A bipolar plate structure optimization method and system based on flow field data analysis

Abstract

The application discloses a bipolar plate structure optimization method and system based on flow field data analysis, and relates to the technical field of fuel cells. The method comprises the following steps: establishing a parameterized bipolar plate flow channel geometric entity model; performing compression deformation simulation of the gas diffusion layer on the bipolar plate flow channel geometric entity model, obtaining physical field data of the gas diffusion layer, and calculating a porosity distribution field and a local mass transfer impedance coefficient field based on the physical field data; performing multi-physical field coupling simulation based on the porosity distribution field and the local mass transfer impedance coefficient field, and obtaining a local performance contribution degree field; constructing a comprehensive loss function according to the local mass transfer impedance coefficient field and the local performance contribution degree field; taking the minimization of the comprehensive loss function as the target, iteratively optimizing the design variables through an optimization algorithm until the convergence condition is met, and outputting the optimal design variable combination. Through parameterized modeling, multi-physical field coupling simulation and comprehensive loss function evaluation, efficient optimization of the bipolar plate design is realized.

Description

Technical Field

[0001] This invention relates to the field of fuel cell technology, specifically to a method and system for optimizing bipolar plate structures based on flow field data analysis. Background Technology

[0002] With the continuous development of fuel cell technology, bipolar plates, as one of the core components of fuel cells, play a crucial role. Bipolar plates are responsible not only for distributing reactant gases but also for multiple functions such as conductivity, cooling, and structural support. To improve the efficiency and performance of fuel cells, the flow channel design of bipolar plates has become particularly important. Traditional bipolar plate design methods often rely on experience or experimentation, making it difficult to fully consider the coupling effects of multiphysics and their impact on overall performance. In recent years, with advancements in computational fluid dynamics, multiphysics simulation technology, and optimization algorithms, optimization design based on numerical simulation and data-driven methods has become an effective means to improve bipolar plate performance. By establishing accurate bipolar plate flow channel geometric models, performing multiphysics coupling simulations, and combining optimization algorithms, the relationship between design variables and performance can be identified more precisely, achieving efficient and accurate bipolar plate design.

[0003] However, existing technologies still face many challenges, such as how to accurately simulate the impact of assembly clamping forces on the gas diffusion layer, how to efficiently combine multiphysics simulation for performance evaluation, and how to effectively optimize in complex design spaces. While existing technologies have provided some solutions, there is still room for improvement in design accuracy, optimization efficiency, and overall performance. Summary of the Invention

[0004] Based on the shortcomings of the prior art described above, the purpose of this invention is to provide a bipolar plate structure optimization method and system based on flow field data analysis to solve the above-mentioned technical problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a bipolar plate structure optimization method based on flow field data analysis, comprising:

[0006] S1: Establish a parameterized bipolar plate flow channel geometric solid model. The design variables of the bipolar plate flow channel geometric solid model include the flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning parts.

[0007] S2: Apply assembly clamping force to the bipolar plate flow channel geometric solid model to simulate the compression deformation of the gas diffusion layer, obtain the physical field data of the gas diffusion layer, and calculate the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data.

[0008] S3: Based on the porosity distribution field and the local mass transfer impedance coefficient field, perform multi-physics field coupling simulation in the fuel cell to obtain the local performance contribution field of the catalyst layer surface;

[0009] S4: Construct a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field;

[0010] S5: With the goal of minimizing the comprehensive loss function, the design variables are iteratively optimized through an optimization algorithm until the convergence condition is met, and the optimal combination of design variables is output.

[0011] The present invention is further configured such that S1 includes:

[0012] Define design variables corresponding to the geometric features of the flow channel and ribs, including flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning points;

[0013] Set constraints for design variables, including equality constraints to maintain the periodic constancy of flow channel width and rib width, and boundary constraints to limit the physical feasible range of each design variable.

[0014] Based on the design variables and their constraints, a corresponding three-dimensional bipolar plate flow channel geometric solid model is generated through a parametric modeling program.

[0015] The present invention is further configured such that the construction process of the bipolar plate flow channel geometric solid model includes: generating a cross-sectional sketch based on the flow channel width and flow channel depth, sweeping the cross-sectional sketch along a preset path, and performing rounded transition processing at the flow channel turning parts according to the radius of curvature.

[0016] The present invention is further configured such that S2 includes:

[0017] Based on the bipolar plate flow channel geometric solid model, a finite element analysis model is established to simulate the compression deformation of the gas diffusion layer under the action of assembly clamping force.

[0018] In the finite element analysis model, the gas diffusion layer region is meshed and its contact behavior with the bipolar plate is defined, and mechanical boundary conditions corresponding to the assembly clamping force are applied.

[0019] The strain field and contact pressure field of the gas diffusion layer are obtained by solving the finite element analysis model.

[0020] Based on the strain field, the porosity distribution field of the gas diffusion layer under compressed conditions is calculated.

[0021] The local mass transfer impedance coefficient field is calculated based on the porosity distribution field, strain field, and contact pressure field.

[0022] The present invention is further configured such that S3 includes:

[0023] In a multiphysics simulation environment, a coupled simulation model is established, and the porosity distribution field and the local mass transfer impedance coefficient field are defined as the porosity property and mass transfer property of the gas diffusion layer, respectively.

[0024] In the coupled simulation model, inlet boundary conditions, outlet boundary conditions, and electrochemical reaction kinetic boundary conditions are set for the reactant gas and the catalyst layer surface.

[0025] The coupled simulation model is solved to obtain the distribution field of reactant gas concentration and local current density on the surface of the catalyst layer;

[0026] The local performance contribution field is calculated based on the local current density distribution field and the reactant gas concentration distribution field.

[0027] The present invention is further configured such that S4 includes:

[0028] Based on the distribution of the local mass transfer impedance coefficient field on the catalyst layer contact surface, the first performance index reflecting the mass transfer uniformity is calculated.

[0029] Based on the distribution of the local performance contribution field, a second performance index reflecting the overall performance loss is calculated by assigning higher weights to low-performance regions.

[0030] Low-performance regions are identified based on the local performance contribution field, and the area ratio of low-performance regions is calculated to obtain a third performance index that reflects the scale of performance bottlenecks.

[0031] Obtain the flow channel pressure drop as a fourth performance indicator reflecting flow power consumption;

[0032] The first, second, third, and fourth performance indicators are combined according to preset weights to construct a comprehensive loss function.

[0033] The present invention is further configured such that the low-performance region refers to the region where the local performance contribution field is less than a preset threshold.

[0034] The present invention is further configured such that S5 includes:

[0035] Based on the design variables and corresponding comprehensive loss function values ​​of the initial sample set, a Gaussian process surrogate model is constructed to establish the mapping relationship between the design variables and the comprehensive loss function;

[0036] Based on the prediction results of the Gaussian process surrogate model, the optimization potential of unsampled points is evaluated through the acquisition function, and the combination of new design variables with the greatest potential is selected.

[0037] The new design variable combination is substituted into the compression deformation simulation, multiphysics coupling simulation and comprehensive loss function calculation process to obtain its corresponding comprehensive loss function value;

[0038] The newly obtained combination of design variables and their combined loss function values ​​are added to the sample set to update the Gaussian process proxy model.

[0039] Repeat the above selection, calculation and update process until the preset convergence condition is met, and output the combination of design variables that minimizes the comprehensive loss function as the optimal solution.

[0040] The present invention is further configured to generate a corresponding bipolar plate flow channel geometric solid model based on the optimal combination of design variables.

[0041] This invention also provides a bipolar plate structure optimization system based on flow field data analysis, the system comprising:

[0042] Parametric modeling module: Establishes a parametric bipolar plate flow channel geometric solid model. The design variables of the bipolar plate flow channel geometric solid model include the flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning parts.

[0043] Deformation field mapping module: Apply assembly clamping force to the geometric solid model of the bipolar plate flow channel to simulate the compression deformation of the gas diffusion layer, obtain the physical field data of the gas diffusion layer, and calculate the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data.

[0044] Performance field analysis module: Based on the porosity distribution field and the local mass transfer impedance coefficient field, multi-physics field coupling simulation is performed in the fuel cell to obtain the local performance contribution field of the catalyst layer surface;

[0045] Function construction module: Constructs a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field;

[0046] The optimization iteration module aims to minimize the comprehensive loss function. It iteratively optimizes the design variables using an optimization algorithm until the convergence condition is met, and outputs the optimal combination of design variables.

[0047] This invention provides a bipolar plate structure optimization method and system based on flow field data analysis. The method involves: S1: establishing a parameterized bipolar plate flow channel geometric solid model, where design variables include flow channel width, rib width, flow channel depth, and the radius of curvature at flow channel bends; S2: applying assembly clamping force to the bipolar plate flow channel geometric solid model to simulate the compression deformation of the gas diffusion layer, obtaining physical field data of the gas diffusion layer, and calculating the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data; S3: performing multi-physics coupling simulation within the fuel cell based on the porosity distribution field and local mass transfer impedance coefficient field to obtain the local performance contribution field of the catalyst layer surface; S4: constructing a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field; S5: iteratively optimizing the design variables using an optimization algorithm with the goal of minimizing the comprehensive loss function until convergence conditions are met, outputting the optimal combination of design variables. The beneficial effects include:

[0048] 1. By establishing a parameterized bipolar plate flow channel geometric solid model and multiphysics coupling simulation, the design of the bipolar plate can be optimized efficiently. By accurately simulating the compression deformation, porosity distribution and local mass transfer impedance coefficient field of the gas diffusion layer, the mass transfer efficiency of the fuel cell can be improved, the performance of the catalyst layer can be enhanced, and thus the power output and energy efficiency of the overall system can be improved.

[0049] 2. By constructing a comprehensive loss function and combining multiple performance indicators such as local mass transfer impedance coefficient field, local performance contribution field, low-performance region area ratio, and flow channel pressure drop, the performance of bipolar plates can be effectively evaluated comprehensively. This method can reflect the optimization effect of bipolar plates, ensure the balance of mass transfer uniformity, performance distribution, and flow power consumption, thereby improving the reliability of the optimization design.

[0050] 3. The Bayesian optimization method based on the Gaussian process surrogate model and sampling function can quickly evaluate the potential of unsampled design points during the optimization process, select the combination of design variables most likely to bring optimization, reduce unnecessary computation, and improve optimization efficiency; through iterative optimization, it gradually approaches the optimal solution and can achieve good design results with less computational resources.

[0051] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0053] Figure 1 A flowchart illustrating a bipolar plate structure optimization method based on flow field data analysis is shown as an exemplary embodiment of the present invention.

[0054] Figure 2 This is a schematic diagram of a bipolar plate structure optimization system based on flow field data analysis, which is an exemplary embodiment of the present invention. Detailed Implementation

[0055] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0056] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0057] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0058] Example 1:

[0059] A bipolar plate structure optimization method based on flow field data analysis, such as... Figure 1 As shown, it includes:

[0060] S1: Establish a parameterized bipolar plate flow channel geometric solid model. The design variables of the bipolar plate flow channel geometric solid model include the flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning parts.

[0061] S2: Apply assembly clamping force to the bipolar plate flow channel geometric solid model to simulate the compression deformation of the gas diffusion layer, obtain the physical field data of the gas diffusion layer, and calculate the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data.

[0062] S3: Based on the porosity distribution field and the local mass transfer impedance coefficient field, perform multi-physics field coupling simulation in the fuel cell to obtain the local performance contribution field of the catalyst layer surface;

[0063] S4: Construct a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field;

[0064] S5: With the goal of minimizing the comprehensive loss function, the design variables are iteratively optimized through an optimization algorithm until the convergence condition is met, and the optimal combination of design variables is output.

[0065] The present invention is further configured such that S1 includes:

[0066] Define design variables corresponding to the geometric features of the flow channel and ribs, including flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning points;

[0067] Set constraints for design variables, including equality constraints to maintain the periodic constancy of flow channel width and rib width, and boundary constraints to limit the physical feasible range of each design variable.

[0068] Based on design variables and their constraints, a corresponding three-dimensional bipolar plate flow channel geometric solid model is generated through a parametric modeling program. The invention further specifies that the construction process of the bipolar plate flow channel geometric solid model includes: generating a cross-sectional sketch based on the flow channel width and depth, sweeping the cross-sectional sketch along a preset path, and performing rounded corner transition processing at the flow channel turning points according to the radius of curvature. Specifically, the bipolar plate structure optimization method of the invention will be described in detail below using a multi-channel serpentine flow field structure as an example. This embodiment takes a fuel cell bipolar plate with a battery active area of ​​300 cm² and a single flow channel period width of 2.0 mm as the implementation object. First... Parametric modeling is performed, using channel width, rib width, channel depth, and the radius of curvature at channel bends as core geometric design variables. The channel width is the channel groove width, the rib width is the width of the protrusion between adjacent channels, the channel depth is the vertical depth of the channel groove, and the radius of curvature at channel bends is the smooth radius at U-shaped corners. Regarding constraint settings, an equality constraint is set requiring the sum of the channel width and rib width to equal the periodic width of a single channel to ensure periodic channel arrangement. Boundary constraints are also set, including a channel width range of 0.8 mm to 1.2 mm, a channel depth range of 0.5 mm to 1.0 mm, and a bend radius. The lower limit is 0.3 mm; the purpose of setting the above boundary constraints is to balance the manufacturability and mass transfer performance requirements of the bipolar plate: the channel width range is used to ensure the smoothness of the channel and the mechanical strength of the bipolar plate; the channel depth range is used to optimize the balance between the transport of reactant gas and the flow resistance; the lower limit of the radius of curvature is used to prevent flow separation caused by sharp angles; then, a parametric model is constructed, and the application interface of the open-source geometry kernel OpenCASCADE is used to automatically model the model using the Python programming language. The specific construction process includes: first, drawing a rectangular sketch representing the cross-section of a single channel on a two-dimensional plane according to the currently set channel width and channel depth; Then, based on the direction of the serpentine flow field, a continuous centerline composed of straight segments and circular arc segments is drawn in three-dimensional space as the sweep path, where the radius of the circular arc segment is the preset radius of curvature of the flow channel turning part. Next, the created rectangular section is translated and rotated along the defined sweep path to generate a three-dimensional bipolar plate flow channel geometry with a continuous flow channel structure. Finally, at the flow channel turning point, the program automatically fills the edges of the swept entity with rounded corners according to the preset radius of curvature to make the turning point smooth and reduce flow resistance. Finally, the three-dimensional bipolar plate flow channel geometry model containing the complete flow channel and rib structure is output and saved in the standard STEP file format.

[0069] The present invention is further configured such that S2 includes:

[0070] Based on the bipolar plate flow channel geometric solid model, a finite element analysis model is established to simulate the compression deformation of the gas diffusion layer under the action of assembly clamping force.

[0071] In the finite element analysis model, the gas diffusion layer region is meshed and its contact behavior with the bipolar plate is defined, and mechanical boundary conditions corresponding to the assembly clamping force are applied.

[0072] The strain field and contact pressure field of the gas diffusion layer are obtained by solving the finite element analysis model.

[0073] Based on the strain field, the porosity distribution field of the gas diffusion layer under compressed conditions is calculated.

[0074] Based on the porosity distribution field, strain field, and contact pressure field, the local mass transfer impedance coefficient field is calculated. Specifically, this implementation process is based on preset assembly clamping force, initial thickness of the gas diffusion layer, and basic porosity parameters. First, the parameterized bipolar plate flow channel geometric solid model is imported into the finite element analysis environment, and the mechanical properties of the bipolar plate and the gas diffusion layer are defined respectively: the bipolar plate is set as a rigid body, and the gas diffusion layer is set as a deformable body, and its stress-strain relationship data obtained through material experiments is input. Then, mesh generation is performed, the gas diffusion layer region is discretized using hexahedral mesh elements, and the mesh is refined in the contact surface region with the bipolar plate to ensure that the mesh density in this region can accurately analyze the stress-strain gradient. When establishing the contact relationship, the contact properties between the lower surface of the bipolar plate and the upper surface of the gas diffusion layer need to be defined: normal contact is set as a hard contact that does not allow penetration, and tangential contact is set as a penalty function contact model that considers the friction effect. When applying mechanical boundary conditions, all degrees of freedom of the bipolar plate are completely constrained, and a uniformly distributed pressure load calculated based on the total assembly clamping force and the area of ​​action is applied to the upper surface of the gas diffusion layer. After completing the settings, submit the static analysis calculation. After the solution is completed, extract the strain field and contact pressure field distribution data of the gas diffusion layer. Then, calculate and map the material property field. Based on the obtained strain field data, use the post-processing function of the finite element software to calculate the porosity distribution field after compression. The calculation is based on the following physical relationship: the current porosity distribution field at each point is obtained by subtracting the product of the strain field value at that point and the experimentally calibrated compression coefficient from the initial porosity, thus reasonably reflecting the influence of the spatial distribution of strain on porosity. Then, write a Python script to call the post-processing application interface of the finite element software to read the porosity distribution field, normal strain field and contact pressure field, and calculate the local mass transfer impedance coefficient distribution field based on the mass transfer impedance theory model that considers the coupling effect of the three. This calculation process performs iterative calculations on each element node of the gas diffusion layer mesh model. This model quantitatively characterizes the influence of the porosity distribution field on the tortuosity of the diffusion path, the correction of the effective diffusion distance by the normal strain field and the contribution of the contact pressure field to the densification effect of the medium, and finally generates a local mass transfer impedance coefficient distribution field that reflects the ease of gas transport at each spatial point.

[0075] The present invention is further configured such that S3 includes:

[0076] In a multiphysics simulation environment, a coupled simulation model is established, and the porosity distribution field and the local mass transfer impedance coefficient field are defined as the porosity property and mass transfer property of the gas diffusion layer, respectively.

[0077] In the coupled simulation model, inlet boundary conditions, outlet boundary conditions, and electrochemical reaction kinetic boundary conditions are set for the reactant gas and the catalyst layer surface.

[0078] The coupled simulation model is solved to obtain the distribution field of reactant gas concentration and local current density on the surface of the catalyst layer;

[0079] Based on the local current density distribution field and the reactant gas concentration distribution field, the local performance contribution field is calculated. Specifically, a coupled simulation model is established in a multiphysics simulation environment. The porosity distribution field and the local mass transfer impedance coefficient field are defined as the porosity attribute and mass transfer attribute of the gas diffusion layer, respectively. The porosity distribution field is used to control the permeability parameter in the Darcy's law interface, while the local mass transfer impedance coefficient field is used to dynamically modulate the effective diffusion coefficient in the rare substance transfer interface. The coupled simulation model achieves bidirectional coupling through three core physical interfaces: rare substance transfer, Darcy's law, and secondary current distribution. The substance concentration affects the reaction rate on the Butler-Volmer electrochemical reaction kinetic boundary, and the reaction rate determines the local current density distribution. In terms of boundary condition settings, the inlet is set as the velocity inlet calculated based on the working current density and stoichiometry, the outlet is set as the pressure outlet of ambient atmospheric pressure, and the catalyst layer surface is set with a Butler-Volmer electrochemical boundary containing experimentally determined kinetic parameters. In the solution process, a fully coupled solver is used and a step-by-step loading strategy is implemented to enhance convergence, i.e., the flow field and concentration outside the electrochemical reaction are solved first. After obtaining a stable initial solution for the degree field, the complete electrochemical reaction is activated for fully coupled solution. A preset tolerance threshold and a maximum iteration limit are used as convergence criteria. Convergence is determined when the residual curve shows a stable downward trend and falls below the preset tolerance threshold. After obtaining the reactive gas concentration distribution field and local current density distribution field on the catalyst layer surface, the local performance contribution field is calculated through the following process: First, the local current density value and reactive gas concentration value of each grid cell on the catalyst layer surface are extracted. The average current density of all cells is calculated as the performance benchmark value, and the reactive gas concentration at the inlet of the flow channel is set as the reactant supply benchmark concentration. Then, normalization is performed on each grid cell. The ratio of the cell's local current density to the average current density is defined as the current density ratio, and the ratio of the cell's local reactive gas concentration to the inlet benchmark concentration is defined as the concentration ratio. The product of these two is used as the local performance contribution of the cell, reflecting the efficiency of the local region's contribution to the overall battery performance. Finally, the local performance contribution of all cells is mapped to the geometric model of the catalyst layer surface to generate a spatially continuous local performance contribution distribution field.

[0080] The present invention is further configured such that S4 includes:

[0081] Based on the distribution of the local mass transfer impedance coefficient field on the catalyst layer contact surface, the first performance index reflecting the mass transfer uniformity is calculated.

[0082] Based on the distribution of the local performance contribution field, a second performance index reflecting the overall performance loss is calculated by assigning higher weights to low-performance regions.

[0083] Low-performance regions are identified based on the local performance contribution field, and the area ratio of low-performance regions is calculated to obtain a third performance index that reflects the scale of performance bottlenecks.

[0084] Obtain the flow channel pressure drop as a fourth performance indicator reflecting flow power consumption;

[0085] The first, second, third, and fourth performance indicators are combined according to preset weights to construct a comprehensive loss function. The invention further specifies that the low-performance region refers to the region where the local performance contribution field is less than a preset threshold. Specifically, the construction of the comprehensive loss function includes the calculation and weighted integration of four performance indicators. First, based on the local mass transfer impedance coefficient field, the arithmetic mean and standard deviation of all points on the catalyst layer contact surface are calculated. The ratio of the standard deviation to the mean is used as the first performance indicator; the lower the value of this first performance indicator, the better the mass transfer uniformity. Second, based on the local performance contribution field, a second performance indicator reflecting the comprehensive performance loss is constructed: a differentiated weighting coefficient is set, assigning a higher weighting coefficient to low-performance regions where the local performance contribution field is lower than the preset threshold, and assigning a baseline weighting coefficient to the remaining regions. All grid cells on the catalyst layer surface are traversed, and the performance loss value of each cell (defined as 1 minus the local performance contribution of that cell) is calculated. The square of this performance loss value, the corresponding weighting coefficient of the cell, and the cell's weighting coefficient are then used to calculate the performance loss. Multiplying the areas yields the weighted performance loss of the unit. Summing the weighted performance losses of all units and dividing by the total area of ​​the catalyst layer yields a second performance index. This second performance index focuses on evaluating system bottlenecks by emphasizing the contribution of low-performance regions. A third performance index is calculated: the ratio of the total area of ​​low-performance regions to the total area of ​​the catalyst layer. This ratio quantifies the spatial scale of the performance bottleneck. The low-performance region area is obtained through the following process: marking all low-performance units based on a preset threshold, then using a region growing algorithm to identify connected regions (merging all face-adjacent or edge-adjacent low-performance units), and finally summing the areas of all connected regions. Normalizing the flow channel pressure drop by dividing it by a preset reference pressure drop yields a fourth performance index reflecting flow power consumption. These four performance indices, representing mass transfer uniformity, overall performance loss, bottleneck scale, and flow power consumption respectively, are weighted according to the optimization objective and summed to form a comprehensive loss function. A smaller comprehensive loss function value indicates better overall performance of the bipolar plate flow field structure.

[0086] The present invention is further configured such that S5 includes:

[0087] Based on the design variables and corresponding comprehensive loss function values ​​of the initial sample set, a Gaussian process surrogate model is constructed to establish the mapping relationship between the design variables and the comprehensive loss function;

[0088] Based on the prediction results of the Gaussian process surrogate model, the optimization potential of unsampled points is evaluated through the acquisition function, and the combination of new design variables with the greatest potential is selected.

[0089] The new design variable combination is substituted into the compression deformation simulation, multiphysics coupling simulation and comprehensive loss function calculation process to obtain its corresponding comprehensive loss function value;

[0090] The newly obtained combination of design variables and their combined loss function values ​​are added to the sample set to update the Gaussian process proxy model.

[0091] Repeat the above selection, calculation, and update process until the preset convergence condition is met, and output the combination of design variables that minimizes the comprehensive loss function as the optimal solution. Specifically, this embodiment uses a Bayesian optimization framework for automatic optimization. The optimization process starts with a preset number of initial samples, such as 20 initial samples, and sets convergence conditions, such as when the relative improvement of the comprehensive loss function value of the optimal solution in a preset number of consecutive iterations is lower than a preset threshold. First, the Latin hypercube sampling method is used to generate 20 initial samples in the design variable space. Each sample includes the channel width, rib width, and channel depth. A specific combination of degree and radius of curvature is used; these sample points are sequentially substituted into the complete physical simulation process of steps S1 to S4 above, and parametric modeling, compression deformation simulation, multiphysics coupling simulation, and comprehensive loss function calculation are performed respectively to obtain the comprehensive loss function value corresponding to each set of design variables, forming an initial training sample set; a Gaussian process surrogate model is constructed based on the initial sample set, which establishes the mapping relationship from design variables to comprehensive loss function through probabilistic statistical methods; during model training, the maximum likelihood estimation method is used to automatically adjust the kernel function parameters so that the model can both fit the function values ​​of known sample points and reasonably predict future results. The uncertainty of sampling points; the trained Gaussian process surrogate model has the ability to quickly predict the loss function value corresponding to any new design variable; then it enters the active learning loop stage. Based on the prediction results of the current Gaussian process surrogate model, the expected improvement sampling function is used to search for the most valuable evaluation point in the entire design space. This sampling function comprehensively considers the magnitude of the predicted comprehensive loss function value and the degree of prediction uncertainty, and prioritizes those regions that have the potential to obtain a lower comprehensive loss function value but also have greater uncertainty. The numerical optimization algorithm finds the combination of new design variables that maximizes the sampling function value; the selected new design variables are substituted into the complete physical simulation process to perform high-fidelity simulation calculations and obtain their corresponding comprehensive loss function values; the newly obtained design variables and their comprehensive loss function values ​​are added to the training sample library, and the parameters of the Gaussian process surrogate model are updated, so that the model gradually improves the recognition accuracy of the target function landscape; the loop process of "sampling function optimization - physical simulation verification - model update" is repeated. When the relative improvement in the preset number of iterations is lower than the preset threshold, it indicates that the optimization has reached a stable state; finally, the combination of design variables corresponding to the minimum comprehensive loss function value obtained in the entire optimization process is output as the optimal bipolar plate structure parameters.

[0092] The present invention is further configured to generate a corresponding bipolar plate flow channel geometric solid model based on the optimal combination of design variables. Specifically, after obtaining the optimal combination of design variables, a parametric modeling program is first run, and the optimal parameter values ​​are passed as input to a pre-developed Python script. The script calls the application programming interface of the open-source geometry kernel OpenCASCADE to automatically execute the three-dimensional geometry reconstruction process: generating a rectangular cross-section sketch based on the flow channel width and depth, performing a sweep operation according to a serpentine path, applying a specified radius of curvature at the turning points to perform rounded corner transitions, and generating a three-dimensional bipolar plate flow channel geometric solid model containing a complete flow channel and rib structure.

[0093] Example 2:

[0094] Please see Figure 2 This exemplary bipolar plate structure optimization system based on flow field data analysis includes:

[0095] Parametric modeling module: Establishes a parametric bipolar plate flow channel geometric solid model. The design variables of the bipolar plate flow channel geometric solid model include the flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning parts.

[0096] Deformation field mapping module: Apply assembly clamping force to the geometric solid model of the bipolar plate flow channel to simulate the compression deformation of the gas diffusion layer, obtain the physical field data of the gas diffusion layer, and calculate the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data.

[0097] Performance field analysis module: Based on the porosity distribution field and the local mass transfer impedance coefficient field, multi-physics field coupling simulation is performed in the fuel cell to obtain the local performance contribution field of the catalyst layer surface;

[0098] Function construction module: Constructs a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field;

[0099] The optimization iteration module aims to minimize the comprehensive loss function. It iteratively optimizes the design variables using an optimization algorithm until the convergence condition is met, and outputs the optimal combination of design variables.

[0100] It should be noted that the bipolar plate structure optimization system based on flow field data analysis provided in the above embodiments and the bipolar plate structure optimization method based on flow field data analysis provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the bipolar plate structure optimization system based on flow field data analysis provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A bipolar plate structure optimization method based on flow field data analysis, characterized in that, include: S1: Establish a parameterized bipolar plate flow channel geometric solid model. The design variables of the bipolar plate flow channel geometric solid model include the flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning parts. S2: Apply assembly clamping force to the bipolar plate flow channel geometric solid model to simulate the compression deformation of the gas diffusion layer, obtain the physical field data of the gas diffusion layer, and calculate the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data. S3: Based on the porosity distribution field and the local mass transfer impedance coefficient field, perform multi-physics field coupling simulation in the fuel cell to obtain the local performance contribution field of the catalyst layer surface; S4: Construct a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field; S5: With the goal of minimizing the comprehensive loss function, the design variables are iteratively optimized through an optimization algorithm until the convergence condition is met, and the optimal combination of design variables is output.

2. The bipolar plate structure optimization method based on flow field data analysis according to claim 1, characterized in that, S1 includes: Define design variables corresponding to the geometric features of the flow channel and ribs, including flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning points; Set constraints for design variables, including equality constraints to maintain the periodic constancy of flow channel width and rib width, and boundary constraints to limit the physical feasible range of each design variable. Based on the design variables and their constraints, a corresponding three-dimensional bipolar plate flow channel geometric solid model is generated through a parametric modeling program.

3. The bipolar plate structure optimization method based on flow field data analysis according to claim 2, characterized in that, The construction process of the bipolar plate flow channel geometric solid model includes: generating a cross-sectional sketch based on the flow channel width and flow channel depth, sweeping the cross-sectional sketch along a preset path, and performing rounded transition processing at the flow channel turning points according to the radius of curvature.

4. The bipolar plate structure optimization method based on flow field data analysis according to claim 1, characterized in that, S2 includes: Based on the bipolar plate flow channel geometric solid model, a finite element analysis model is established to simulate the compression deformation of the gas diffusion layer under the action of assembly clamping force. In the finite element analysis model, the gas diffusion layer region is meshed and its contact behavior with the bipolar plate is defined, and mechanical boundary conditions corresponding to the assembly clamping force are applied. The strain field and contact pressure field of the gas diffusion layer are obtained by solving the finite element analysis model. Based on the strain field, the porosity distribution field of the gas diffusion layer under compressed conditions is calculated. The local mass transfer impedance coefficient field is calculated based on the porosity distribution field, strain field, and contact pressure field.

5. The bipolar plate structure optimization method based on flow field data analysis according to claim 1, characterized in that, S3 includes: In a multiphysics simulation environment, a coupled simulation model is established, and the porosity distribution field and the local mass transfer impedance coefficient field are defined as the porosity property and mass transfer property of the gas diffusion layer, respectively. In the coupled simulation model, inlet boundary conditions, outlet boundary conditions, and electrochemical reaction kinetic boundary conditions are set for the reactant gas and the catalyst layer surface. The coupled simulation model is solved to obtain the distribution field of reactant gas concentration and local current density on the surface of the catalyst layer; The local performance contribution field is calculated based on the local current density distribution field and the reactant gas concentration distribution field.

6. The bipolar plate structure optimization method based on flow field data analysis according to claim 1, characterized in that, S4 includes: Based on the distribution of the local mass transfer impedance coefficient field on the catalyst layer contact surface, the first performance index reflecting the mass transfer uniformity is calculated. Based on the distribution of the local performance contribution field, a second performance index reflecting the overall performance loss is calculated by assigning higher weights to low-performance regions. Low-performance regions are identified based on the local performance contribution field, and the area ratio of low-performance regions is calculated to obtain a third performance index that reflects the scale of performance bottlenecks. Obtain the flow channel pressure drop as a fourth performance indicator reflecting flow power consumption; The first, second, third, and fourth performance indicators are combined according to preset weights to construct a comprehensive loss function.

7. The bipolar plate structure optimization method based on flow field data analysis according to claim 6, characterized in that, The low-performance region refers to the region where the local performance contribution field is less than a preset threshold.

8. The bipolar plate structure optimization method based on flow field data analysis according to claim 1, characterized in that, S5 includes: Based on the design variables and corresponding comprehensive loss function values ​​of the initial sample set, a Gaussian process surrogate model is constructed to establish the mapping relationship between the design variables and the comprehensive loss function; Based on the prediction results of the Gaussian process surrogate model, the optimization potential of unsampled points is evaluated through the acquisition function, and the combination of new design variables with the greatest potential is selected. The new design variable combination is substituted into the compression deformation simulation, multiphysics coupling simulation and comprehensive loss function calculation process to obtain its corresponding comprehensive loss function value; The newly obtained combination of design variables and their combined loss function values ​​are added to the sample set to update the Gaussian process proxy model. Repeat the above selection, calculation and update process until the preset convergence condition is met, and output the combination of design variables that minimizes the comprehensive loss function as the optimal solution.

9. The bipolar plate structure optimization method based on flow field data analysis according to claim 8, characterized in that, Based on the optimal combination of design variables, the corresponding bipolar plate flow channel geometric solid model is generated.

10. A bipolar plate structure optimization system based on flow field data analysis, used to implement the bipolar plate structure optimization method based on flow field data analysis as described in any one of claims 1-9, characterized in that, include: Parametric modeling module: Establishes a parametric bipolar plate flow channel geometric solid model. The design variables of the bipolar plate flow channel geometric solid model include the flow channel width, rib width, flow channel depth, and radius of curvature of the flow channel turning parts. Deformation field mapping module: Apply assembly clamping force to the geometric solid model of the bipolar plate flow channel to simulate the compression deformation of the gas diffusion layer, obtain the physical field data of the gas diffusion layer, and calculate the porosity distribution field and local mass transfer impedance coefficient field based on the physical field data. Performance field analysis module: Based on the porosity distribution field and the local mass transfer impedance coefficient field, multi-physics field coupling simulation is performed in the fuel cell to obtain the local performance contribution field of the catalyst layer surface; Function construction module: Constructs a comprehensive loss function based on the distribution of the local mass transfer impedance coefficient field and the local performance contribution field; The optimization iteration module aims to minimize the comprehensive loss function. It iteratively optimizes the design variables using an optimization algorithm until the convergence condition is met, and outputs the optimal combination of design variables.

Images