Method for mixed variable hierarchical multi-objective optimization of servo drive heat sink

Abstract

The embodiment of the application relates to the technical field of computer-aided engineering and thermal design, and provides a servo driver radiator hybrid variable layered multi-objective optimization method, which comprises the following steps: obtaining basic parameters and initial optimization variable values to generate an initialization data set; assembling a target function and constraint conditions according to the data set to obtain a multi-objective optimization model and a constraint matrix; traversing the value of an outer-layer integer variable, reconstructing the target function and the constraint conditions, and obtaining a sub-optimization problem set; performing selective injection of an initial population based on the matching of initial values and integer values, calling a multi-objective evolutionary algorithm for iterative optimization, and obtaining a Pareto frontier solution set; performing normalization processing on the solution set and calculating the Euclidean distance, screening the optimal solution under each integer value, and globally comparing to obtain a globally optimal solution; performing full-dimension constraint verification and visual mapping to obtain an optimization result that passes the verification. The method realizes the collaborative optimization of radiator thermal resistance, pressure drop and mass multi-objectives and the precise design of engineering compatibility.

Description

Technical Field

[0001] This invention relates to the fields of computer-aided engineering and thermal design, and in particular to a hybrid variable hierarchical multi-objective optimization method for the heat sink of a servo driver. Background Technology

[0002] Heat sinks are core heat dissipation components in high heat flux density equipment such as electronic devices, industrial power equipment, and new energy equipment. Their performance directly determines the operational stability and service life of the equipment. The core performance indicators of heat sinks include thermal resistance (reflecting heat dissipation efficiency), fluid pressure drop (reflecting the energy consumption of the heat dissipation system), and their own mass (reflecting the level of lightweighting). In engineering applications, it is necessary to simultaneously achieve the multi-objective optimization requirements of "minimum thermal resistance, minimum pressure drop, and minimum mass".

[0003] The design variables of a radiator include two core dimensions: continuous variables (such as length, width, height, base thickness, fin spacing, and air velocity) and integer variables (such as the number of air ducts). Traditional approaches mainly include: manually fixing integer variables before optimizing continuous variables, or directly using multi-objective evolutionary algorithms (such as NSGA-II) to optimize the mixed "integer + continuous" variables. The former relies excessively on engineering experience, cannot traverse all feasible integer values, and is prone to missing the global optimum; the latter disrupts the continuity of solutions when performing integer rounding operations, leading to slow algorithm convergence and a tendency to get trapped in local optima. Furthermore, existing solutions suffer from insufficient constraint verification of the optimized solution, and the selection of the Pareto front solution (i.e., in multi-objective optimization, the set of all solutions that cannot further improve a certain objective without harming other objectives) lacks quantitative standards, is highly subjective, and fails to meet the precise design requirements of actual engineering projects. Summary of the Invention

[0004] This invention provides a hybrid variable hierarchical multi-objective optimization method for servo driver heat sinks, which solves the defects of existing heat sink multi-objective optimization methods, such as low efficiency of hybrid integer-continuous variable optimization, strong subjectivity in solution selection, insufficient constraint verification, and lack of global analysis capability. It achieves synergistic optimization of multiple objectives such as heat sink thermal resistance, voltage drop, and mass, as well as precise design with engineering compatibility.

[0005] This invention provides a hierarchical multi-objective optimization method for a servo driver's heat sink using mixed variables, comprising: Obtain the basic parameters for radiator optimization and the initial optimization variable values ​​input by the user; perform format validation and boundary rationality verification on the basic parameters and initial optimization variable values; mark abnormal data; and generate a qualified initial dataset. Based on the initial dataset, objective functions are assembled and constraints are generated to obtain a multi-objective optimization model and constraint matrix with the common optimization objectives of minimizing radiator thermal resistance, minimizing voltage drop, and minimizing mass. Determine the integer search range for the number of air ducts, traverse all feasible integer values ​​within this range, and according to the multi-objective optimization model, traverse and take values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values, reconstruct the objective function and constraints, and obtain the sub-optimization problem set corresponding to each value of the outer integer variable of the number of air ducts. Based on the sub-optimization problem set, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts. A multi-objective evolutionary algorithm is called to iteratively optimize and obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. Based on the Pareto front solution set, normalization and Euclidean distance calculations are performed to select the optimal solution for each outer integer variable value of the number of air ducts and perform a global comparison to obtain the global optimal solution; Based on the global optimal solution, perform full-dimensional constraint verification and visualization mapping: substitute the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix to perform matrix operations and error calculations, complete the full-dimensional constraint verification, and obtain the verified multi-objective optimization results and visualization charts.

[0006] In one possible implementation, the method further includes: The initialization dataset includes a basic parameter set and an initial optimization variable value set; The basic parameter set includes material properties, fluid properties, structural parameters, upper limit of size constraints, integer variable constraints, continuous variable constraints, and optimization algorithm parameters; The initial set of optimized variable values ​​includes a multidimensional array consisting of length, width, height, base thickness, number of air ducts, fin spacing, and wind speed.

[0007] In one possible implementation, the method further includes: The multi-objective optimization model includes thermal resistance objective function, pressure drop objective function, and mass objective function; Substitute the length, width, base thickness, and thermal conductivity into the heat conduction model, and substitute the number of air ducts, fin spacing, and wind speed into the convection heat transfer model to couple and generate the thermal resistance objective function. By substituting air density, friction coefficient, wind speed, and geometric parameters into a friction resistance model based on fluid mechanics, a pressure drop objective function is generated. By combining the material density and volume calculation model, and substituting the length, width, height, base thickness, number of air ducts, and fin spacing, a mass objective function is generated. Based on the aforementioned thermal resistance objective function, voltage drop objective function, and mass objective function, a multi-objective optimization model is generated with the common optimization objectives of minimizing radiator thermal resistance, minimizing voltage drop, and minimizing mass.

[0008] In one possible implementation, the method further includes: The optimization variables of the multidimensional array are limited in range based on the upper limit of size constraints, integer variable constraints, and continuous variable constraints, and a boundary constraint matrix is ​​generated. Based on the geometric relationship between the base thickness and the total height, inequality conditions are established, and a linear inequality constraint matrix is ​​generated. Based on the geometric compatibility formula that the width equals the product of the number of air ducts and the fin spacing plus the product of the number of air ducts plus one and the fin thickness, an equality condition is established, and a nonlinear equality constraint matrix is ​​generated.

[0009] In one possible implementation, the method further includes: The traversal values ​​of the outer integer variable of the number of air ducts include full traversal or optimized traversal using greedy algorithms or genetic algorithms; Extract the traversal values ​​of the outer integer variable of the number of air ducts, substitute the fixed values ​​into the multi-objective optimization model to reduce the dimensionality of the variables, and remove the dimension of the number of air ducts to obtain the objective function of the continuous variable after dimensionality reduction. The optimization variables after dimensionality reduction are 6-dimensional continuous variables, including length, width, height, base thickness, fin spacing and wind speed. Adjusting the dimension of the constraint matrix based on fixed values, retaining boundary constraints, linear inequality constraints, and reconstructed nonlinear equality constraints, to obtain sub-constraints that fit the current integer values; By combining the dimensionality-reduced continuous variable objective function with the sub-constraints, a set of sub-optimization problems corresponding to the outer integer variable values ​​of each air duct number is generated.

[0010] In one possible implementation, the method further includes: Based on the sub-optimization problem set, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the current air duct number values. Compare the number of air ducts in the initial optimization variable values ​​with the number of air ducts currently being traversed; If the number of wind tunnels in the initial optimization variable value is the same as the number of wind tunnels currently being traversed and the other continuous variables satisfy the boundary constraints, then the corresponding dimension-reduced continuous variables are injected into the initial population. If the number of wind tunnels in the initial optimization variable value is inconsistent with the number of wind tunnels currently being traversed, then an initial population is randomly generated based on the boundary constraints. The initial population is subjected to selection, crossover, and mutation operations by calling the NSGA-II algorithm, MOPSO algorithm, or SPEA2 multi-objective evolutionary algorithm. The population is iteratively updated to a preset number of iterations, and the non-dominated solution set is output as the Pareto front solution set corresponding to the outer integer variable value of each wind channel number. The non-dominated solution set is the solution set that cannot improve other objectives without compromising any optimization objective.

[0011] In one possible implementation, the method further includes: The thermal resistance, pressure drop, and mass target values ​​of each Pareto front solution set are min-max normalized and mapped to the [0,1] interval to obtain the normalized target vector; Calculate the Euclidean distance from each normalized objective vector to the ideal origin (0,0,0), and select the solution corresponding to the minimum distance as the optimal solution under the current outer integer variable value of the air duct number; Iterate through the optimal solutions for each outer integer variable value of the wind tunnel number, repeat normalization and distance calculation, and select the solution corresponding to the smallest global Euclidean distance as the global optimal solution.

[0012] In one possible implementation, the method further includes: The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, the linear inequality constraint matrix, and the nonlinear equality constraint matrix for matrix operations and error calculation. When the calculation error is within the preset engineering tolerance range, the verification is deemed qualified. After the verification is qualified, the optimization parameters are output. The normalized distance of the optimal solution under the outer integer variable values ​​of each air duct number is stored in an array, and a bar chart is generated to compare the positions of the global optimal solutions. The three-dimensional target values ​​of each Pareto front solution set are mapped to spatial coordinates to generate a three-dimensional scatter plot. The three-dimensional target value is projected onto the thermal resistance-pressure drop, thermal resistance-mass, and pressure drop-mass planes to generate a two-dimensional scatter projection map.

[0013] The present invention also provides a hybrid variable hierarchical multi-objective optimization device for a servo driver heat sink, comprising the following modules: The acquisition module is used to acquire the basic parameters for radiator optimization and the initial optimization variable values ​​input by the user, perform format validation and boundary rationality verification on the basic parameters and initial optimization variable values, mark abnormal data, and generate a qualified initial dataset. The generation module is used to assemble objective functions and generate constraints based on the initial dataset, so as to obtain a multi-objective optimization model and constraint matrix with the common optimization objectives of minimizing heat sink thermal resistance, minimizing pressure drop and minimizing mass. The reconstruction module is used to determine the integer search range of the number of air ducts, traverse all feasible integer values ​​within the range, and, based on the multi-objective optimization model, traverse and take values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values, reconstruct the objective function and constraints, and obtain the sub-optimization problem set corresponding to each value of the outer integer variable of the number of air ducts. The optimization module is used to selectively inject the initial population based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts, according to the sub-optimization problem set, and call a multi-objective evolutionary algorithm to iteratively optimize and obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. The optimization module is also used to perform normalization processing and Euclidean distance calculation based on the Pareto front solution set, filter the optimal solution under the outer integer variable values ​​of each air duct number and perform global comparison to obtain the global optimal solution; The optimization module is used to perform full-dimensional constraint verification and visualization mapping based on the global optimal solution. It substitutes the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix to perform matrix operations and error calculations, thereby completing the full-dimensional constraint verification and obtaining the verified multi-objective optimization results and visualization charts.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a hybrid variable hierarchical multi-objective optimization method for a servo driver heat sink as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a servo driver heat sink hybrid variable hierarchical multi-objective optimization method as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements a servo driver heat sink hybrid variable hierarchical multi-objective optimization method as described above.

[0017] The present invention provides a multi-objective optimization method for heat sinks of servo drives using a hybrid variable hierarchical approach. This method acquires basic optimization parameters and user-inputted initial optimization variable values ​​for the heat sink. It then performs format validation and boundary rationality verification on the basic parameters and initial optimization variable values, marks abnormal data, and generates a validated initial dataset. Based on the initial dataset, it assembles objective functions and generates constraints to obtain a multi-objective optimization model and constraint matrix with the combined optimization objectives of minimizing heat sink thermal resistance, voltage drop, and mass. Finally, it determines the integer search range for the number of air ducts, iterates through all feasible integer values ​​within this range, and, based on the multi-objective optimization model, iterates through the outer integer variables of the number of air ducts in the initial optimization variable values ​​to reconstruct the objective function and constraints, obtaining the corresponding values ​​of each outer integer variable for the number of air ducts. The algorithm iteratively optimizes a set of sub-optimization problems. Based on this set, it selectively injects an initial population based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts. It then uses a multi-objective evolutionary algorithm to iteratively optimize the population, obtaining the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. Based on the Pareto front solution set, it performs normalization and Euclidean distance calculations, filters the optimal solutions for each outer integer variable value of the number of air ducts, and performs a global comparison to obtain the global optimal solution. Based on the global optimal solution, it performs full-dimensional constraint verification and visualization mapping, substituting the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculations. This completes the full-dimensional constraint verification, yielding the verified multi-objective optimization results and visualization charts. Compared to existing multi-objective optimization methods for heat sinks, which suffer from low efficiency in mixed integer-continuous variable optimization, strong subjectivity in solution selection, insufficient constraint verification, and lack of global analysis capabilities, this solution achieves precise design with synergistic optimization of multiple objectives (thermal resistance, voltage drop, and mass) and engineering compatibility by employing a hierarchical decoupling strategy of outer-layer integer variable traversal and inner-layer continuous variable optimization, a selective injection mechanism for the initial population based on matching judgment, a Pareto solution selection method based on normalized Euclidean distance quantization, and collaborative processing of full-dimensional constraint verification and visualization mapping. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the hybrid variable hierarchical multi-objective optimization method for the heat sink of the servo driver provided by the present invention.

[0020] Figure 2 This is a comparison chart of the normalized distance of the optimal solution under different numbers of air ducts provided by this invention.

[0021] Figure 3 This is a three-dimensional Pareto front distribution map provided by the present invention.

[0022] Figure 4 This is a schematic diagram of the two-dimensional Pareto front projection and heat sink structure parameters provided by the present invention.

[0023] Figure 5 This is a model diagram of the R5 heat sink provided by the present invention for actual processing.

[0024] Figure 6 This is a schematic diagram of the structure of the heat sink hybrid variable hierarchical multi-objective optimization device for servo drivers provided by the present invention.

[0025] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0028] Figure 1 This is a flowchart illustrating the hybrid variable hierarchical multi-objective optimization method for the heat sink of the servo driver provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: S11. Obtain the basic parameters for radiator optimization and the initial optimization variable values ​​input by the user. Perform format verification and boundary rationality verification on the basic parameters and initial optimization variable values, mark abnormal data, and generate a qualified initial dataset.

[0029] In this embodiment of the invention, firstly, the basic parameters for radiator optimization are obtained, including material properties (thermal conductivity, material density), fluid properties (air density, friction coefficient), structural parameters (fin thickness), upper limit of size constraints (maximum length, maximum width, maximum height), integer variable constraints (upper limit of number of air ducts), continuous variable constraints (upper limit of wind speed), and optimization algorithm parameters (population size, maximum number of iterations).

[0030] Obtain the initial optimization variable values ​​input by the user: length x1, width x2, height x3, base thickness x4, number of air ducts x5, fin spacing x6, and wind speed x7.

[0031] Optionally, the input data can be format-validated and boundary-based validation can be performed to verify whether the values ​​of each initial optimization variable meet the corresponding basic parameter constraints, output the validation results and mark abnormal data, and generate a valid initial dataset.

[0032] S12. Based on the initial dataset, assemble the objective function and generate the constraint conditions to obtain a multi-objective optimization model and constraint condition matrix with the objectives of minimizing the radiator thermal resistance, minimizing the voltage drop, and minimizing the mass as the common optimization objectives.

[0033] Based on the material properties, fluid properties, and structural parameters in the initial dataset, the objective function is assembled by calling the heat conduction model, the fluid dynamics friction model, and the volume calculation model: Substitute the length, width, base thickness, and thermal conductivity into the heat conduction model, and substitute the number of air ducts, fin spacing, and wind speed into the convection heat transfer model to couple and generate the thermal resistance objective function. By substituting air density, friction coefficient, wind speed, and geometric parameters into the friction resistance model, a pressure drop objective function is generated. By combining the material density and volume calculation model, and substituting the length, width, height, base thickness, number of air ducts, and fin spacing, a mass objective function is generated. A multi-objective optimization model is generated based on the objective functions of thermal resistance, pressure drop, and mass.

[0034] Specifically, a multi-objective optimization function (a three-objective optimization function is used as an example in this embodiment) is constructed to minimize thermal resistance F1, pressure drop F2, and mass F3: Thermal resistance F1 (K / W): Based on the heat conduction and convection heat transfer model, the formula is: Pressure drop F2 (Pa): Based on the fluid dynamics friction loss model, the formula is: Mass F3 (kg): Based on the volume-density relationship, the formula is: Meanwhile, the optimization variables of the multidimensional array are limited in range based on the upper limit of size constraints, integer variable constraints, and continuous variable constraints, and a boundary constraint matrix is ​​generated. Based on the geometric relationship between the base thickness and the total height, inequality conditions are established, and a linear inequality constraint matrix is ​​generated. Based on the geometric compatibility formula that the width equals the product of the number of air ducts and the fin spacing plus the product of the number of air ducts plus one and the fin thickness, an equality condition is established, a nonlinear equality constraint matrix is ​​generated, and the constraint condition matrix is ​​obtained by combining them.

[0035] Specifically, set constraints: Boundary constraints: (lb=[0.215,0.095,0.065,0.01,20,0.002,8]mm, ub is the upper limit of each variable); Linear inequality constraints: (Base thickness ≤ total height); Nonlinear equality constraints: (Width = number of air ducts × fin spacing + (number of air ducts + 1) × fin thickness, to ensure geometric compatibility).

[0036] S13. Determine the integer search range for the number of air ducts, traverse all feasible integer values ​​within this range, and according to the multi-objective optimization model, traverse and take values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values, reconstruct the objective function and constraints, and obtain the sub-optimization problem set corresponding to each outer integer variable value of the number of air ducts.

[0037] Determine the search range (e.g., 20~26) of the outer integer variable, i.e., the number of air ducts x5, and iterate through each feasible integer value; extract the iterated values ​​of the current outer integer variable, substitute the fixed values ​​into the multi-objective optimization model to reduce the dimensionality of the variables, and remove the air duct number dimension to obtain the dimensionality-reduced continuous variable objective function; adjust the dimension of the constraint matrix according to the fixed values, retain the boundary constraints, linear inequality constraints, and reconstructed nonlinear equality constraints (substitute the fixed air duct number into the width geometric compatibility formula), and obtain the sub-constraints that fit the current integer values; combine the dimensionality-reduced continuous variable objective function and the sub-constraints to generate the sub-optimization problem set corresponding to each outer integer variable value.

[0038] Specifically, the outer integer variable is traversed: the integer search range (20~26) of the number of air ducts x5 is determined, and each feasible integer value is traversed; Inner-layer continuous variable optimization: For each fixed x5, simplify the optimization variables into 6-dimensional continuous variables (x1, x2, x3, x4, x6, x7), and perform the following operations: Construct the inner objective function: Substitute the fixed x5 into the three objective function of step 2 to generate an objective function containing only continuous variables; Adjust inner constraints: Remove the constraint dimension corresponding to x5, and retain boundary constraints, linear inequality constraints, and nonlinear equality constraints; Initial value adaptation: If the initial value input by the user matches the current x5 and satisfies the boundary, it is included in the initial population; otherwise, the population is randomly initialized. Perform multi-objective optimization: Use the NSGA-II algorithm (gamultiobj) to optimize continuous variables and output Pareto front solutions.

[0039] S14. Based on the sub-optimization problem set, selectively inject the initial population based on the matching of the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts, call the multi-objective evolutionary algorithm to iteratively optimize, and obtain the Pareto front solution set corresponding to the outer integer variable values ​​of each number of air ducts.

[0040] For each sub-optimization problem, the number of airways in the initial optimization variable values ​​is extracted and compared with the current outer integer variable values ​​of the number of airways. If they are consistent and the other continuous variables satisfy the boundary constraints, the corresponding dimensionality-reduced continuous variables are injected into the initial population. If they are inconsistent, the initial population is randomly generated based on the boundary constraints. The NSGA-II algorithm is called to perform selection, crossover, and mutation operations on the initial population, iteratively updating it to the preset number of iterations, and outputting the non-dominated solution set as the Pareto front solution set corresponding to the values ​​of each outer integer variable.

[0041] S15. Based on the Pareto front solution set, perform normalization processing and Euclidean distance calculation, select the optimal solution under the outer integer variable values ​​of each air duct number and perform global comparison to obtain the global optimal solution.

[0042] The thermal resistance, pressure drop, and mass target values ​​of each Pareto front solution set are min-max normalized and mapped to the [0,1] interval to obtain the normalized target vector. The Euclidean distance from each normalized target vector to the ideal origin (0,0,0) is calculated, and the solution corresponding to the minimum distance is selected as the optimal solution under the current outer integer variable value. The optimal solutions under the outer integer variable values ​​of each air duct number are traversed, and the normalization and distance calculation are repeated. The solution corresponding to the global minimum Euclidean distance is selected as the global optimal solution.

[0043] Specifically, the inner optimal solution: for each Pareto front solution corresponding to x5, the Euclidean distance is calculated after normalizing the objective function value, and the solution with the smallest distance is selected as the optimal solution under x5; Global optimal solution: Iterate through all inner optimal solutions of x5, filter feasible solutions, and select the solution with the smallest normalized Euclidean distance as the global optimal solution.

[0044] It should be noted that the selection of the optimal solution can be replaced by the entropy weight method, TOPSIS method, or grey relational analysis multi-attribute decision method, while the core hierarchical optimization logic remains unchanged.

[0045] S16. Based on the global optimal solution, perform full-dimensional constraint verification and visualization mapping processing. Substitute the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix to perform matrix operations and error calculations, complete the full-dimensional constraint verification, and obtain the verified multi-objective optimization results and visualization charts.

[0046] The global optimal solution's variable values ​​are substituted into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculation. After verification, the optimization parameters are output. The normalized distances of the optimal solutions under the outer integer variable values ​​of each air duct are stored in an array, generating a bar chart comparison and marking the position of the global optimal solution. The three-dimensional objective values ​​of each Pareto front solution set are spatially mapped to generate a three-dimensional scatter plot. The three-dimensional objective values ​​are projected onto the thermal resistance-pressure drop, thermal resistance-mass, and pressure drop-mass planes to generate a two-dimensional scatter projection plot, obtaining the verified multi-objective optimization results and visualization charts.

[0047] Specifically, full-dimensional constraint verification: verify the boundary constraints, linear inequality constraints, geometric constraints (x4≤x3), and nonlinear equality constraints of the global optimal solution, and output the verification results; Visual analysis: Plot a comparison of the normalized distance of the optimal solution under different numbers of air ducts, and label the global optimal solution, such as... Figure 2 As shown; Plot a 3D Pareto front distribution map to show the solution distribution for different airway numbers, such as... Figure 3 As shown; Plot two-dimensional Pareto front projections (F1-F2, F1-F3, F2-F3) to visually compare the overall performance of the solutions, such as... Figure 4 As shown.

[0048] Figure 5 This is a model diagram of the R5 heatsink provided by this invention used in actual manufacturing. For example... Figure 5 As shown, this model is a three-dimensional solid model that can be directly used for engineering processing, generated after optimizing the heat sink of a certain type of servo driver using the multi-objective optimization method described in this invention.

[0049] It should be noted that the multi-objective optimization algorithm for inner continuous variables in the embodiments of the present invention can be replaced by MOPSO (multi-objective particle swarm optimization), SPEA2 (intensity Pareto evolutionary algorithm), etc. Only the algorithm calling interface needs to be modified, and the hierarchical optimization framework, objective function and constraints remain unchanged, so that efficient optimization of mixed variables can still be achieved.

[0050] The "normalized Euclidean distance" can be replaced with multi-attribute decision-making methods such as entropy weight method, TOPSIS method, and grey relational analysis to quantify and sort the Pareto solutions, select the optimal solution, and the core hierarchical optimization logic remains unchanged, thus ensuring the overall optimality of the solution.

[0051] The traversal method for the outer integer air duct number can be replaced by "greedy algorithm / genetic algorithm traversal" (instead of full traversal), which reduces the number of traversals while ensuring the quality of the solution. This method is suitable for scenarios with a large range of integer variables and can still obtain the global optimal solution.

[0052] Nonlinear equality constraints can be incorporated into the objective function through the "penalty function method" (instead of defining constraints separately). By adding geometric compatibility error as a penalty term to the objective function, the optimization solution can still be guaranteed to meet engineering constraints without affecting the multi-objective optimization effect.

[0053] The beneficial effects brought about by the embodiments of the present invention include: Optimize efficiency and improve accuracy: The hierarchical strategy avoids the rounding error of direct optimization of mixed variables, the outer traversal ensures global coverage of integer variables, and the inner continuous variable optimization improves the convergence speed and can obtain the global optimal solution; Solution selection is quantitatively controllable: Pareto solutions are selected based on normalized Euclidean distance to avoid subjective decision-making and ensure the overall optimality of thermal resistance, pressure drop, and mass; Engineering compatibility assurance: Full-dimensional constraint verification (including geometric compatibility constraints) ensures that the optimized solution meets the processing requirements, solving the problem that existing solutions are "algorithmically feasible but engineering infeasible"; Enhanced design flexibility: Multi-dimensional visualization analysis allows for intuitive comparison of the optimization effects of different air duct numbers, supporting engineers to adjust design strategies as needed; Improved engineering adaptability: The initial value adaptation mechanism takes into account engineering experience, shortens the algorithm convergence time, and optimizes the results to better fit the actual application scenario; Multi-objective balance optimization: The precisely constructed three-objective function covers the core requirements of heat dissipation efficiency, energy consumption, and lightweight design, meeting the comprehensive heat dissipation design requirements of high-power devices.

[0054] The present invention provides a method and apparatus for hierarchical multi-objective optimization of heat sinks for servo drives. The method involves acquiring basic parameters for heat sink optimization and initial optimization variable values ​​input by the user; performing format verification and boundary rationality verification on the basic parameters and initial optimization variable values; marking abnormal data; and generating a verified initial dataset. Based on the initial dataset, the method assembles objective functions and generates constraints to obtain a multi-objective optimization model and constraint matrix with the common optimization objectives of minimizing heat sink thermal resistance, voltage drop, and mass. The method then determines the integer search range for the number of air ducts, iterates through all feasible integer values ​​within this range, and, based on the multi-objective optimization model, iterates through the outer integer variables of the number of air ducts in the initial optimization variable values ​​to reconstruct the objective function and constraints, thereby obtaining the outer integer variable values ​​for each number of air ducts. The corresponding sub-optimization problem set is determined. Based on the sub-optimization problem set, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts. A multi-objective evolutionary algorithm is invoked for iterative optimization to obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. Based on the Pareto front solution set, normalization processing and Euclidean distance calculation are performed to select the optimal solution under each outer integer variable value of the number of air ducts and perform a global comparison to obtain the global optimal solution. Based on the global optimal solution, full-dimensional constraint verification and visualization mapping processing are performed. The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculation to complete the full-dimensional constraint verification and obtain the verified multi-objective optimization results and visualization charts. Compared to existing multi-objective optimization methods for heat sinks, which suffer from low efficiency in mixed integer-continuous variable optimization, strong subjectivity in solution selection, insufficient constraint verification, and lack of global analysis capabilities, this solution achieves precise design with synergistic optimization of multiple objectives (thermal resistance, voltage drop, and mass) and engineering compatibility by employing a hierarchical decoupling strategy of outer-layer integer variable traversal and inner-layer continuous variable optimization, a selective injection mechanism for the initial population based on matching judgment, a Pareto solution selection method based on normalized Euclidean distance quantization, and collaborative processing of full-dimensional constraint verification and visualization mapping.

[0055] The following describes the multi-objective optimization device for the servo driver heat sink provided by the present invention. The multi-objective optimization device for the servo driver heat sink described below can be referred to in correspondence with the hybrid variable hierarchical multi-objective optimization method for the servo driver heat sink described above.

[0056] Figure 6 This is a schematic diagram of the structure of the servo driver heat sink hybrid variable hierarchical multi-objective optimization device provided by the present invention, specifically including: The acquisition module 601 is used to acquire the basic parameters for radiator optimization and the initial optimization variable values ​​input by the user, perform format validation and boundary reasonableness verification on the basic parameters and initial optimization variable values, mark abnormal data, and generate a valid initial dataset. For detailed explanations, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0057] The generation module 602 is used to assemble objective functions and generate constraints based on the initialization dataset, resulting in a multi-objective optimization model and constraint matrix with the common optimization objectives of minimizing heat sink thermal resistance, minimizing voltage drop, and minimizing mass. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0058] The reconstruction module 603 is used to determine the integer search range for the number of air ducts, traverse all feasible integer values ​​within this range, and, based on the multi-objective optimization model, traverse and retrieve values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values. It then reconstructs the objective function and constraints to obtain the sub-optimization problem set corresponding to each value of the outer integer variable of the number of air ducts. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0059] The optimization module 604 is used to selectively inject an initial population based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts, according to the sub-optimization problem set, and to iteratively optimize by calling a multi-objective evolutionary algorithm to obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. For detailed explanations, please refer to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0060] The optimization module 604 is further configured to perform normalization processing and Euclidean distance calculation based on the Pareto front solution set, filter the optimal solutions for each outer integer variable value of the number of air ducts, and perform a global comparison to obtain the globally optimal solution. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0061] The optimization module 605 is used to perform full-dimensional constraint verification and visualization mapping processing based on the global optimal solution. It substitutes the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculation, completing the full-dimensional constraint verification and obtaining the verified multi-objective optimization results and visualization charts. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0062] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute a multi-objective optimization method for the servo driver's heatsink using a hybrid variable hierarchical structure. This method includes: acquiring basic parameters for heatsink optimization and initial optimization variable values ​​input by the user; performing format verification and boundary rationality verification on the basic parameters and initial optimization variable values; marking abnormal data; generating a verified initialization dataset; assembling objective functions and generating constraints based on the initialization dataset to obtain a multi-objective optimization model and constraint matrix with the radiator's thermal resistance, pressure drop, and mass as the collaborative optimization objectives; determining the integer search range for the number of air ducts; traversing all feasible integer values ​​within this range; and, based on the multi-objective optimization model, traversing and taking values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values; reconstructing the objective function and constraints; and obtaining the values ​​for each air duct. The algorithm identifies a set of sub-optimization problems corresponding to the outer integer variables of the number of air ducts. Based on these sub-optimization problems, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the outer integer variables of the number of air ducts. A multi-objective evolutionary algorithm is then used for iterative optimization to obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. Based on the Pareto front solution set, normalization and Euclidean distance calculations are performed to select the optimal solution for each outer integer variable value of the number of air ducts and conduct a global comparison to obtain the global optimal solution. Based on the global optimal solution, full-dimensional constraint verification and visualization mapping are performed. The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculations to complete the full-dimensional constraint verification. The verified multi-objective optimization results and visualization charts are then obtained.

[0063] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0064] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the servo drive heat sink hybrid variable hierarchical multi-objective optimization method provided by the above methods. This method includes: obtaining the basic parameters of heat sink optimization and the initial optimization variable values ​​input by the user; performing format verification and boundary rationality verification on the basic parameters and the initial optimization variable values; marking abnormal data; and generating a verified initialization dataset; assembling objective functions and generating constraints based on the initialization dataset to obtain a multi-objective optimization model and constraint matrix with the minimum heat sink thermal resistance, minimum pressure drop, and minimum mass as the collaborative optimization objectives; determining the integer search range of the number of air ducts; traversing all feasible integer values ​​within this range; and, based on the multi-objective optimization model, optimizing the initial optimization variable values... The outer integer variables of the number of air ducts are iterated and their values ​​are taken to reconstruct the objective function and constraints, resulting in a set of sub-optimization problems corresponding to each value of the outer integer variables of the number of air ducts. Based on the sub-optimization problem set, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the values ​​of the outer integer variables of the number of air ducts. A multi-objective evolutionary algorithm is called to iteratively optimize and obtain the Pareto front solution set corresponding to each value of the outer integer variables of the number of air ducts. Based on the Pareto front solution set, normalization processing and Euclidean distance calculation are performed to select the optimal solution under each value of the outer integer variables of the number of air ducts and perform a global comparison to obtain the global optimal solution. Based on the global optimal solution, full-dimensional constraint verification and visualization mapping processing are performed. The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculation to complete the full-dimensional constraint verification and obtain the verified multi-objective optimization results and visualization charts.

[0065] Furthermore, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, this computer program implements a hierarchical multi-objective optimization method for a servo driver with mixed variables, as described above. This method includes: acquiring basic parameters for radiator optimization and initial optimization variable values ​​input by the user; performing format validation and boundary rationality verification on the basic parameters and initial optimization variable values; marking abnormal data; and generating a validated initialization dataset; assembling objective functions and generating constraints based on the initialization dataset to obtain a multi-objective optimization model and constraint matrix with the radiator's thermal resistance, pressure drop, and mass as the co-optimization objectives; determining the integer search range for the number of air ducts; traversing all feasible integer values ​​within this range; and, based on the multi-objective optimization model, optimizing the outer integer variables of the number of air ducts in the initial optimization variable values. The process involves iterating through the values, reconstructing the objective function and constraints, and obtaining a set of sub-optimization problems corresponding to the outer integer variables for each air duct number. Based on these sub-optimization problem sets, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the outer integer variable values. A multi-objective evolutionary algorithm is then invoked for iterative optimization to obtain a Pareto front solution set corresponding to each outer integer variable value. Based on the Pareto front solution set, normalization and Euclidean distance calculations are performed to select the optimal solution for each outer integer variable value and conduct a global comparison to obtain the global optimal solution. Based on the global optimal solution, full-dimensional constraint verification and visualization mapping are performed. The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculations to complete the full-dimensional constraint verification, resulting in a verified multi-objective optimization result and a visualization chart.

[0066] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0067] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A hierarchical multi-objective optimization method for a servo driver's heat sink using mixed variables, characterized in that, include: Obtain the basic parameters for radiator optimization and the initial optimization variable values ​​input by the user; perform format validation and boundary rationality verification on the basic parameters and initial optimization variable values; mark abnormal data; and generate a qualified initial dataset. Based on the initial dataset, objective functions are assembled and constraints are generated to obtain a multi-objective optimization model and constraint matrix with the common optimization objectives of minimizing radiator thermal resistance, minimizing voltage drop, and minimizing mass. Determine the integer search range for the number of air ducts, traverse all feasible integer values ​​within this range, and according to the multi-objective optimization model, traverse and take values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values, reconstruct the objective function and constraints, and obtain the sub-optimization problem set corresponding to each value of the outer integer variable of the number of air ducts. Based on the sub-optimization problem set, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts. A multi-objective evolutionary algorithm is called to iteratively optimize and obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. Based on the Pareto front solution set, normalization and Euclidean distance calculations are performed to select the optimal solution for each outer integer variable value of the number of air ducts and perform a global comparison to obtain the global optimal solution; Based on the global optimal solution, full-dimensional constraint verification and visualization mapping are performed. The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculation to complete the full-dimensional constraint verification and obtain the verified multi-objective optimization results and visualization charts.

2. The method according to claim 1, characterized in that, The initialization dataset includes a basic parameter set and an initial optimization variable value set; The basic parameter set includes material properties, fluid properties, structural parameters, upper limit of size constraints, integer variable constraints, continuous variable constraints, and optimization algorithm parameters; The initial set of optimized variable values ​​includes a multidimensional array consisting of length, width, height, base thickness, number of air ducts, fin spacing, and wind speed.

3. The method according to claim 2, characterized in that, The multi-objective optimization model includes thermal resistance objective function, pressure drop objective function, and mass objective function; Based on the initial dataset, objective functions are assembled to obtain a multi-objective optimization model with the joint optimization objectives of minimizing heat sink thermal resistance, minimizing voltage drop, and minimizing mass, including: Substitute the length, width, base thickness, and thermal conductivity into the heat conduction model, and substitute the number of air ducts, fin spacing, and wind speed into the convection heat transfer model to couple and generate the thermal resistance objective function. By substituting air density, friction coefficient, wind speed, and geometric parameters into a friction resistance model based on fluid mechanics, a pressure drop objective function is generated. By combining the material density and volume calculation model, and substituting the length, width, height, base thickness, number of air ducts, and fin spacing, a mass objective function is generated. Based on the aforementioned thermal resistance objective function, voltage drop objective function, and mass objective function, a multi-objective optimization model is generated with the common optimization objectives of minimizing radiator thermal resistance, minimizing voltage drop, and minimizing mass.

4. The method according to claim 2, characterized in that, Constraints are generated based on the initial dataset to obtain a constraint matrix, including: The optimization variables of the multidimensional array are limited in range based on the upper limit of size constraints, integer variable constraints, and continuous variable constraints, and a boundary constraint matrix is ​​generated. Based on the geometric relationship between the base thickness and the total height, inequality conditions are established, and a linear inequality constraint matrix is ​​generated. Based on the geometric compatibility formula that the width equals the product of the number of air ducts and the fin spacing plus the product of the number of air ducts plus one and the fin thickness, an equality condition is established, and a nonlinear equality constraint matrix is ​​generated.

5. The method according to claim 2, characterized in that, The traversal values ​​of the outer integer variable of the number of air ducts include full traversal or optimized traversal using greedy algorithms or genetic algorithms; The process involves determining the integer search range for the number of air ducts, traversing all feasible integer values ​​within that range, and, based on the multi-objective optimization model, iterating through the outer integer variables of the air duct number in the initial optimization variable values ​​to reconstruct the objective function and constraints. This yields a set of sub-optimization problems corresponding to each outer integer variable value of the air duct number, including: Extract the traversal values ​​of the outer integer variable of the number of air ducts, substitute the fixed values ​​into the multi-objective optimization model to reduce the dimensionality of the variables, and remove the dimension of the number of air ducts to obtain the objective function of the continuous variable after dimensionality reduction. The optimization variables after dimensionality reduction are 6-dimensional continuous variables, including length, width, height, base thickness, fin spacing and wind speed. Adjusting the dimension of the constraint matrix based on fixed values, retaining boundary constraints, linear inequality constraints, and reconstructed nonlinear equality constraints, to obtain sub-constraints that fit the current integer values; By combining the dimensionality-reduced continuous variable objective function with the sub-constraints, a set of sub-optimization problems corresponding to the values ​​of each outer integer variable of the number of air ducts is generated.

6. The method according to claim 5, characterized in that, The process involves selectively injecting an initial population based on the matching between the initial optimization variable values ​​and the outer integer variables of the number of air ducts, according to the sub-optimization problem set. A multi-objective evolutionary algorithm is then invoked for iterative optimization to obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts, including: Based on the sub-optimization problem set, selective initial population injection is performed based on the matching between the initial optimization variable values ​​and the current air duct number values. Compare the number of air ducts in the initial optimization variable values ​​with the number of air ducts currently being traversed; If the number of wind tunnels in the initial optimization variable value is the same as the number of wind tunnels currently being traversed and the other continuous variables satisfy the boundary constraints, then the corresponding dimension-reduced continuous variables are injected into the initial population. If the number of wind passages in the initial optimization variable value is inconsistent with the number of wind passages currently being traversed, an initial population is randomly generated based on boundary constraints. The NSGA-II algorithm, MOPSO algorithm, or SPEA2 multi-objective evolutionary algorithm is called to perform selection, crossover, and mutation operations on the initial population, and iteratively updated to a preset number of iterations. The non-dominated solution set is output as the Pareto front solution set corresponding to the outer integer variable value of each wind passage number. The non-dominated solution set is the set of solutions that cannot improve other objectives without compromising any optimization objective.

7. The method according to claim 6, characterized in that, The process involves normalizing and calculating Euclidean distance based on the Pareto front solution set, selecting the optimal solution for each outer integer variable value of the wind tunnel number, and performing a global comparison to obtain the globally optimal solution, including: The thermal resistance, pressure drop, and mass target values ​​of each Pareto front solution set are min-max normalized and mapped to the [0,1] interval to obtain the normalized target vector; Calculate the Euclidean distance from each normalized objective vector to the ideal origin (0,0,0), and select the solution corresponding to the minimum distance as the optimal solution under the current outer integer variable value of the air duct number; Iterate through the optimal solutions for each outer integer variable value of the wind tunnel number, repeat normalization and distance calculation, and select the solution corresponding to the smallest global Euclidean distance as the global optimal solution.

8. The method according to claim 7, characterized in that, The process involves performing full-dimensional constraint verification and visualization mapping based on the global optimal solution. This includes substituting the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix for matrix operations and error calculations. This completes the full-dimensional constraint verification, yielding verified multi-objective optimization results and visualization charts, including: The variable values ​​of the global optimal solution are substituted into the boundary constraint matrix, the linear inequality constraint matrix, and the nonlinear equality constraint matrix for matrix operations and error calculation. When the calculation error is within the preset engineering tolerance range, the verification is deemed qualified. After the verification is qualified, the optimization parameters are output. The normalized distance of the optimal solution under the outer integer variable values ​​of each air duct number is stored in an array, and a bar chart is generated to compare the positions of the global optimal solutions. The three-dimensional target values ​​of each Pareto front solution set are mapped to spatial coordinates to generate a three-dimensional scatter plot. The three-dimensional target value is projected onto the thermal resistance-pressure drop, thermal resistance-mass, and pressure drop-mass planes to generate a two-dimensional scatter projection map.

9. A hybrid variable hierarchical multi-objective optimization device for a servo driver's heat sink, characterized in that, include: The acquisition module is used to acquire the basic parameters for radiator optimization and the initial optimization variable values ​​input by the user, perform format validation and boundary rationality verification on the basic parameters and initial optimization variable values, mark abnormal data, and generate a qualified initial dataset. The generation module is used to assemble objective functions and generate constraints based on the initial dataset, so as to obtain a multi-objective optimization model and constraint matrix with the common optimization objectives of minimizing heat sink thermal resistance, minimizing pressure drop and minimizing mass. The reconstruction module is used to determine the integer search range of the number of air ducts, traverse all feasible integer values ​​within the range, and, based on the multi-objective optimization model, traverse and take values ​​for the outer integer variables of the number of air ducts in the initial optimization variable values, reconstruct the objective function and constraints, and obtain the sub-optimization problem set corresponding to each value of the outer integer variable of the number of air ducts. The optimization module is used to selectively inject the initial population based on the matching between the initial optimization variable values ​​and the outer integer variable values ​​of the number of air ducts, according to the sub-optimization problem set, and call a multi-objective evolutionary algorithm to iteratively optimize and obtain the Pareto front solution set corresponding to each outer integer variable value of the number of air ducts. The optimization module is also used to perform normalization processing and Euclidean distance calculation based on the Pareto front solution set, filter the optimal solution under the outer integer variable values ​​of each air duct number and perform global comparison to obtain the global optimal solution; The optimization module is used to perform full-dimensional constraint verification and visualization mapping based on the global optimal solution. It substitutes the variable values ​​of the global optimal solution into the boundary constraint matrix, linear inequality constraint matrix, and nonlinear equality constraint matrix to perform matrix operations and error calculations, thereby completing the full-dimensional constraint verification and obtaining the verified multi-objective optimization results and visualization charts.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the heat sink hybrid variable hierarchical multi-objective optimization method for servo drives as described in any one of claims 1 to 8.

Images