Structural optimization design method of intelligent bearing multilayer heterogeneous microsystem

Abstract

The invention relates to an intelligent bearing multilayer heterogeneous microsystem structure optimization design method comprising the following steps: building a physical model of an intelligent bearing multilayer heterogeneous microsystem, and carrying out finite element analysis; determining a to-be-optimized design parameter and a value boundary thereof according to a finite element analysis result, and extracting a plurality of samples; the design parameters are used as optimization variables, the set design index optimization is used as an optimization target, a digital model is obtained through iteration fitting by adopting a multi-target optimization algorithm, the digital model and finite element analysis are compared in the iteration process, and calculation results of the design indexes are obtained through calculation according to the extracted samples. And if the difference between the two parameters is within a set range, outputting the current design parameter as an optimal parameter. According to the invention, the structure optimization of the multi-layer heterogeneous microsystem of the intelligent bearing under a complex working condition can be effectively realized, so that the bearing efficiency and reliability of the intelligent bearing are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent bearing design, and particularly to a structural optimization design method for a multi-layer heterogeneous microsystem of an intelligent bearing. Background Art

[0002] The multi-layer heterogeneous microsystem is the core component for an intelligent bearing to achieve source-end monitoring, signal acquisition, processing, and transmission. It can realize real-time monitoring of the bearing operating state under complex working conditions. This system uses high-precision sensor components to efficiently collect key parameters including temperature, vibration, stress, and rotational speed, and instantaneously processes, analyzes, and transmits these signals through an embedded processing unit, which is an indispensable core technology in the intelligent bearing system.

[0003] In the design of intelligent bearings, the optimization design of the multi-layer heterogeneous microsystem occupies a prominent position and is the key to improving the overall performance and adaptability of intelligent bearings. Existing optimization methods have problems such as insufficient consideration of complex working conditions, complex optimization processes, and poor optimization effects. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a structural optimization design method for a multi-layer heterogeneous microsystem of an intelligent bearing, which can effectively realize the structural optimization of the multi-layer heterogeneous microsystem of an intelligent bearing under complex working conditions.

[0005] The technical solution adopted by the present invention to solve its technical problems is: to provide a structural optimization design method for a multi-layer heterogeneous microsystem of an intelligent bearing, including:

[0006] Establish a physical model of the multi-layer heterogeneous microsystem of the intelligent bearing, and perform finite element analysis based on this physical model to obtain the physical response data of the multi-layer heterogeneous microsystem of the intelligent bearing under different working conditions;

[0007] Determine the design parameters to be optimized and their value boundaries according to the finite element analysis results, and extract a number of samples;

[0008] Taking the design parameters as optimization variables and the optimal setting of the design index as the optimization goal, use a multi-objective optimization algorithm to iteratively fit a digital model, and compare the calculation results of the design index calculated by the digital model and the finite element analysis according to the extracted samples during the iteration process. If the difference between the two is within the set range, output the current design parameters as the optimal parameters.

[0009] Further, the determination of the design parameters to be optimized according to the finite element analysis results includes:

[0010] Based on the physical response data obtained from the finite element analysis, establish a response matrix between the design parameters and the performance indicators;

[0011] Based on the response matrix, calculate the contribution degree of each design parameter to the performance index;

[0012] Select the design parameters with contribution degrees greater than the set value as the design parameters to be optimized.

[0013] Further, taking the design parameters as optimization variables and the optimal set design index as the optimization goal, use a multi-objective optimization algorithm to iteratively fit a digital model, including:

[0014] Taking the samples of the design parameters as inputs and the set design index as the output, establish an approximation model;

[0015] Based on the approximation model, solve the multi-objective optimization problem with the design parameters as optimization variables and the optimal set design index as the optimization goal.

[0016] Further, use the Kriging method or the polynomial response surface model to establish the approximation model.

[0017] Further, use at least one of the multi-objective genetic algorithm, particle swarm algorithm, and simulated annealing algorithm to solve the multi-objective optimization problem.

[0018] Further, the multi-objective genetic algorithm selects parent individuals for the next iteration through the Pareto sorting mechanism, introduces key performance indicators into the Pareto sorting mechanism for individual evaluation, and sorts according to the evaluation results.

[0019] Further, the step of introducing key performance indicators into the Pareto sorting mechanism for individual evaluation includes:

[0020] Extract key performance indicators from the physical response data obtained from the finite element analysis;

[0021] Eliminate the individuals whose key performance indicators do not meet the set conditions, and calculate the weighted sum of the key performance indicators of each remaining individual as the evaluation value.

[0022] Further, the multi-objective genetic algorithm generates offspring individuals through crossover operations and mutation operations, and sets the crossover probability and mutation probability according to the structural characteristics of the intelligent bearing multi-layer heterogeneous microsystem.

[0023] Further, the finite element analysis includes static analysis, dynamic analysis, and heat conduction analysis.

[0024] Further, the design index includes at least one of strength, life, and heat dissipation performance.

[0025] Furthermore, the extraction of several samples is achieved by the optimal Latin hypercube sampling method, and the number of samples is determined according to the dimension of the design parameters and the optimization requirements.

[0026] Beneficial effects

[0027] Due to the adoption of the above technical solution, compared with the prior art, the present invention has the following advantages and positive effects: By performing finite element analysis on the intelligent bearing multi-layer heterogeneous microsystem, the design parameters to be optimized and their value boundaries of the microsystem structure under real working conditions are determined. Then, several samples are extracted according to the value range of the design parameters, and an approximate model is constructed and multi-objective optimization is carried out. The design index results calculated using the optimized digital model are compared with those obtained from finite element analysis to obtain the finally optimized design parameters, which can effectively realize the design of the intelligent bearing multi-layer heterogeneous microsystem under complex working conditions, optimize the material selection, component layout and structure design, enhance the strength, life and heat dissipation capacity of the bearing and the microsystem, improve the durability, reduce the failure risk, and ensure safe, stable and reliable operation in complex environments, so as to exert its best efficacy. Brief description of the drawings

[0028] Figure 1 is the flowchart of the embodiment of the present invention;

[0029] Figure 2 is the structure diagram of the intelligent bearing multi-layer heterogeneous microsystem of the embodiment of the present invention;

[0030] Figure 3 is the schematic diagram of the finite element mesh model of the intelligent bearing multi-layer heterogeneous microsystem of the embodiment of the present invention;

[0031] Figure 4 is the schematic diagram of the finite element result of the fatigue life analysis of the intelligent bearing multi-layer heterogeneous microsystem of the embodiment of the present invention;

[0032] Figure 5 is the schematic diagram of the finite element result of the strength analysis of the intelligent bearing multi-layer heterogeneous microsystem of the embodiment of the present invention;

[0033] Figure 6 is the schematic diagram of the finite element result of the thermal analysis of the energy bearing multi-layer heterogeneous microsystem of the embodiment of the present invention;

[0034] Figure 7 is the schematic diagram of the fitting result of the Isight surrogate model construction of the embodiment of the present invention;. Specific embodiments

[0035] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of this application.

[0036] The embodiment of the present invention relates to a structural optimization design method for a multi-layer heterogeneous microsystem for intelligent bearings, as Figure 1 shown, including the following steps:

[0037] Step S1: Establish a physical model of the multi-layer heterogeneous microsystem for intelligent bearings, perform pre-processing and import it into the finite element analysis software;

[0038] Step S2: According to the working characteristics of the intelligent bearing under complex working conditions (such as high temperature, high speed, strong vibration and other environments), perform finite element analysis. By dividing the mesh of the physical geometry model, applying boundary conditions, setting initial conditions, establishing a finite element model, solving, and analyzing the results, etc., obtain the fatigue life, stress distribution and heat conduction response results of the microsystem;

[0039] Step S3: Based on the finite element analysis results, perform parametric modeling, define the optimization design variables and objective functions, and determine the optimization design objectives and parameter boundaries;

[0040] Step S4: Use the optimization algorithm for iterative calculation to obtain the optimization design scheme. Through iterative design and multi-objective optimization strategy, continuously update the model until the optimal parameter combination is found to ensure that the comprehensive design objectives such as life, strength and heat conduction of the microsystem reach the best;

[0041] Step S5: Verify the optimization design scheme to ensure that the optimization results meet the design requirements. Through the finite element model proximity check, if the digital model is close to the finite element model, output the optimal parameter combination optimization design model, otherwise adjust the digital model optimization algorithm.

[0042] Among them, the physical model in step S1 is established through a parametric modeling tool, including geometric modeling, material property definition and boundary condition setting.

[0043] The finite element analysis in step S2 includes static analysis, dynamic analysis and heat conduction analysis:

[0044] S201 performs a thermal analysis on the bearing and the microsystem, establishes a heat transfer model based on the principles of heat transfer, and calibrates the model through experimental data; predicts the heat distribution based on the calibrated model, studies the performance dependence of the microsystem materials at different temperatures, nonlinear effects (such as thermal expansion and contraction), the performance of materials under thermal fatigue and thermal creep, the analysis of thermal boundary conditions, and the failure mechanisms of materials and systems in extreme thermal environments;

[0045] S202 performs a strength analysis on the microsystem, determines the stress state of the microsystem when stressed through stress analysis, and evaluates the deformation degree of the microsystem through strain analysis; conducts a fatigue analysis to predict the life of the microsystem under repeated loads, and evaluates the development risk of cracks based on fracture mechanics analysis; at the same time, conducts a thermodynamic analysis considering the influence of temperature changes on the material properties between different layers of the microsystem;

[0046] S203 performs a reliability assessment on the microsystem, and uses Ansys software for structural reliability analysis. The specific steps include establishing a geometric model, meshing, defining materials and loads, establishing a finite element model, defining reliability indicators, and performing reliability calculations.

[0047] In some specific embodiments, the establishment of the heat transfer model in step S201 is based on the basic principles of heat transfer and calibrated through experimental data to ensure the accuracy of the model, and the heat distribution prediction is carried out using the finite element method; the strength analysis in step S202 includes stress analysis, strain analysis, fatigue analysis and fracture mechanics analysis, as well as thermodynamic analysis considering temperature changes, all of which are carried out using the finite element method. And during the analysis process, factors such as the nonlinear characteristics of the materials of each layer of the intelligent bearing multi-layer heterogeneous microsystem, the thermal resistance and mechanical properties of the contact interface are considered to improve the accuracy of the analysis results. The reliability assessment in step S203 is carried out using Ansys software, and a finite element model is established through establishing a geometric model, meshing, defining materials and loads for reliability calculation.

[0048] The parametric modeling in step S3 includes the selection of optimized design variables and the definition of the objective function. The objective function can include strength, life and heat dissipation performance, etc. Among them, the strength index is used to ensure that the intelligent bearing does not fail when bearing complex loads; the life index is directly related to the maintenance cycle and usage cost of the intelligent bearing; the heat dissipation performance index ensures that the intelligent bearing can operate stably under high-temperature working conditions. The corresponding weights and optimized target values of each index are determined according to the specific application scenarios of the intelligent bearing (such as aerospace, industrial manufacturing and other fields).

[0049] In step S4, the optimization algorithm adopted is one or a combination of genetic algorithm, particle swarm algorithm, or simulated annealing algorithm. When using these algorithms, improvements are made to the initial population generation, search strategy, etc. of the algorithm according to the structural and performance characteristics of the intelligent bearing multi-layer heterogeneous microsystem. For example, in the multi-objective genetic algorithm, the probability and range of genetic operations are adjusted according to the performance requirements of the key parts of the system to improve the convergence speed and optimization effect of the algorithm. In step S5, the verification of the optimized design scheme is carried out by comparing the finite element analysis results with the digital proxy model results to ensure that it meets the accuracy requirements of fatigue life, stress distribution, and heat conduction response.

[0050] Steps S4 and S5 perform the optimization design process. More specific methods include:

[0051] Design parameter sampling: Generate sample points by sampling and combining design parameters;

[0052] Parametric model update: Perform parametric update on the simulation model to generate an updated 3D model;

[0053] Finite element model proximity check: Compare the proximity of the digital model and the finite element model. If they are close, proceed to the next step; otherwise, adjust the optimization algorithm;

[0054] Ansys Workbench calculation and design index output: Perform finite element calculation in Ansys Workbench and output the design index;

[0055] Optimization based on the digital model: Use Isight software to fit the digital model between the design parameters and the design index and perform optimization;

[0056] Optimal design index: Determine the optimal design parameters according to the optimization results to achieve the purpose of improving the reliability and performance of the equipment.

[0057] Among them, the optimal Latin hypercube sampling method can be used for design parameter sampling. The number of sampling points is determined according to the dimension of the design parameters and the optimization requirements. The parametric model update is carried out in 3D modeling software. The finite element model proximity check is carried out by comparing the results of the digital model and the finite element model. The Ansys Workbench calculation and design index output are carried out in Ansys Workbench software. The optimization based on the digital model is carried out using Isight software. The optimal design index is set according to the optimization results and the actual engineering requirements, comprehensively considering various factors such as the reliability, performance, and cost of the equipment to determine the optimal combination of design parameters.

[0058] Ansys Workbench calculations and design metric outputs include, but are not limited to, reliability metrics such as stress, strain, fatigue life, and other design metrics determined based on optimization objectives. Optimization based on the digital model uses Isight software to fit the digital model between design parameters and design metrics. The fitting method can be a polynomial response surface model or other suitable fitting methods (such as selecting the Kriging method for approximate modeling) to ensure the accuracy and reliability of the optimization results. The Kriging method can achieve the optimal unbiased estimation within a limited area through spatial local interpolation and has good results when fitting highly nonlinear problems.

[0059] The construction principle of the model is as follows:

[0060]

[0061] In the above formula, x0 is the vector point to be estimated, and x i is the adjacent known vector point, is the estimated value, and λ i is the weighting coefficient to be solved. More specifically, the known vector point x i represents the sample data points that have been obtained through finite element analysis using Ansys Workbench. The vector point x0 to be estimated represents the supplementary data points that need to be interpolated using the Kriging method to construct a continuous surrogate model. Solve the undetermined weighting coefficient λ under the following constraints i :

[0062]

[0063] The multi-objective optimization problem consists of several objective functions and constraint functions. Among them, the constraint functions can be equality constraints or inequality constraints. However, the optimization objectives are often contradictory, and it is impossible to control multiple objectives to the minimum or maximum at the same time. Therefore, it is often necessary to sacrifice one optimization objective to the extreme to satisfy multiple objectives simultaneously. The multi-objective optimization problem can be summarized as the following formula:

[0064] min{f1(x), f2(x), …, f m (x)}

[0065] s.t. g i (x) ≤ 0, i = 1, 2, …, p

[0066] h j (x) ≤ 0, j = 1, 2, …, q

[0067] x1 ≤ x ≤ x n , x = (x1, x2, …, x n )

[0068] Among them, f1(x), f2(x), …, f m (x) represent different objective functions, s.t. represents the constraint conditions, and g i (x) represents p possible linear inequality constraints, and h j (x) represents q possible non - linear inequality constraints.

[0069] The optimization process of the multi - objective genetic algorithm is to search in the solution space through genetic operations. Its basic principle is based on natural selection and genetic mechanisms, including operations such as selection, crossover, and mutation, to solve multi - objective optimization problems.

[0070] In multi - objective optimization, our goal is to find a set of decision variables x such that the values of multiple objective functions f1(x), f2(x), …, f m (x) reach the optimal simultaneously (usually minimized, but can also be maximized or meet other specific conditions).

[0071] These objective functions can be combined into an objective function vector F(x):

[0072] F(x) = m(f1(x), f2(x), …, f m (x))

[0073] Our task is to find the value of x that makes each component in F(x) as small as possible.

[0074] In the multi - objective genetic algorithm, each solution is represented as an individual, and the fitness of the individual is jointly determined by multiple objective function values. The algorithm retains excellent individuals through the selection operation and generates new individuals through the crossover and mutation operations to continuously approach the optimal solution set. The fitness of the individual can be calculated by the weighted sum method, Pareto ranking, or other methods.

[0075] The key of the multi - objective genetic algorithm lies in how to balance the conflicts between multiple objective functions and how to maintain the diversity of the solution set. For this purpose, the algorithm introduces strategies such as weight sum, Pareto ranking, crowding degree calculation, etc. to ensure the convergence and distribution of the solution set.

[0076] The mathematical representation of Pareto ranking is:

[0077] In the multi - objective genetic algorithm, when comparing two individuals x1 and x2, if the following two conditions are met:

[0078] For all objective functions f i (x) (where i = 1, 2, …, m), the objective function value of individual x1 is not greater than that of individual x2, that is, f i (x1) ≤ f i(x2) holds for all i.

[0079] And there is at least one objective function f j (x) (where j is a specific objective function index) such that the objective function value of individual x1 is less than that of individual x2, i.e., f j (x1) < f j (x2).

[0080] Then, it is said that individual x1 is Pareto superior to individual x2. This means that without making the values of other objective functions worse, individual x1 performs better than individual x2 in at least one objective function.

[0081] Through this Pareto ranking mechanism, the multi-objective genetic algorithm can continuously select those individuals that perform well (i.e., Pareto superior) in each objective function during the iterative process, thereby promoting the evolution of the solution set towards the optimal direction while maintaining the diversity and convergence of the solution set.

[0082] Before each iteration, the algorithm will select excellent individuals as parents through a selection operation according to the state of the current solution set. The selection operation is usually based on the fitness value of the individuals, but other factors such as the diversity of the solution set will also be considered. The selection operation can be implemented by methods such as roulette wheel selection and tournament selection. In some preferred embodiments, four key performance indicators extracted by finite element analysis can be combined in the Pareto ranking process for hierarchical evaluation to select more targeted and advantageous parent individuals: 1) The maximum equivalent stress value of the sensor layer (obtained through static analysis); 2) The temperature gradient of the encapsulation layer (obtained through heat conduction analysis); 3) The coefficient of variation of the interface contact pressure distribution (obtained through dynamic analysis); 4) The fundamental frequency offset of the microsystem (obtained through modal analysis). The evaluation method uses the constraint weighting method: when any index exceeds the allowable threshold of the material, it is directly eliminated, and the remaining individuals are sorted according to the weighted sum of the indexes, and the weight distribution is 0.4:0.3:0.2:0.1.

[0083] The crossover operation is the main way to generate new individuals in the genetic algorithm. By exchanging some genes of the parent individuals, new individuals with new characteristics can be generated. The way and probability of the crossover operation will affect the search efficiency and solution set quality of the algorithm. The crossover operation can be implemented by methods such as single-point crossover, two-point crossover or uniform crossover, and its mathematical representation is:

[0084] Select the crossover point: First, randomly select one or more crossover points. These crossover points divide the genes of the parent individuals into multiple segments.

[0085] Exchange gene segments: Then, based on the selected crossover points, we exchange the gene segments of two parent individuals. Specifically, we can combine the gene segments of p1 before a certain crossover point with the gene segments of p2 after that crossover point to form a new offspring individual c1; similarly, we can combine the gene segments of p2 before the crossover point with the gene segments of p1 after the crossover point to form a new offspring individual c2.

[0086] Generate offspring individuals: Finally, through the above exchange operations, we obtain two new offspring individuals c1 and c2, which inherit part of the genes of the parent individuals respectively.

[0087] The mutation operation is an important means to maintain the diversity of the solution set and jump out of local optima in genetic algorithms. By randomly changing some genes in an individual, new individuals different from the parent can be generated. The probability and method of the mutation operation also need to be adjusted according to the specific problem. The mutation operation can be achieved by methods such as simple mutation, complex mutation, or adaptive mutation.

[0088] In some preferred embodiments, structural features can be incorporated into genetic operators in the following ways: 1) The crossover strategy adopts hierarchical grouping crossover. The parameters are grouped according to the structural levels (encapsulation layer / sensing layer / substrate layer). Arithmetic crossover is performed within the group, and the crossover probability is set to 0.8; the parameters between groups remain independent, and the crossover probability is reduced to 0.2; 2) The mutation probability is adaptively adjusted according to sensitivity. According to the results of parameter sensitivity analysis, a low mutation probability of 0.05 - 0.1 is set for key parameters (such as the thickness of the copper cover) with a contribution degree > 15%, and a high mutation probability of 0.2 - 0.3 is set for auxiliary parameters with a contribution degree < 5%. The monotonic change characteristics of key structural dimensions are ensured through mutation direction constraints.

[0089] During the iterative process of the multi-objective genetic algorithm, the solution set is continuously updated, and excellent individuals are retained. By comparing new individuals with the individuals in the current solution set, the algorithm decides whether to add the new individuals to the solution set or replace some individuals in the solution set. This is usually achieved through Pareto sorting and crowding degree calculation.

[0090] In the optimal design of microsystems, the multi-objective genetic algorithm can simultaneously consider multiple performance indicators, such as power consumption, speed, area, etc., and find the best compromise among these indicators. By continuously optimizing the solution set, the algorithm can approximate the true Pareto front and provide designers with multiple feasible optimization schemes.

[0091] The mathematical representation of the Pareto front is:

[0092] Solution set P: First, we have a solution set P, which contains all possible solutions (or individuals). Each solution is a point in a multi-dimensional space, and its coordinates are composed of the values of each performance indicator.

[0093] Set of objective function values F(P): For each solution in the solution set P, we can calculate its objective function value, that is, the values of various performance metrics. These values form the set of objective function values F(P).

[0094] Pareto optimal points: In F(P), some points are particularly special and are called Pareto optimal points. These points satisfy the following conditions: There does not exist other points that can improve any one performance metric without making other performance metrics worse. In other words, these points achieve the best compromise among multiple performance metrics.

[0095] Pareto front: Finally, the Pareto front is the set of all Pareto optimal points in F(P). It forms a surface (or curve in the case of only two performance metrics) in a multi-dimensional space, and each point on this surface (or curve) represents a solution that achieves the best compromise among multiple performance metrics.

[0096] In some other specific embodiments, the optimization based on the digital model can be performed by fitting a polynomial response surface model and optimizing according to the fitting results to obtain the optimal design parameters. The 3D modeling software used for parametric model update can be Solidworks or other similar software to ensure the accuracy and efficiency of model update. The standard for finite element model proximity check is that the result difference between the digital model and the finite element model is within an acceptable range, and the specific difference range is determined according to the engineering practice and optimization requirements.

[0097] The following is further illustrated through a specific embodiment.

[0098] 1. Simulation analysis and preprocessing:

[0099] Select the analysis system: Select appropriate simulation analysis tools, such as Electric and Static Structural, etc.

[0100] Define engineering data: Define the required engineering data in the simulation analysis tool.

[0101] Establish the physical model of the intelligent bearing multi-layer heterogeneous microsystem: Use the simulation analysis tool (such as the DM model or import through the CAD model) to establish the physical model, and its structure is as Figure 2 shown.

[0102] Preprocessing: Perform necessary preprocessing work on the simulation model, such as mesh generation, material property definition, application of boundary conditions, application of load conditions, setting of initial conditions, contact definition, mesh quality check, solution settings, etc. The finite element mesh model of the intelligent bearing multi-layer heterogeneous microsystem is as Figure 3 shown.

[0103] 2. Simulation solution and post-processing:

[0104] Solution: Simulate and solve the pre-processed model.

[0105] Post-processing: Process the simulation data, including micro-system fatigue life analysis, micro-system strength assessment, and micro-system thermal analysis, etc. Specifically as follows:

[0106] ① Conduct reliability assessment on the micro-system. Use Ansys for structural reliability analysis. Through steps such as establishing a geometric model, meshing, defining materials and loads, establishing a finite element model, defining reliability indicators, performing reliability calculations, and optimizing the design, conduct reliability analysis on the system, improve the reliability of the system, and obtain the fatigue life of the micro-system. The finite element results of the fatigue life analysis of the multi-layer heterogeneous micro-system of the intelligent bearing are as Figure 4 shown.

[0107] ② Conduct strength analysis on the micro-system. Determine the stress state of the micro-system when stressed through stress analysis, and evaluate the deformation degree of the micro-system through strain analysis; predict the life of the micro-system under repeated loads through fatigue analysis, and evaluate the risk of crack development based on fracture mechanics analysis; consider the influence of temperature changes on the material properties of different layers of the micro-system through thermodynamics analysis. The finite element results of the strength analysis of the multi-layer heterogeneous micro-system of the intelligent bearing are as Figure 5 shown.

[0108] ③ Conduct thermal analysis on the bearing and the micro-system. Through the establishment and verification of the heat transfer model, model calibration based on experimental data, and prediction of heat distribution, study the performance dependence of the micro-system materials at different temperatures, non-linear effects (such as thermal expansion and contraction), the performance of materials under thermal fatigue and thermal creep, the analysis of thermal boundary conditions, and the failure mechanisms of materials and systems in extreme thermal environments, so as to optimize the design of the heat dissipation capacity of the micro-system in high-temperature environments. The finite element results of the thermal analysis of the multi-layer heterogeneous micro-system of the intelligent bearing are as Figure 6 shown.

[0109] 3. Optimization design process:

[0110] Selection of design parameters: Achieved through quantitative analysis of parameter sensitivity. First, establish a parameter-index response matrix, use the variance decomposition method to calculate the contribution of each structural parameter to the strength, life, and heat dissipation performance indicators, and screen out the key parameters whose total contribution exceeds 70%. For example, in this embodiment, the contribution of the copper cover thickness to the thermal stress distribution reaches 42.3%, and the contribution of the elastic modulus of the PCB board to the vibration mode reaches 35.6%. Therefore, select the two as optimization variables.

[0111] Design parameter sampling: Combine the sampled design parameters to generate sample points. The optimal Latin hypercube sampling method is used for sampling, and the number of samples satisfies N≥(k + 1)(k + 2) / 2, where k is the dimension of the design parameters, ensuring the fitting accuracy of the response surface model.

[0112] Ansys Workbench calculation and design index output: Perform calculations in Ansys Workbench and output the design indexes. The sample points obtained by sampling and the simulation calculation design indexes are as follows:

[0113]

[0114] Optimization based on the digital model: Use Isight to fit the design parameter - design index digital model for optimization.

[0115] The optimization variables selected in this optimization example are the copper cover thickness (potting glue layer thickness) and the elastic modulus of the PCB board, and the parameter value boundaries are determined: Assume that the total thickness of the copper cover and the potting glue is 1.5 mm, and assume that the minimum thickness of the copper cover and the potting glue is 0.3 mm, that is, the design range of the copper cover thickness is 0.3 - 1.2 mm, and the selection range of the elastic modulus of the PCB board is 35 - 40 MPa. The optimization design goal is set to the optimal combination of the micro - system life and the safety factor.

[0116] Optimal design index: Obtain the optimal design parameter combination according to the calculation results. If the optimization design process is not completed, add the optimal parameter combination to the sample points.

[0117] The fitting result of the Isight surrogate model construction in this optimization example is as Figure 7 shown. According to the calculation results of the mathematical model, when the copper cover thickness is 0.7 mm and the elastic modulus of the PCB board is 35 Mpa, the combination of the micro - system life and the safety factor is optimal. At this time, the life of the weakest part of the micro - system calculated by the surrogate model is 143,910 cycles.

[0118] Optimize the micro - system. Through steps such as design parameter sampling, Ansys Workbench calculation and design index output, optimization based on the digital model, and optimal design index, perform precise simulation and optimization design of the multi - layer heterogeneous micro - system, aiming to improve the bearing performance while extending its service life and ensuring its safety factor. The design process is scientific and reasonable, ensuring the reliability and practicality of the optimization results.

[0119] 5. End of the optimization design process:

[0120] When the calculation results of the mathematical model are close to those of the finite - element model, output the optimal design parameter combination; otherwise, add it back to the sample points and update the parametric model of the intelligent bearing multi - layer heterogeneous micro - system, and repeat the above optimization process.

[0121] Finite element model proximity check: If the digital model is close to the finite element model, proceed to the next step; otherwise, adjust the optimization algorithm.

[0122] Parametric model update: When the error between the calculation results of the mathematical model and the finite element model is too large, perform parametric update on the simulation model and repeat the above optimization process.

[0123] Determination of the optimal design parameter combination: When the calculation results of the mathematical model and the finite element model are close, output the optimal design parameter combination and determine the final optimal design parameter combination.

[0124] This example only takes the micro-system life and safety factor as the optimization objectives. Referring to this example and adopting this optimization design method and process, it is also possible to achieve the optimization design objectives with the optimal micro-system fatigue life, micro-system structural strength, and micro-system heat transfer effect.

[0125] In this example, the life of the weakest part of the micro-system calculated by the digital model is 143,910 cycles, and the life of the weakest part of the micro-system calculated by the simulation model is 143,950 cycles. The error between the two is 0.028%. It can be considered that the mathematical model is close to the finite element model. At this time, the optimal design parameters are output: the copper cover thickness is 0.7 mm, and the elastic modulus of the PCB board is 35 Mpa.

Claims

1. A structural optimization design method for an intelligent bearing multi-layer heterogeneous microsystem, characterized in that Including: Establish a physical model of the intelligent bearing multi-layer heterogeneous microsystem, and perform finite element analysis based on this physical model to obtain the physical response data of the intelligent bearing multi-layer heterogeneous microsystem under different working conditions; Determine the design parameters to be optimized and their value boundaries according to the finite element analysis results, and extract a number of samples; Taking the design parameters as optimization variables and the optimal setting of the design indicators as the optimization goal, use a multi-objective optimization algorithm to iteratively fit a digital model, and compare the calculation results of the design indicators calculated by the digital model and the finite element analysis respectively according to the extracted samples during the iteration. If the difference between the two is within the set range, output the current design parameters as the optimal parameters.

2. The method according to claim 1, characterized in that, The determining the design parameters to be optimized according to the finite element analysis results includes: Based on the physical response data obtained from the finite element analysis, establish a response matrix between the design parameters and the performance indicators; Based on the response matrix, calculate the contribution degree of each design parameter to the performance indicators; Select the design parameters with a contribution degree greater than the set value as the design parameters to be optimized.

3. The method according to claim 1, wherein The taking the design parameters as optimization variables and the optimal setting of the design indicators as the optimization goal, and using a multi-objective optimization algorithm to iteratively fit a digital model includes: Taking the samples of the design parameters as the input and the set design indicators as the output, establish an approximate model; Based on the approximate model, solve the multi-objective optimization problem with the design parameters as optimization variables and the optimal setting of the design indicators as the optimization goal.

4. The method according to claim 3, characterized in that, Use the Kriging method or the polynomial response surface model to establish the approximate model.

5. The method according to claim 3, characterized in that, Use at least one of the multi-objective genetic algorithm, particle swarm algorithm and simulated annealing algorithm to solve the multi-objective optimization problem.

6. The method according to claim 5, wherein The multi-objective genetic algorithm selects parent individuals for the next iteration through the Pareto sorting mechanism, introduces key performance indicators into the Pareto sorting mechanism for individual evaluation, and sorts according to the evaluation results.

7. The method according to claim 6, wherein The introducing key performance indicators into the Pareto sorting mechanism for individual evaluation includes: Extract key performance indicators from the physical response data obtained from the finite element analysis; Eliminate the individuals whose key performance indicators do not meet the set conditions, and calculate the weighted sum of the key performance indicators of each remaining individual as the evaluation value.

8. The method according to claim 6, wherein The multi-objective genetic algorithm generates offspring individuals through crossover operations and mutation operations, and sets the crossover probability and mutation probability according to the structural characteristics of the intelligent bearing multi-layer heterogeneous microsystem.

9. The method according to claim 8, wherein The finite element analysis includes static analysis, dynamic analysis and heat conduction analysis.

10. The method according to claim 1, characterized in that The extracting a number of samples is realized by the optimal Latin hypercube sampling method, and the number of samples is determined according to the dimension of the design parameters and the optimization requirements.

Images