Optimization design method and system for printed circuit board heat exchanger airfoil fins

Abstract

The present invention belongs to the technical field related to the optimization of printed circuit board heat exchangers. It discloses a method and system for optimizing the airfoil fin of a printed circuit board heat exchanger, comprising: (1) parametric modeling of the airfoil fin PCHE; (2) dimensionless parameters of the airfoil profile control points and the shrinkage proportional factor of the cross section in the fin are used as optimization variables, and the value range is determined; (3) design test points are generated using an experimental design method, and a sample set is obtained through CFD calculation; (4) SVR prediction models for heat transfer coefficient and pressure drop are respectively constructed, and hyperparameter optimization is performed to optimize the SVR prediction model; (5) with the heat transfer coefficient and pressure drop as optimization targets, a multi-objective optimization algorithm is used to optimize the optimization variables in the design space based on the optimized SVR prediction model to construct a Pareto optimal frontier; (6) an optimization solution is obtained from the Pareto optimal frontier using a decision method. The present invention can effectively improve the efficiency of optimization design.

Description

Technical Field

[0001] The present invention belongs to the technical field related to the optimization design of printed circuit board type heat exchangers, and more specifically, relates to an optimization design method and system for airfoil fins of a printed circuit board heat exchanger. Background Art

[0002] The supercritical CO2 Brayton cycle has broad application prospects in a wide range of fields, including high-temperature reactors in fourth-generation nuclear power plants, new gas turbines, and high-temperature solar power generation systems. As a key component of the supercritical CO2 Brayton cycle, the performance of the heat exchanger is directly related to the efficiency and safety of the entire system.

[0003] The printed circuit heat exchanger (PCHE) is a new type of high-efficiency, compact heat exchanger manufactured using chemical etching and diffusion welding techniques. This eliminates the need for internal joints and welds in the heat exchanger core, resulting in mechanical properties identical to the original material, withstanding high temperatures, high pressures, and corrosion resistance. Based on these characteristics, the PCHE is considered the most promising intermediate heat exchanger in the supercritical carbon dioxide Brayton cycle. PCHE channel structures can be divided into two categories: continuous and discontinuous. Continuous channels primarily include straight, zigzag, and wavy channels, while discontinuous channels primarily include S-shaped fins and airfoil-shaped fins. Airfoil-shaped PCHEs have become a research focus due to their superior flow and heat transfer performance.

[0004] Currently, there are few studies on the optimization of PCHE of airfoil fins. Existing research mainly focuses on the influence of the arrangement of existing fins and different fin cross-sectional shapes on the flow and heat transfer performance. There is no overall optimization research on the airfoil fin profile and the three-dimensional shape changes along the height direction.

[0005] The most common optimization method for airfoil fins within PCHE channels is to conduct comparative studies using experimental or numerical simulation methods, comparing different existing airfoil fin arrangements (horizontal spacing, vertical spacing, and staggered spacing) with pre-defined fixed fin cross-sectional shapes (rectangular, circular, elliptical, airfoil, and swordfish). Airfoil shape significantly influences the flow and heat transfer of PCHEs. Existing studies on airfoil shape optimization fail to fully consider the variations in the airfoil profile and three-dimensional shape along the height direction, thus limiting the effectiveness of optimizing the heat transfer performance of airfoil fins in PCHEs. Summary of the Invention

[0006] In response to the above defects or improvement needs of the prior art, the present invention provides an optimization design method and system for the airfoil fins of a printed circuit board heat exchanger to solve the problem of optimizing the three-dimensional shape of the airfoil fins, aiming to improve the flow and heat transfer performance of the airfoil fin PCHE.

[0007] To achieve the above objectives, according to one aspect of the present invention, a method for optimizing the design of airfoil fins of a printed circuit board heat exchanger is provided, the method mainly comprising the following steps:

[0008] (1) Construct the airfoil profile and change the cross-section size of the airfoil fin, perform parametric modeling on the airfoil fin by lofting, and then establish a three-dimensional geometric model of the airfoil fin PCHE;

[0009] (2) Non-dimensionalizing the geometric parameters of the airfoil fin and determining the optimal design variables and their value ranges; the optimal design variables include the shrinkage proportional factor of the cross section of the airfoil fin and the non-dimensionalized parameters of the airfoil profile control points;

[0010] (3) The experimental design method is used to generate design test point samples with different optimization variable parameter level combinations. The airfoil fin PCHE with the corresponding geometric structure of each design test point sample is modeled and the heat transfer coefficient h and pressure drop ΔP of the airfoil fin PCHE under given working conditions are obtained through CFD calculation, thereby obtaining a sample set;

[0011] (4) Preprocess the sample set, construct and train the SVR prediction model with heat transfer coefficient h and pressure drop ΔP as output, and use the grid search method with cross-validation to perform hyperparameter optimization to optimize the SVR prediction model;

[0012] (5) Based on the optimized SVR prediction model, maximizing the heat transfer coefficient h and minimizing the pressure drop ΔP are used as optimization objectives. A multi-objective optimization algorithm is used to optimize the design variables in the design space to construct the Pareto optimal frontier.

[0013] (6) Use decision-making methods to obtain optimization solutions from the Pareto optimal frontier and then determine the optimal optimization solution.

[0014] Furthermore, a non-uniform B-spline curve is used to construct the airfoil profile. The two-dimensional profile of the airfoil fin is modified by changing the coordinates of the airfoil profile control points. At the same time, the proportional size of the cross-section in the airfoil fin can be adjusted by changing the shrinkage factor. The three-dimensional structure of the airfoil fin is formed by cross-section lofting.

[0015] Furthermore, the airfoil chord length L c As the characteristic size, the airfoil profile control point P i (xi,y i )’s coordinates are dimensionless:

[0016]

[0017] At the same time, the shrinkage ratio factor is introduced to measure the ratio of the cross section of the airfoil fin to the upper and lower cross sections.

[0018] Furthermore, the optimized Latin square hypercube experimental design method is used to sample a set number of design test points with different optimization variable parameter level combinations from the multidimensional variable space composed of the optimized design variables. The airfoil fin PCHE with the corresponding geometric structure of each design test point is modeled, and CFD calculations are performed under given operating conditions to obtain the corresponding heat transfer coefficient h and pressure drop △P, thereby obtaining a sample set; the sample consists of optimized design variables and performance evaluation indicators, and the performance evaluation indicators include the heat transfer coefficient h and pressure drop △P.

[0019] Furthermore, the sample set data is normalized, and the normalization calculation formula is:

[0020]

[0021] Where y i Represents the original data, y min ,y max Represent the minimum and maximum values ​​of each column of data respectively, and y represents the normalized data in the range [0,1].

[0022] Furthermore, the sample set is divided into a training set and a test set, and SVR prediction models with heat transfer coefficient h and pressure drop △P as output are established and trained respectively. The K-fold cross-validation grid search method is used to traverse and optimize the hyperparameters of the SVR prediction model. For the two hyperparameters of the penalty factor C and the bandwidth coefficient γ of the RBF kernel function in the SVR prediction model, multiple levels are selected to form multiple groups of hyperparameters. Each group of hyperparameters is substituted into the SVR prediction model for training. Each training iteration is K times. In each iteration, the training set is evenly divided into K groups, of which 1 group is used as a validation set to test the accuracy of the SVR prediction model, and the remaining K-1 groups are used as training data to train the SVR prediction model. The model performance of each group of hyperparameters is evaluated on the validation set, and the average value of the K iteration results is obtained. Finally, the optimized SVR prediction model is obtained by comparing and selecting the best hyperparameter combination (C, γ); the prediction performance of the trained SVR model for the heat transfer coefficient h and pressure drop △P of the airfoil fin PCHE in the test set and the generalization ability of the new samples are tested, and the relative error error, mean square error MSE and correlation coefficient R are introduced. 2 As an evaluation indicator, the expression is:

[0023]

[0024]

[0025]

[0026] Where y i,SVR Represents the predicted value of the SVR model, y i,CFD Represents the CFD calculated value, and N represents the number of test set samples.

[0027] Furthermore, in step (4), the optimization problem is expressed as:

[0028] Objective function:

[0029] Constraints: X i ∈[0,1 / 3],Y i ∈[0.05,0.15],SF∈[0.5,1]

[0030] Where SF is the shrinkage factor; △P is the pressure drop; h is the heat transfer coefficient; (X i ,Y i ) is the airfoil profile control point P i Coordinates, i=1,2,3.

[0031] Furthermore, the multi-objective optimization algorithm is a multi-objective genetic algorithm, a multi-objective evolutionary algorithm or a multi-objective particle swarm algorithm; the decision-making method is a hierarchical analysis method, a superior-inferior solution distance method or a clustering algorithm.

[0032] Furthermore, the K-Medoids center point clustering algorithm is introduced to find the optimization solution corresponding to the representative solution from the Pareto optimal frontier.

[0033] The present invention also provides an optimization design system for the airfoil fins of a printed circuit board type heat exchanger. The optimization design system includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it executes the optimization design method for the airfoil fins of the printed circuit board type heat exchanger as described above.

[0034] In general, compared with the prior art, the above technical solutions conceived by the present invention provide a method and system for optimizing the design of airfoil fins for a printed circuit board heat exchanger, which has the following beneficial effects:

[0035] 1. The present invention overcomes the defect of not being able to obtain the optimal fin shape by pre-setting a fixed cross-sectional shape of the fin and ignoring the change of the fin along the height direction by comprehensively optimizing the airfoil fin profile and the three-dimensional shape along the height direction.

[0036] 2. The present invention adopts the SVR prediction model to predict the performance parameters of each generation of population, and optimizes the SVR prediction model to improve its prediction accuracy. It can replace a large number of numerical simulation calculation processes, save computing resources, shorten the optimization research and development cycle, and improve the efficiency of the optimization process.

[0037] 3. This invention optimizes the optimization variables using a multi-objective genetic algorithm, obtaining a Pareto frontier. A clustering algorithm is then employed to identify representative solutions, making the resulting optimized structure more convincing and effectively improving the overall performance of the heat exchanger. In practical applications, the optimal design under the corresponding conditions can be selected based on the priorities and weights of different sub-objectives, ensuring that each sub-objective meets the requirements as much as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic flow chart of an optimization design method for airfoil fins of a printed circuit board type heat exchanger provided by the present invention;

[0039] Figure 2 It is a schematic diagram of the geometric model of the airfoil fin PCHE;

[0040] Figure 3 (a) and (b) are schematic diagrams of the airfoil fin profile and three-dimensional shape parameterization, respectively;

[0041] Figure 4 (a), (b), (c), and (d) are schematic diagrams of the SVR training and testing prediction results of the heat transfer coefficient h and pressure drop △P, respectively;

[0042] Figure 5 It is the Pareto optimal frontier diagram of heat transfer coefficient h and pressure drop △P;

[0043] Figure 6 Cluater A, Cluater B, Cluater C, Cluater D, and Cluater E are schematic diagrams of the three-dimensional shapes of the airfoil fins corresponding to the cluster center solutions. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0045] See also Figure 1 and Figure 2The present invention provides an optimization design method for airfoil fins of a printed circuit board type heat exchanger, the design method mainly comprising the following steps:

[0046] Step 1: Use non-uniform B-spline curves to construct airfoil profiles. Modify the two-dimensional profiles of the airfoil fins by changing the coordinates of the airfoil profile control points. At the same time, adjust the proportional size of the cross-section in the airfoil fins by changing the shrinkage scale factor. The three-dimensional structure of the airfoil fins is formed by cross-section lofting.

[0047] In this embodiment, due to the periodicity and symmetry of the flow and to save computing resources, this embodiment adopts a single channel composed of five minimum periodic structures along the fluid flow direction as the heat exchange section. The fins are completely staggered. To eliminate the influence of the inlet effect, inlet and outlet sections are set at both ends of the heat exchange section. SolidWorks is used to establish a three-dimensional model. The airfoil fin parameter description is as follows: Figure 3 As shown, the airfoil profile is constructed using non-uniform B-spline curves, and the airfoil fin L c The chord length is 4mm, by changing the airfoil profile control point P i (x i ,y i )(i=1,2,3) coordinates are used to modify the airfoil fin profile. At the same time, the cross-sectional size of the airfoil fin can be changed to achieve the three-dimensional shape change of the airfoil fin. The shrinkage scale factor (SF) is introduced to measure the proportion of the cross-sectional size of the airfoil fin relative to the upper and lower cross-sectional sizes. The three-dimensional structure of the airfoil fin is formed by cross-sectional lofting.

[0048] Step 2: Non-dimensionalize the geometric parameters of the airfoil and determine the optimized design variables and their value ranges; the optimized design variables include the shrinkage factor of the cross section in the airfoil fin and the dimensionless parameters of the airfoil profile control points.

[0049] Among them, the airfoil chord length L c As the characteristic size, the airfoil profile control point P i (x i ,y i )(i=1,2,…m) dimensionless coordinates:

[0050]

[0051] The selected optimization variables are the dimensionless coordinates of the airfoil profile control points As well as the airfoil section shrinkage factor (SF), in order to avoid the intersection of control points, the X coordinate range is obtained by dividing the airfoil chord length at equal intervals.

[0052] In this embodiment, the airfoil chord length L cAs the characteristic size, the airfoil profile control point P i (x i ,y i )(i=1,2,3) dimensionless coordinates, where the airfoil fin L c The string length is fixed at 4mm:

[0053]

[0054] The optimization variables selected in this embodiment are the airfoil profile control point P i (X i ,Y i )(i=1,2,3) and the airfoil section shrinkage factor (SF). To avoid the intersection of control points, the X-coordinate range is obtained by dividing the airfoil chord length by equal intervals. The optimization variables and their value ranges are shown in Table 1.

[0055] Table 1 Optimization variable selection and value range

[0056]

[0057] Step three, using the experimental design method to sample a set number of design test points with different optimization variable parameter level combinations from the multidimensional variable space composed of the optimization design variables, modeling the airfoil fin PCHE with the corresponding geometric structure of each design test point, performing CFD calculations under given operating conditions, and obtaining the heat transfer coefficient h and pressure drop △P of the airfoil fin PCHE, thereby obtaining a sample set; the optimization variables and performance evaluation indicators together constitute the sample set, and the performance evaluation indicators include the heat transfer coefficient h and pressure drop △P.

[0058] Among them, the experimental design methods include full factorial design method, orthogonal design method, optimized Latin hypercube design method, etc.

[0059] In this embodiment, an optimized Latin hypercube experimental design method is used to obtain 300 experimental design samples from the multidimensional variable space composed of optimization variables. The airfoil fin PCHE with the corresponding geometric structure of each sample point is modeled, and CFD calculations are performed under given operating conditions to obtain two performance evaluation indicators of the heat transfer coefficient h and pressure drop △P of the airfoil fin PCHE. The optimization variables and performance evaluation indicators together constitute the sample set.

[0060] For this embodiment, supercritical carbon dioxide (S-CO2) is used as the working fluid for calculation, and the relevant physical parameters of S-CO2 are obtained by calling the NIST database; the shear stress transport (SST) k-ω model is selected as the turbulence model. The inlet boundary condition is set to the mass flow inlet, the mass flow rate is determined by the inlet Reynolds number Re, and the inlet temperature is 380K; the outlet is set to the pressure outlet, and the outlet pressure is 7.73MPa; the two sides of the channel are set to symmetrical boundary conditions, and the upper and lower walls and the airfoil fin surfaces are set to constant wall temperature conditions, and the wall temperature is 430K. The SIMPLEC algorithm is used to solve the coupled equations of pressure and velocity, and the second-order upwind scheme is used for the discretization settings of turbulent kinetic energy, turbulent dissipation rate, and energy. When the residuals of the continuity equation, momentum equation, and energy equation are all less than 10 -6 , and the residual value tends to be stable, and the monitored outlet mass flow rate and temperature remain unchanged, the numerical calculation is considered to have converged.

[0061] Step 4: Preprocess the sample set and divide it into training and test sets. SVR prediction models with heat transfer coefficient h and pressure drop ΔP as output are constructed and trained respectively. A grid search method with cross-validation is used to optimize hyperparameters to optimize the SVR prediction model and improve the model prediction accuracy. The optimized SVR prediction model is then tested to verify its generalization ability to new samples.

[0062] It includes the following sub-steps:

[0063] Step 4.1) Data preprocessing: Normalize the sample set data. The normalization calculation formula is:

[0064]

[0065] Where y i Represents the original data, y min ,y max Represent the minimum and maximum values ​​of each column of data respectively, and y represents the normalized data in the range [0,1].

[0066] Step 4.2) Construct an SVR prediction model: Use the optimized variables normalized in step 4.1) as input data and the performance evaluation index as output data, and use the support vector regression (SVR) machine to build a prediction model.

[0067] Step 4.3) Use the grid search method with cross-validation to optimize the hyperparameters of the SVR prediction model: To obtain the optimal hyperparameter combination (C, γ) of the penalty factor C and the bandwidth coefficient γ of the RBF kernel function in the SVR prediction model, a grid search algorithm with K-fold cross-validation is used to traverse and search for the optimal result, that is, iterate K times. In each iteration, the training dataset is evenly divided into K groups, of which one group is used as validation data to test the performance of the prediction model, and the remaining K-1 groups are used as training data to train the prediction model.

[0068] Step 4.4) Test the performance of the SVR prediction model: In order to test the prediction performance of the trained SVR prediction model for the test set samples and its generalization ability for new samples, the relative error (error), mean square error (MSE) and correlation coefficient (R) are introduced. 2 ) is used as the evaluation index, and the expression is as follows:

[0069]

[0070]

[0071]

[0072] Where y i,SVR Represents the predicted value of the SVR prediction model, y i,CFD represents the CFD calculation value, and N represents the number of test set samples. The smaller the error and MSE values ​​are, the better the R 2 The closer the value is to 1, the higher the prediction accuracy of the prediction model.

[0073] In this embodiment, the 270 training set samples and 30 test set samples obtained by experimental design and CFD calculation are used as the training set and test set of the SVR prediction model, respectively. The grid search algorithm with 5-fold cross-validation is used to obtain the optimal hyperparameter combination (C, γ) of the penalty factor C and the bandwidth coefficient γ of the RBF kernel function in the SVR prediction model, and the optimized SVR prediction model is constructed. The comparison of the prediction results of the SVR prediction model with the CFD calculation results in the training set and test set is shown in Figure 2. Figure 4 As shown in the figure, the mean square error (MSE) of the heat transfer coefficient h and the pressure drop △P are 2.01×10 -4 and 1.68×10 -4 , correlation coefficient R 2The results are 0.995092 and 0.993102, respectively. This result shows that the prediction results of the SVR prediction model and the CFD calculation results are highly correlated. Therefore, it is proved that the constructed SVR prediction model is accurate and feasible for predicting the heat transfer coefficient h and pressure drop ΔP of PCHE. At the same time, in order to verify the generalization ability of the SVR prediction model of this embodiment for new samples, the prediction results of the SVR prediction model on the test set samples were compared with the CFD calculation results. The maximum relative error corresponding to the heat transfer coefficient h is 1.5%, and the mean square error (MSE) is 1.35×10 -4 , correlation coefficient R 2 is 0.993711. Figure 4 It can be obtained that the maximum relative error corresponding to the voltage drop △P is 5.8%, and the mean square error MSE is 1.24×10 -4 , correlation coefficient R 2 The SVR prediction model constructed in this example has high prediction accuracy and generalization ability, and can be used to predict the flow and heat transfer performance of airfoil fin PCEH.

[0074] Step 5: Based on the optimized SVR prediction model, with maximizing the heat transfer coefficient h and minimizing the pressure drop △P as the optimization objectives, a multi-objective optimization algorithm is used to optimize the optimization design variables in the design space to construct the Pareto optimal frontier.

[0075] Multi-objective optimization algorithms include multi-objective genetic algorithm (MOGA), multi-objective evolutionary algorithm (MOEA), multi-objective particle swarm optimization (MOPSO), etc.

[0076] The optimization objectives of this embodiment are the heat exchanger heat transfer coefficient h and pressure drop ΔP, seeking to maximize the heat transfer coefficient h and minimize the pressure drop ΔP, and selecting the airfoil profile control point P i (X i ,Y i )(i=1,2,3) and the airfoil cross-sectional shrinkage factor (SF) are used as optimization variables. The SVR prediction model for heat transfer coefficient h and pressure drop ΔP obtained from the above training is used as the objective function. The optimization problem can be described as follows:

[0077] Objective function:

[0078] Constraints: X i ∈[0,1 / 3],Y i ∈[0.05,0.15],SF∈[0.5,1]

[0079] This example uses NSGA-II for multi-objective optimization design. The parameters are set as follows: population size of 100, maximum number of generations of 500, crossover probability of 0.9, mutation probability of 0.1, and termination criteria set when the maximum number of generations is reached or when the population fitness converges and stagnates. Finally, a Pareto optimal front is constructed, and the K-Medoids center point clustering algorithm is introduced to identify optimization solutions corresponding to representative solutions within the Pareto optimal front.

[0080] Step six: Use decision-making methods to obtain optimization solutions from the Pareto optimal frontier, and then determine the optimal optimization solution.

[0081] Decision-making methods include hierarchical analysis method, top-to-bottom solution distance method (TOPSIS method), clustering algorithm, etc.

[0082] Figure 5 The Pareto optimal frontier obtained by multi-objective optimization can be seen. The optimized Pareto optimal solution set forms a concave curve. Each data point on the curve represents a feasible optimization scheme, which is the result of continuous reconciliation of the heat transfer coefficient h and the pressure drop △P. The K-medoids clustering method is used to obtain 5 cluster center points from the Pareto optimal frontier, representing 5 different optimization schemes. The corresponding design variable parameters are shown in Table 2 and Figure 6 . In addition, in order to evaluate the accuracy of the multi-objective optimization method proposed in this embodiment, the airfoil fin PCHE with geometric structural parameters corresponding to the five cluster center points was modeled and numerically simulated respectively, and the predicted values ​​of the SVR prediction model and the CFD numerical simulation results were analyzed and compared. The results are shown in Table 3. It can be seen that the errors are basically within 5%. From this, it can be judged that the predicted values ​​of the SVR prediction model are in good agreement with the calculated values ​​of the numerical simulation, which proves that the optimization design method adopted in this embodiment is feasible. It can be seen from the multi-objective optimization results that in terms of heat transfer performance, the Cluster E scheme has the highest heat transfer coefficient h; in terms of flow characteristics, the Cluster A scheme has the lowest pressure drop △P; and the flow heat transfer performance of Cluster BD is between the two. In the actual design process, the required optimization scheme can be selected according to the requirements of heat transfer performance and flow characteristics.

[0083] Table 2 Cluster center design variables

[0084]

[0085]

[0086] Table 3 Cluster center objective function

[0087]

[0088] The present invention also provides an optimization design system for the airfoil fins of a printed circuit board type heat exchanger. The optimization design system includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it executes the optimization design method for the airfoil fins of the printed circuit board type heat exchanger as described above.

[0089] The present invention also provides a computer-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned optimized design method for the airfoil fins of the printed circuit board heat exchanger.

[0090] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing the design of airfoil fins for a printed circuit board heat exchanger, characterized in that: The method comprises the following steps: (1) Construct the airfoil profile and change the cross-section size of the airfoil fin, perform parametric modeling on the airfoil fin by lofting, and then establish a three-dimensional geometric model of the airfoil fin PCHE; (2) Non-dimensionalizing the geometric parameters of the airfoil fin and determining the optimal design variables and their value ranges; the optimal design variables include the shrinkage proportional factor of the cross section of the airfoil fin and the non-dimensionalized parameters of the airfoil profile control points; (3) The experimental design method is used to generate design test point samples with different optimization variable parameter level combinations. The airfoil fin PCHE with the corresponding geometric structure of each design test point sample is modeled and the heat transfer coefficient h and pressure drop ΔP of the airfoil fin PCHE under given working conditions are obtained through CFD calculation, thereby obtaining a sample set; (4) Preprocess the sample set, construct and train the SVR prediction model with heat transfer coefficient h and pressure drop ΔP as output, and use the grid search method with cross-validation to perform hyperparameter optimization to optimize the SVR prediction model; (5) Based on the optimized SVR prediction model, maximizing the heat transfer coefficient h and minimizing the pressure drop ΔP are used as optimization objectives. A multi-objective optimization algorithm is used to optimize the design variables in the design space to construct the Pareto optimal frontier. (6) Use decision-making methods to obtain optimization solutions from the Pareto optimal frontier and then determine the optimal optimization solution.

2. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to claim 1, wherein: The airfoil profile is constructed using non-uniform B-spline curves. The two-dimensional profile of the airfoil fin is modified by changing the coordinates of the airfoil profile control points. At the same time, the proportion of the cross-section in the airfoil fin can be adjusted by changing the shrinkage factor. The three-dimensional structure of the airfoil fin is formed by cross-section lofting.

3. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to claim 1, wherein: The chord length of the airfoil is L c As the characteristic size, the airfoil profile control point P i (x i ,y i )’s coordinates are dimensionless: Where i = 1, 2, ... m; At the same time, the shrinkage ratio factor is introduced to measure the ratio of the cross section of the airfoil fin to the upper and lower cross sections.

4. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to claim 1, wherein: The optimized Latin square hypercube experimental design method is used to sample a set number of design test points with different optimization variable parameter level combinations from the multidimensional variable space composed of the optimization design variables. The PCHE of the airfoil fin with the corresponding geometric structure of each design test point is modeled. CFD calculations are performed under given operating conditions to obtain the corresponding heat transfer coefficient h and pressure drop ΔP, thus obtaining a sample set. The sample consists of optimized design variables and performance evaluation indicators, and the performance evaluation indicators include heat transfer coefficient h and pressure drop △P.

5. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to claim 1, wherein: The sample set data is normalized, and the normalization calculation formula is: Where y i Represents the original data, y min ,y max Represent the minimum and maximum values ​​of each column of data respectively, and y represents the normalized data in the range [0,1].

6. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to claim 5, wherein: The sample set is divided into a training set and a test set, and SVR prediction models with heat transfer coefficient h and pressure drop △P as output are established and trained respectively. The K-fold cross-validation grid search method is used to traverse and optimize the hyperparameters of the SVR prediction model. For the two hyperparameters of the penalty factor C and the bandwidth coefficient γ of the RBF kernel function in the SVR prediction model, multiple levels are selected to form multiple groups of hyperparameters. Each group of hyperparameters is substituted into the SVR prediction model for training. Each training iteration is K times. In each iteration, the training set is evenly divided into K groups, of which one group is used as a validation set to test the accuracy of the SVR prediction model, and the remaining K-1 groups are used as training data to train the SVR prediction model. The model performance of each group of hyperparameters is evaluated on the validation set, and the average value of the K iteration results is obtained. Finally, the optimal hyperparameter combination (C, γ) is selected by comparison to obtain the optimized SVR prediction model; the prediction performance of the trained SVR model for the heat transfer coefficient h and pressure drop △P of the airfoil fin PCHE in the test set and its generalization ability for new samples are tested, and relative error, mean square error MSE and correlation coefficient R are introduced. 2 As an evaluation indicator, the expression is: Where y i,SVR Represents the predicted value of the SVR model, y i,CFD Represents the CFD calculated value, and N represents the number of test set samples.

7. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to claim 1, wherein: In step (4), the optimization problem is expressed as: Objective function: Constraints: X i ∈[0,1 / 3],Y i ∈[0.05,0.15],SF∈[0.5,1] Where SF is the shrinkage factor; △P is the pressure drop; h is the heat transfer coefficient; (X i ,Y i ) is the airfoil profile control point P i Coordinates, i=1,2,3.

8. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to any one of claims 1 to 7, wherein: The multi-objective optimization algorithm is a multi-objective genetic algorithm, a multi-objective evolutionary algorithm or a multi-objective particle swarm algorithm; the decision-making method is a hierarchical analysis method, a superior-inferior solution distance method or a clustering algorithm.

9. The method for optimizing the design of airfoil fins for a printed circuit board type heat exchanger according to any one of claims 1 to 7, wherein: The K-Medoids center point clustering algorithm is introduced to find the optimization solution corresponding to the representative solution from the Pareto optimal frontier.

10. An optimization design system for airfoil fins of a printed circuit board heat exchanger, characterized in that: The optimization design system includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the optimization design method for the airfoil fin of the printed circuit board type heat exchanger according to any one of claims 1 to 9 is executed.

Images