) are completely dissolved; after stirring, cool to room temperature; mix 3.5g of NaCl with 10mL of deionized water, stir thoroughly until the NaCl is completely dissolved, and then slowly add the NaCl solution dropwise to the mixture of deionized water and polyvinyl alcohol, place it in a thermostatic magnetic stirrer, and continue stirring at 90℃ for 1h to finally form a translucent gel electrolyte, which is the PVA / NaCl solid electrolyte.

[0044] b) Preparation of substrate-free CNT / PTFE electrode: The three materials were measured in a ratio of carbon nanotubes: acetylene: polytetrafluoroethylene = 4:1:1, mixed, and ground with an appropriate amount of deionized water to obtain a slurry. The slurry was dried with a hair dryer until semi-dry, placed on an electric heating platform, and rolled evenly with a glass rod for about 15 minutes (if the electrode surface becomes dry, rough, cracked, or powdery, it means that the material is too dry. If it is too dry, add deionized water to neutralize it; if it is too wet, heat it slightly. This is to improve the dispersion uniformity of the material, increase the packing density, and reduce the powdering effect of the capacitor during the cycling process). The rolled electrode was obtained. Weighing paper was placed on the rolled electrode, and a small amount of deionized water was dripped on it to help maintain the shape, cleanliness, and stability of the sample. The electrode film was transferred to the weighing paper, the irregular edges were trimmed, and the electrode was dried in a 60°C oven for 12 hours to obtain a substrate-free CNT / PTFE electrode.

[0045] c) Cut the electrodes prepared in step b) to a size of 20*20 mm, leaving the tabs. Evenly coat the translucent gel electrolyte prepared in step a) onto the CNT / PTFE electrodes, place them in a drying oven until semi-dry, then place two electrolyte-coated electrodes face to face together and continue drying in the drying oven until the solid electrolyte is completely solidified. The finished product is shown below. Figure 1 As shown.

[0046] d) Simulation of electrochemical performance of planar supercapacitor before and after deformation:

[0047] In this embodiment, a double-layer supercapacitor model of "porous carbon-solid electrolyte-porous carbon" is established, and the electrochemical performance changes of the solid supercapacitor after deformation are analyzed using COMSOL simulation. Figure 2 Figures (a)-(e) show the cyclic volt-ampere characteristic curves of the solid-state supercapacitor (volt-ampere characteristics under different loads of 0 MPa, 5 MPa, 10 MPa, 15 MPa, and 20 MPa). It can be seen that the CV curves under different loads do not exhibit significant distortion and are all close to rectangular, indicating that the supercapacitor has good reversibility and ideal capacitance performance. Figure 2 As shown in (f), the areal capacitance of the undeformed solid-state supercapacitor remains essentially unchanged compared to that of the supercapacitor subjected to deformation, with a variation not exceeding 0.03 F / cm². 2 This indicates that the planar supercapacitor has good recyclability and can be used as a material for subsequent molding.

[0048] 2) Establish a finite element simulation model of the mold and the overall flat-plate supercapacitor.

[0049] The metal plate, blank holder, and mold all adopt a conical shell model, and the thickness of the blank is assigned by creating a material interface. The focus is on analyzing the relevant performance parameters of the metal plate and the planar supercapacitor during the forming process, without considering the deformation of the mold and blank holder; therefore, both the mold and blank holder are defined as rigid bodies. The mold is fixed in the initial step and the dynamic analysis step, and the blank holder is fixed in the initial step, but is subjected to a blank holder force of 150N in the dynamic analysis step. Based on the characteristics of hydroforming, the load of the liquid medium on the plate is distributed inside the blank holder and perpendicular to the plate surface. The load application is set to be linear, rising to 40MPa within 1 second. The assembled planar supercapacitor is considered as a single-layer film structure (both electrodes are of the same material), with a thickness set to 0.2. The overall material parameters are assigned using the electrode material parameters to simplify the simulation calculation. If the two electrode materials are different, the overall material parameters are assigned using the combined properties of the two electrode materials, including elastic modulus, Poisson's ratio, density, and yield stress. The contact conditions and friction coefficient are set according to the material properties. In this embodiment, the contact condition is designated as "punish," and the friction coefficient is set to 0.11. A high-precision three-dimensional nonlinear dynamic finite element model is established using ABAQUS finite element software, such as... Figure 4 As shown, the analysis examines the material deformation and stress distribution after hydroforming.

[0050] 3) Create a dataset:

[0051] Finite element modeling was used to conduct finite element forming analysis under different loads and mold features. The cell IDs, x, y, and z coordinates of each cell node, and the corresponding material thickness STH after forming were extracted from the finite element analysis results, i.e., the data "def_x", "def_y", "def_z", and "STH". This resulted in a subset of data for machine learning under different mold features and loads for the same mold.

[0052] Data cleaning is performed. In the ABAQUS simulation results, if the mold is a flat-bottomed cylinder, the upper part of the flat-bottomed cylinder has a large flat surface, resulting in a large amount of data. Moreover, the thickness does not change much before and after molding. Deleting the data from the part of the flat-bottomed cylinder with a large flat surface can shorten the training time and improve the accuracy of the model.

[0053] The performance of molds under different loads varies due to processing techniques. Assuming the mold remains constant, the effects of thickness and external loads are considered. Following the same approach, models with different thicknesses and external loads are trained under a fixed mold. Different loads are then applied to the established finite element simulation model to perform molding, and load data is added.

[0054] Data sets are formed by combining subsets of data from different types of molds; the type of mold is related to the three-dimensional shape of the solid-state supercapacitor device required by the target (such as flat-bottomed cylindrical mold, concave-bottomed cylindrical mold, stepped mold, etc., the shape of the mold determines the shape of the capacitor after molding); the mold features are inflection points, chamfers, depth and radius.

[0055] 4) Establish a machine learning model, train the machine learning model using the dataset, and obtain an AI model that reflects the thickness distribution of the supercapacitor molding. The input of the AI ​​model is the xyz coordinates of each cell node and the external load, and the output is the thickness of the supercapacitor. The mapping relationship between the mold features under the action of external load and the thickness distribution of the supercapacitor molding is obtained through training.

[0056] During training, the data is split into a training set and a test set, with the training set accounting for 80% of the dataset and the test set accounting for 20% of the dataset, in order to prevent overfitting and improve the model's performance and generalization ability.

[0057] Feature values ​​are scaled to a similar range, typically between 0 and 1 or between -1 and 1. This avoids the effect of feature values ​​exceeding orders of magnitude having an excessive impact on the model.

[0058] The machine learning model can be linear regression, elastic regression network, random forest, SVM, gradient boosting decision tree, etc. Multivariate linear regression models are mainly used to consider data correlation issues. Single coordinate variables are relatively concentrated, but do not have a clear linear relationship with thickness. Using multivariate linear regression models, a multivariate regression model can be found. ( ) = 2 1 + 2 2 + ... + The parameter 0 is used to determine the relationship between the three coordinates, load and thickness. X represents the variable, w represents the weight, and n represents the number of variables. In the example, f represents the thickness, and x represents the coordinates and load.

[0059] Five models—linear regression, elastic network regression, random forest, SVM, and gradient boosting decision tree—were evaluated, and the random forest model was selected for training because it had the smallest error.

[0060] A machine learning model is trained using the training set to obtain the final AI model for the thickness distribution of the direct reaction supercapacitor.

[0061] The input consists of the xyz coordinates of each cell node, the external load, and the thickness of the supercapacitor. The training process obtains the mapping relationship between the mold features (inflection points, chamfers, depth, and radius) under external load and the thickness of the supercapacitor.

[0062] After training, the accuracy of the model was tested using the test set. The fitting curve is shown in Figure 5, and the accuracy is 0.973.

[0063] 5) Determine and model the initial mold based on the 3D shape of the target solid-state supercapacitor device. Obtain the ID of each cell in the modeled mold and the xyz coordinate information (def_x, def_y, def_z) of each cell node, as well as the external load data. Input these into the AI ​​model to obtain the thickness distribution of the solid-state supercapacitor device under different loads in the current mold as the prediction result. If the prediction result indicates that the capacitor is damaged or too thin (less than 0.1mm), record the location of the local damage or thinning and adjust the mold features until the requirements are met. If the prediction result is qualified, record the external load condition at this time and perform finite element simulation to confirm whether the mold is qualified. If qualified, produce the mold.

[0064] 6) Use the mold obtained in step 5) to perform step-by-step hydroforming (e.g. Figure 6 (As shown): The metal plate (load-bearing) and the planar supercapacitor are formed separately. First, two metal plates are stacked and placed between the mold and the pressure ring for forming. For molds without draft, an auxiliary plate needs to be added between the two metal plates during forming to create a vertical gap between them, allowing space for the electrodes and electrolyte materials. Additionally, applying lubricant or lubricating oil to the workpiece surface can reduce the coefficient of friction between it and the tool to some extent. After the metal plate is formed, the upper metal plate and auxiliary plate are removed, leaving the lower metal plate in the mold. The planar supercapacitor is then placed in, and the solid-state supercapacitor with a complex morphology is formed according to the hydroforming steps. Since the electrodes and electrolyte cannot come into contact with liquid, if a liquid medium is used, an insulating film must be covered on the upper surface of the supercapacitor to prevent liquid leakage. Because flexible electrode and electrolyte materials are easily deformed, blow molding technology can be used during hydroforming, and the liquid medium can be replaced with a gas medium to solve the problem of the liquid medium wetting the electrode and electrolyte materials. Finally, the supercapacitor and metal plate are bonded together with insulating adhesive to form a finished supercapacitor with a complex three-dimensional morphology.

[0065] The above embodiments are only used to illustrate and not limit the technical solutions of the invention. Although the above embodiments have described the invention in detail, those skilled in the art should understand that modifications or equivalent substitutions can be made to the invention, but any modifications and partial substitutions that do not depart from the spirit and scope of the invention should be covered within the scope of the claims of the invention.

Claims

1. A method for integrated preparation of a solid-state supercapacitor device with both load bearing and energy storage, characterized in that, The prepared solid electrolyte and electrodes are combined to form a planar supercapacitor with good recycling performance. The molding mold is guided by machine learning and a step-by-step molding process is adopted. The metal plate and the planar supercapacitor are respectively hydraulically molded to synthesize a solid supercapacitor device that combines load carrying and energy storage. The preparation method includes the following specific steps: 1) Fabricate a planar supercapacitor with good recyclability, ensuring that the areal capacitance of the undeformed solid-state supercapacitor remains essentially unchanged compared to the deformed supercapacitor, with a variation not exceeding 0.03 F / cm². 2 ; 2) Establish a finite element simulation model of the mold and the entire flat-plate supercapacitor; 3) Create a dataset: Using a finite element simulation model of a mold, we conduct finite element forming analysis under different loads and mold features. We extract the ID of the cell and the xyz coordinate information def_x, def_y, def_z of each cell node, as well as the corresponding material thickness STH after forming, from the finite element forming analysis results. This forms a data subset for machine learning under different mold features and loads of the same mold. Data subsets of different types of molds are combined to form a dataset; The mold features include inflection points, chamfers, depth, and radius; 4) Establish a machine learning model, train the machine learning model using the dataset, and obtain an AI model that directly reflects the thickness distribution of the supercapacitor molding. The input of the AI ​​model is the xyz coordinates of each cell node and the external load, and the output is the thickness of the supercapacitor. The mapping relationship between the mold features under the action of external load and the thickness distribution of the supercapacitor molding is obtained through training. 5) Determine and model the initial mold based on the 3D shape of the target solid-state supercapacitor device. Obtain the ID of each cell in the modeled mold and the xyz coordinate information (def_x, def_y, def_z) of each cell node, as well as the external load data. Input these into the AI ​​model to obtain the thickness distribution of the solid-state supercapacitor device under different loads in the current mold. Compare this with the actual thickness distribution of the target solid-state supercapacitor device. If the requirements are met, the mold is confirmed to be qualified, and the external load situation at this time is recorded. If the requirements are not met and local damage or excessive thinness occurs, record the location of the local damage or excessive thinness and adjust the mold features until the requirements are met. 6) Use the mold obtained in step 5) to perform step-by-step hydraulic forming: stack two metal plates and place them between the mold and the pressure ring for forming. For molds without draft, an auxiliary plate needs to be added between the two metal plates for forming at the same time, so that there is a certain gap between the two metal plates in the vertical direction. After the metal plate is formed, the upper metal plate and auxiliary plate are removed, the lower metal plate is left in the mold, a flat supercapacitor is placed in it, and the hydraulic forming process is carried out according to the external load conditions recorded in step 5). The hydraulically formed flat supercapacitor is assembled with the metal plate to obtain a target solid-state supercapacitor device with complex morphological features.

2. The production method according to claim 1, characterized by, In step 6), if a liquid medium is used for hydraulic forming, an insulating film is covered on the upper surface of the flat supercapacitor to prevent the liquid medium from flowing out; if a gas medium is used for hydraulic forming, blow molding is used for preparation.

3. The preparation method according to claim 1, characterized in that, The machine learning model is at least one of linear regression, elastic network regression, random forest, SVM, and gradient boosting decision tree.

4. The method of claim 1, wherein, After step 5), the mold determined in step 5) is finally verified using finite element simulation software to obtain a more accurate mold model for step 6) to perform step-by-step hydraulic forming.

5. The preparation method according to claim 1, characterized in that, The construction process of the mold finite element simulation model described in step 2 is as follows: the metal plate, the blank holder, and the mold all adopt a conical shell model, and the thickness of the blank is given by creating a material interface; the deformation of the mold and the blank holder is not considered, and both the mold and the blank holder are defined as rigid bodies; the mold is fixed in the initial step and the dynamic analysis step, the blank holder is fixed in the initial step, and it is subjected to a blank holder force of 150N in the dynamic analysis step; according to the characteristics of hydraulic forming, the load of the liquid medium on the plate is distributed on the inner side of the blank holder and perpendicular to the plate surface, and the load application value is set to linear, rising to 40MPa within 1s; the assembled flat plate supercapacitor is regarded as a single-layer membrane structure that is integrally bound, the thickness of the single-layer membrane structure is set, and the overall material parameters are given according to the comprehensive performance of the two electrode materials; the contact conditions and friction coefficient are set, and a high-precision three-dimensional nonlinear dynamic finite element model is established using ABAQUS finite element software as the mold finite element simulation model to analyze the material deformation and stress distribution after hydraulic forming.

6. The preparation method according to claim 1, characterized in that, The types of molds include flat-bottomed cylindrical molds, concave-bottomed cylindrical molds, and stepped molds.

Images