Spring stiffness characteristic simulation optimization method and system

Abstract

The application relates to a spring stiffness characteristic simulation optimization method and system. The spring stiffness characteristic simulation optimization method comprises the following steps: obtaining first stiffness characteristic curve data of a spring through multi-physical field simulation, and extracting first physical form features of the spring based on the first stiffness characteristic curve data; a prediction model between spring input parameters and the first physical form features is established; in the robust design optimization iteration, corresponding second physical form features are obtained according to the geometric design parameters and the uncertainty parameters and based on the prediction model, and second stiffness characteristic curve data of the spring is reconstructed based on the second physical form features; the robust performance evaluation result of the design scheme under the influence of uncertainty is obtained through statistical analysis based on the second stiffness characteristic curve data, and the geometric design parameters are iteratively adjusted according to the robust performance evaluation result until a spring design scheme meeting the robustness requirement is obtained.

Description

Technical Field

[0001] This application relates to the field of spring design optimization, and in particular to a simulation optimization method and system for spring stiffness characteristics. Background Technology

[0002] In modern industrial design, springs, as a key elastic component, directly impact the overall performance of equipment. To ensure stable and reliable operation of products under various complex conditions, engineers typically utilize advanced simulation optimization systems to accurately predict and optimize the stiffness characteristics of springs. However, as design requirements become increasingly stringent, especially when dealing with precision springs subjected to high-speed, high-frequency vibrations and extreme temperature environments, traditional simulation optimization methods often encounter bottlenecks. These bottlenecks stem not only from insufficient consideration of material dynamic response and temperature effects but also from tolerances and assembly deviations in actual manufacturing, which can cause performance fluctuations in theoretically perfect designs in actual products.

[0003] In the process of simulating and optimizing the stiffness characteristics of springs, when it is necessary to simultaneously consider the mechanical response of the material under stress, the changes in material properties at different temperatures, and the impact of manufacturing tolerances and assembly deviations on the final performance, how to efficiently perform robust design optimization has become an urgent technical problem to be solved.

[0004] Specifically, the challenge in spring design optimization lies in balancing the accuracy of simulation calculations that consider various physical phenomena (such as the effect of temperature on the structure) with the enormous computational demands of statistical analysis methods like Monte Carlo methods when assessing uncertainties during optimization iterations. The goal is to find a spring design that satisfies the target stiffness characteristics while also exhibiting good resistance to uncertainties in actual production and assembly within a limited timeframe, avoiding the optimization process becoming impractical due to excessive computational time. Summary of the Invention

[0005] This application provides a simulation optimization method and system for spring stiffness characteristics, which at least solves the problem that existing spring stiffness characteristic simulation optimization methods are difficult to perform robust design optimization efficiently when considering material dynamic response, temperature influence, manufacturing tolerance and assembly deviation, as well as the problem of balancing simulation calculation accuracy and uncertainty assessment computational load in optimization iteration.

[0006] Firstly, this application provides a simulation optimization method for spring stiffness characteristics, the method comprising:

[0007] The first stiffness characteristic curve data of the spring is obtained by multiphysics simulation, and the first physical morphological characteristics of the spring are extracted based on the first stiffness characteristic curve data. The multiphysics simulation is based on the material dynamic response and temperature influence of the spring.

[0008] A predictive model is established between the spring input parameters and the first physical morphological feature, wherein the input parameters include geometric design parameters, operating temperature range, manufacturing tolerances, and assembly deviations;

[0009] In the robust design optimization iteration, based on the geometric design parameters and uncertainty parameters and the prediction model, the corresponding second physical morphological features are obtained, and based on the second physical morphological features, the second stiffness characteristic curve data of the spring is reconstructed.

[0010] Based on the statistical analysis of the second stiffness characteristic curve data, the robust performance evaluation results of the design scheme under the influence of uncertainty are obtained. Based on the robust performance evaluation results, the geometric design parameters are iteratively adjusted until a spring design scheme that meets the robustness requirements is obtained.

[0011] Optionally, in the robust design optimization iteration, based on the geometric design parameters and uncertainty parameters and the prediction model, the corresponding second physical morphological features are obtained, and based on the second physical morphological features, the second stiffness characteristic curve data of the spring is reconstructed, including:

[0012] Based on the geometric design parameters and expected service history information, estimate the cumulative material damage state of the spring;

[0013] Based on the geometric design parameters and the uncertainty parameters, and based on the prediction model, the original physical morphological characteristics of the spring are obtained;

[0014] The original physical morphological characteristics are corrected according to the cumulative damage state of the material to obtain the corrected second physical morphological characteristics;

[0015] Based on the second physical morphological characteristics, the second stiffness characteristic curve data of the spring is reconstructed.

[0016] Optionally, in the robust design optimization iteration, obtaining the corresponding second physical morphological features based on the geometric design parameters and uncertainty parameters and the prediction model includes:

[0017] Evaluate the spatial distribution density of the combination of the geometric design parameters and the uncertainty parameters in the training data of the prediction model;

[0018] When the spatial distribution density is lower than a preset density threshold, a new parameter sub-combination is generated based on the geometric design parameters and the combination of uncertainty parameters;

[0019] The parameter sub-combinations are added to the training dataset of the prediction model, and the prediction model is locally retrained based on the supplemented training dataset to obtain an updated prediction model.

[0020] Based on the updated prediction model, the second physical morphological features corresponding to the geometric design parameters and the uncertainty parameters are obtained.

[0021] Optionally, the step of extracting the first physical morphological features of the spring based on the first stiffness characteristic curve data includes:

[0022] Based on the first stiffness characteristic curve data and the service history information of the spring, the current service stage of the spring is determined;

[0023] For the current service stage, a corresponding feature extraction strategy is selected from a preset feature extraction strategy library, wherein the feature extraction strategy library contains feature extraction methods optimized for different service stages or the material's cumulative damage state.

[0024] Based on the selected feature extraction strategy, the data of the first stiffness characteristic curve is analyzed to identify and quantify the stiffness change characteristics that reflect the evolution of the microstructure of the spring material. The stiffness change characteristics include nonlinear inflection point offset, hysteresis loop area change rate, and stiffness decay factor.

[0025] The stiffness variation characteristics are combined with other physical morphological characteristics of the first stiffness characteristic curve data to obtain the first physical morphological characteristics of the spring.

[0026] Optionally, estimating the cumulative material damage state of the spring based on the geometric design parameters and expected service history information includes:

[0027] Based on the expected service history information, consistency checks and missing data completion are performed to obtain preprocessed service history information;

[0028] Based on the geometric design parameters and the preprocessed service history information, a damage evolution model is dynamically selected and calculated based on the damage evolution model to obtain the material cumulative damage factor of the spring.

[0029] Based on the material cumulative damage factor, the material cumulative damage state of the spring is evaluated and obtained.

[0030] Optionally, determining the current service stage of the spring based on the first stiffness characteristic curve data and the spring's service history information includes:

[0031] The service history information data is standardized and then preprocessed based on the standardized service history information, wherein the preprocessing includes timestamp alignment, interpolation completion, and data cleaning.

[0032] Key parameters are extracted from the preprocessed service history information, including the cumulative load cycle count, high temperature exposure duration, and corrosive environment exposure time.

[0033] The key parameters are compared with preset material damage thresholds and performance degradation curves, and the current service stage of the spring is determined based on the comparison results.

[0034] Optionally, the step of correcting the original physical morphological features based on the cumulative damage state of the material to obtain the corrected second physical morphological features includes:

[0035] Obtain the estimated confidence level information of the cumulative damage state of the material;

[0036] When the estimated confidence level is lower than the preset confidence level threshold, the range of change of each original physical morphological feature under the current damage state is determined based on the material's cumulative damage state.

[0037] Within the range of variation, the original physical morphological features are modified to obtain the modified second physical morphological features;

[0038] The second physical morphological feature is compared with the actual simulation results of a small number of key points, and the second physical morphological feature is corrected based on the comparison results.

[0039] Optionally, the step of extracting the first physical morphological features of the spring based on the first stiffness characteristic curve data includes:

[0040] A global analysis of the first stiffness characteristic curve data is performed to extract features that describe the overall trend and macroscopic shape of the first stiffness characteristic curve data, resulting in global smoothing features.

[0041] Run the local event detection algorithm to identify and quantify the non-smooth locality features on the first stiffness characteristic curve data to obtain local event features;

[0042] The global smoothing feature and the local event feature are combined into a hybrid feature vector, which serves as the first physical morphological feature of the spring.

[0043] Secondly, this application provides a spring stiffness characteristic simulation and optimization system, the system comprising:

[0044] The feature acquisition module is used to acquire the first stiffness characteristic curve data of the spring through multiphysics simulation, and extract the first physical morphological features of the spring based on the first stiffness characteristic curve data, wherein the multiphysics simulation is based on the material dynamic response and temperature influence of the spring.

[0045] The model building module is used to establish a predictive model between the spring input parameters and the first physical morphological features, wherein the input parameters include geometric design parameters, operating temperature range, manufacturing tolerances, and assembly deviations;

[0046] The characteristic curve reconstruction module is used to obtain the corresponding second physical morphological features based on the geometric design parameters and uncertainty parameters and the prediction model during robust design optimization iteration, and to reconstruct the second stiffness characteristic curve data of the spring based on the second physical morphological features.

[0047] The robust performance evaluation module is used to obtain the robust performance evaluation results of the design scheme under the influence of uncertainty based on the statistical analysis of the second stiffness characteristic curve data, and to iteratively adjust the geometric design parameters according to the robust performance evaluation results until a spring design scheme that meets the robustness requirements is obtained.

[0048] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the method provided in the first aspect above.

[0049] Compared with related technologies, the spring stiffness characteristic simulation optimization method and system provided in this application have at least the following technical advantages:

[0050] This application first obtains the first stiffness characteristic curve data of the spring through multiphysics simulation, and extracts the first physical morphological features based on this data. The simulation considers the dynamic response of the material and the influence of temperature. Based on this, a predictive model is established between the spring input parameters (including geometric design parameters, operating temperature range, manufacturing tolerances, and assembly deviations) and the first physical morphological features. Subsequently, in a robust design optimization iteration, based on the geometric design parameters and uncertainty parameters, the second physical morphological features are obtained according to the predictive model, and the second stiffness characteristic curve data is reconstructed. Finally, statistical analysis is performed based on the second stiffness characteristic curve data to obtain robust performance evaluation results, and the geometric design parameters are iteratively adjusted until a spring design scheme that meets the robustness requirements is obtained.

[0051] This application effectively solves the problem in existing technologies where it is difficult to efficiently balance the accuracy of simulation calculations for various physical phenomena with the computational burden of uncertainty assessment during the simulation optimization of spring stiffness characteristics. The method in this application comprehensively considers multiple complex factors such as material dynamic response, temperature influence, manufacturing tolerances, and assembly deviations. By constructing a predictive model and introducing robust design optimization iterations, it significantly reduces the enormous computational resource requirements of traditional statistical analysis methods such as Monte Carlo simulations, avoiding the problem of optimization processes being impractical due to excessive computation time. In summary, this application can efficiently find a spring design scheme that satisfies the target stiffness characteristics and has good resistance to uncertainties in actual production and assembly, thereby improving the efficiency and reliability of spring design.

[0052] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0053] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0054] Figure 1 This is a flowchart illustrating a simulation optimization method for spring stiffness characteristics according to an exemplary embodiment.

[0055] Figure 2 This is a flowchart illustrating step S3 according to an exemplary embodiment.

[0056] Figure 3 This is a flowchart illustrating step S3 according to another exemplary embodiment.

[0057] Figure 4 This is a flowchart illustrating step S314 according to an exemplary embodiment.

[0058] Figure 5 This is a flowchart illustrating step S311 according to an exemplary embodiment.

[0059] Figure 6 This is a flowchart illustrating step S313 according to an exemplary embodiment.

[0060] Figure 7 This is a partial flowchart illustrating step S1 according to an exemplary embodiment.

[0061] Figure 8 This is a block diagram illustrating a spring stiffness characteristic simulation optimization system according to an exemplary embodiment. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0063] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any creative effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0064] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0065] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0066] In related technologies, a key challenge in spring design optimization is balancing the need for high-precision simulations that consider various physical phenomena (such as the effect of temperature on structure) with the computational demands of statistical analysis methods like Monte Carlo methods when assessing uncertainties. The goal is to find a spring design that satisfies the target stiffness characteristics while also exhibiting good resistance to uncertainties in actual production and assembly within a limited timeframe, avoiding the impracticality of the optimization process due to excessive computational time.

[0067] Based on the above, embodiments of the present invention provide a method and system for simulating and optimizing the stiffness characteristics of springs, which will be described in detail below with reference to specific embodiments and accompanying drawings.

[0068] Example 1

[0069] This invention provides a method for simulating and optimizing the stiffness characteristics of a spring. Figure 1 This is a flowchart illustrating a simulation optimization method for spring stiffness characteristics according to an exemplary embodiment. Figure 1 As shown, the method includes:

[0070] S1. Obtain the first stiffness characteristic curve data of the spring through multiphysics simulation, and extract the first physical morphological features of the spring based on the first stiffness characteristic curve data. The multiphysics simulation is based on the material dynamic response and temperature influence of the spring.

[0071] In this embodiment, multiphysics simulation is a simulation technique that can simultaneously consider the interaction of multiple physical phenomena (such as mechanics, thermodynamics, electromagnetism, etc.). In this application, multiphysics simulation is mainly used to simulate the dynamic material response and structural deformation of a spring under the influence of various factors such as force and temperature changes, thereby obtaining more accurate stiffness characteristic curve data.

[0072] Stiffness characteristic curve data refers to the set of stiffness response data of a spring under different loads or displacements. This data is typically presented as a curve, reflecting the elastic performance of the spring. In step S1, the first stiffness characteristic curve data is obtained from initial simulation or experimentation. For example, finite element analysis (FEA) software, combined with a thermo-coupling module, can be used to simulate the deformation and stress distribution of the spring under different loads and temperatures, thereby obtaining its stiffness characteristic curve. In multiphysics simulations, different load steps and temperature gradients can be set to comprehensively capture the mechanical response of the spring. For example, a cyclic load from zero to the maximum working load can be applied to the spring, and its displacement can be recorded at each load point, while considering the change in ambient temperature from room temperature to extreme high temperatures.

[0073] Physical morphological characteristics refer to key parameters extracted from spring stiffness characteristic curve data that can quantitatively describe the physical state and performance of the spring. These characteristics can be the slope of the curve, nonlinear inflection points, hysteresis loop area, etc., and they can reflect the microstructural evolution and macroscopic mechanical behavior of the spring material. Based on the acquired first stiffness characteristic curve data, the first physical morphological characteristics of the spring are extracted. For example, the curve can be mathematically fitted to extract the slope of its linear segment as the initial stiffness, or the location and amplitude of nonlinear inflection points in the curve can be identified.

[0074] S2. Establish a predictive model between the spring input parameters and the first physical characteristics, wherein the input parameters include geometric design parameters, operating temperature range, manufacturing tolerances, and assembly deviations;

[0075] In this embodiment, the prediction model refers to a mathematical model established through machine learning or other data-driven methods, used to describe the mapping relationship between the spring's input parameters (such as geometric design parameters, operating temperature range, manufacturing tolerances, assembly deviations, etc.) and physical morphological characteristics. The prediction model in step S2 can quickly predict the physical morphological characteristics of the spring based on given input parameters, avoiding time-consuming repetitive simulations. Therefore, the prediction model can be constructed using various machine learning algorithms. For example, a Support Vector Machine (SVM) or Gaussian Process Regression (GPR) model can be used, trained with a large amount of simulation or experimental data. During training, geometric design parameters (such as wire diameter, number of turns, free height), operating temperature range, manufacturing tolerances (such as wire diameter tolerance, pitch tolerance), and assembly deviations (such as preload deviation, installation angle deviation) are used as inputs, and the extracted first physical morphological feature is used as the output. For example, spring simulation data under different combinations of geometric design parameters can be collected, and their corresponding stiffness characteristic curves and extracted physical morphological features can be recorded. This data is used to train a neural network model, enabling it to learn the complex nonlinear relationship between input parameters and physical morphological characteristics.

[0076] S3. In the robust design optimization iteration, based on the geometric design parameters and uncertainty parameters and the prediction model, the corresponding second physical form features are obtained, and based on the second physical form features, the second stiffness characteristic curve data of the spring is reconstructed.

[0077] In this embodiment, robust design optimization iteration refers to a design optimization process aimed at finding a design scheme insensitive to uncertainties (such as manufacturing tolerances, assembly deviations, etc.). During this iteration, design parameters are adjusted based on robust performance evaluation results until a preset robustness requirement is met. Monte Carlo simulation methods can be used to introduce uncertainty parameters in robust design optimization iteration. For example, in each iteration, sample values ​​of manufacturing tolerances and assembly deviations are randomly drawn from a preset probability distribution and combined with the current geometric design parameters as input to the prediction model. The prediction model quickly outputs the corresponding second physical characteristics based on these input parameters. For example, in one iteration, the geometric design parameters are set to a set of initial values, and samples of manufacturing tolerances and assembly deviations are randomly drawn from a normal distribution. These parameters are input into a pre-trained prediction model, which immediately outputs the predicted stiffness, nonlinear inflection point, and other second physical characteristics of the spring. Based on the acquired second physical characteristics, interpolation algorithms or curve fitting techniques can be used to reconstruct the second stiffness characteristic curve data of the spring. For example, if the physical morphological features include several key points on the curve and the overall trend parameters of the curve, these features can be transformed into a complete stiffness characteristic curve through spline interpolation or polynomial fitting.

[0078] S4. Based on the statistical analysis of the second stiffness characteristic curve data, the robust performance evaluation results of the design scheme under the influence of uncertainty are obtained. Based on the robust performance evaluation results, the geometric design parameters are iteratively adjusted until a spring design scheme that meets the robustness requirements is obtained.

[0079] In this embodiment, robust performance evaluation results can be obtained through statistical analysis of a large amount of reconstructed second stiffness characteristic curve data. For example, the mean, standard deviation, and probability of exceeding the design margin of the spring stiffness characteristic curve can be calculated under different combinations of uncertainty parameters. These statistics can serve as indicators for robust performance evaluation. For example, in Monte Carlo simulations, the prediction model is run thousands of times, with uncertainty parameters randomly selected each time.

[0080] Then, statistical analysis is performed on all reconstructed stiffness characteristic curves to calculate the average stiffness and fluctuation range within the target load range. If the fluctuation range is too large, it indicates insufficient robustness of the design. Based on the robustness performance evaluation results, optimization algorithms (such as genetic algorithms or particle swarm optimization algorithms) can be used to iteratively adjust the geometric design parameters. For example, if the evaluation results show that the stiffness fluctuation of the current design is large, the optimization algorithm will attempt to adjust geometric design parameters such as wire diameter and number of turns to reduce this fluctuation while ensuring that other performance indicators are met. This iterative process continues until a spring design is found whose performance fluctuation is within an acceptable range under the influence of uncertainty.

[0081] The technical solutions described above combine multiphysics simulation, predictive models, and robust design optimization iterations to form an efficient and accurate spring design optimization process. Firstly, by introducing multiphysics simulation, this application can more accurately capture the dynamic material response and temperature effects of the spring under complex operating conditions, improving the accuracy of initial data. Based on this, by establishing predictive models, this application transforms time-consuming simulation calculations into rapid model predictions, significantly improving the efficiency of optimization iterations. Subsequently, in the robust design optimization iterations, this application can systematically consider uncertainties. Finally, through statistical analysis, the robustness performance of the design scheme is quantitatively evaluated, thereby finding a spring design scheme with good resistance to uncertainties in actual production and assembly. Therefore, this application not only improves the accuracy and reliability of spring design but also significantly shortens the design cycle, providing an effective solution for precision spring design under complex operating conditions.

[0082] In one possible design, Figure 2 This is a flowchart illustrating step S3 according to an exemplary embodiment. (Refer to the attached diagram.) Figure 2 Step S3 includes:

[0083] S311. Estimate the cumulative material damage state of the spring based on the geometric design parameters and expected service history information;

[0084] In this embodiment, geometric design parameters refer to the inherent properties of the spring, such as its structural dimensions and material type; the expected service history information includes the number of load cycles the spring may experience in the actual working environment, the temperature variation range, and the exposure time to corrosive media. By comprehensively analyzing this information, the degree of fatigue, creep, corrosion, and other damage that the spring material may experience during long-term use can be quantitatively assessed, thereby obtaining the cumulative damage state of the material.

[0085] S312. Based on the geometric design parameters and uncertainty parameters, and based on the prediction model, obtain the original physical morphological characteristics of the spring;

[0086] In this embodiment, the original physical morphological characteristics are predictions based on ideal conditions or without considering material damage, reflecting the theoretical physical morphology of the spring under specific input parameters.

[0087] S313. Correct the original physical morphological characteristics based on the cumulative damage state of the material to obtain the corrected second physical morphological characteristics;

[0088] In this embodiment, the correction process aims to take into account the impact of material damage on the physical morphology of the spring. For example, material damage may lead to a reduction in the effective stiffness of the spring or microscopic changes in its geometric dimensions. By introducing damage conditions for correction, the second physical morphological feature can more accurately reflect the true state of the spring under actual service conditions.

[0089] S314. Based on the second physical morphological characteristics, reconstruct the second stiffness characteristic curve data of the spring;

[0090] In this embodiment, the reconstruction process uses the corrected physical morphological features as input and generates curve data that reflects the actual stiffness response of the spring under the condition of considering material damage through a specific mathematical model or simulation method.

[0091] The technical solution described above, by introducing an estimation and correction mechanism for the cumulative material damage state, can significantly improve the accuracy and reliability of the simulation optimization results of spring stiffness characteristics. By incorporating the cumulative material damage state into the prediction model, the obtained second physical morphological characteristics can more realistically reflect the performance degradation of the spring under long-term service conditions, thereby avoiding prediction bias caused by ignoring material damage. This not only helps in designing springs that meet performance requirements throughout their entire lifespan but also effectively reduces the risk of failure caused by material performance degradation, improving the robustness and reliability of spring design and providing more accurate data support for spring life prediction and maintenance strategy formulation.

[0092] In one example, suppose we need to perform robust design optimization for a spring used in a car suspension system. This spring will be subjected to cyclic loads and temperature changes during its service life.

[0093] First, based on the spring's geometric design parameters (e.g., wire diameter, number of coils, material grade) and expected service history information (e.g., expected mileage, road condition type, ambient temperature range), the cumulative damage state of the spring material is estimated. The estimation may involve fatigue damage accumulation models and creep damage models to quantify the extent of fatigue crack initiation and propagation or the cumulative amount of plastic deformation in the material.

[0094] Next, based on the current geometric design parameters and uncertainty parameters (e.g., manufacturing tolerances, assembly deviations), and using a pre-trained prediction model, the original physical characteristics of the spring under ideal conditions are obtained, such as its initial geometric dimensions and material elastic modulus.

[0095] Subsequently, the original physical morphological characteristics are modified using the previously estimated cumulative material damage state. For example, if a certain degree of fatigue damage is estimated in the material, the effective elastic modulus of the material can be reduced according to the degree of damage, or the effective number of coils of the spring can be fine-tuned, thereby obtaining the modified second physical morphological characteristics.

[0096] Finally, based on these revised second physical characteristics, the second stiffness characteristic curve data of the spring considering material damage is reconstructed. For example, by inputting the revised physical characteristics through finite element analysis or analytical models, a stiffness-displacement curve that more closely approximates the actual service condition is obtained. In this way, designers can more accurately evaluate the performance of different design schemes under long-term service during the optimization iteration process, thereby selecting a more robust spring design scheme.

[0097] In another possible design, Figure 3 This is a flowchart illustrating step S3 according to another exemplary embodiment. (Refer to the attached document.) Figure 3 Step S3 includes:

[0098] S321. Evaluate the spatial distribution density of the combination of geometric design parameters and uncertainty parameters in the training data of the prediction model;

[0099] In this embodiment, the spatial distribution density is evaluated using statistical methods or machine learning techniques, such as kernel density estimation (KDE), cluster analysis, or nearest neighbor distance calculation, to quantify the sparsity of the current parameter combination to be evaluated in the existing training data distribution, thereby identifying areas where the prediction model may have prediction uncertainty or decreased accuracy.

[0100] S322. When the spatial distribution density is lower than the preset density threshold, generate a new parameter sub-combination based on the combination of geometric design parameters and uncertainty parameters.

[0101] In this embodiment, the preset density threshold can be set according to the actual application scenario, model accuracy requirements, and available computing resource limits. Generating new parameter sub-combinations can be achieved in various ways, such as performing small-scale random sampling near the current parameter combination, generating new data points based on Latin Hypercube Sampling (LHS) or Monte Carlo Sampling, or using an active learning strategy to select the most informative new data points, thereby specifically increasing the model's data coverage in sparse regions.

[0102] S323. Add the parameter sub-combinations to the training dataset of the prediction model, and retrain the prediction model locally based on the supplemented training dataset to obtain the updated prediction model.

[0103] In this embodiment, local retraining refers to incrementally training or fine-tuning the prediction model using only the newly added parameter sub-combinations and a portion of the original training data nearby, rather than training the entire model from scratch. Local retraining helps improve the model's prediction accuracy in a specific region without significantly increasing computational costs, thereby enabling the prediction model to better adapt to the new parameter space and improve its generalization ability.

[0104] S324. Based on the updated prediction model, obtain the second physical morphological features corresponding to the geometric design parameters and uncertainty parameters;

[0105] In this embodiment, by using a prediction model that has been locally retrained and updated, more accurate and reliable second physical morphological features can be obtained, thereby significantly improving the robustness and accuracy of the spring stiffness characteristic simulation optimization method.

[0106] The technical solution described above, by dynamically evaluating the spatial distribution density of parameters and adaptively retraining the prediction model locally, can significantly improve the robustness and accuracy of the spring stiffness characteristic simulation optimization method. Specifically, by dynamically evaluating the spatial distribution density of parameter combinations and performing local adaptive updates to the prediction model, deviations in optimization results caused by inaccurate predictions in sparse training data regions are avoided. This makes the second physical morphological features obtained in the robust design optimization iteration more accurate, improving the reliability of subsequent reconstruction of the spring's second stiffness characteristic curve data and the accuracy of robust performance evaluation results.

[0107] In one example, suppose that during the simulation optimization of spring stiffness characteristics, the prediction model is initially trained with a finite set of geometric design parameters (e.g., wire diameter, number of turns) and uncertainty parameters (e.g., material elastic modulus fluctuation range, manufacturing tolerances). When the optimization algorithm explores a new design point, such as a spring design with a thin wire diameter, a high number of turns, and a large material elastic modulus fluctuation range, this parameter combination may be very rare in the original training dataset, resulting in an extremely low spatial distribution density.

[0108] At this point, the method of this application first evaluates the spatial distribution density of the new parameter combination in the training data of the prediction model. If the density is found to be lower than a preset threshold, the system will automatically generate a series of parameter sub-combinations around the new parameter combination. For example, dozens of new parameter points can be generated near the range of fine wire diameter, high coil count, and high fluctuation, by slightly adjusting the fluctuation range of wire diameter, coil count, and elastic modulus. These new parameter sub-combinations are then added to the training dataset of the prediction model. Next, the prediction model will use these new data for local retraining to enhance its predictive ability in that specific region. In this way, the updated prediction model can more accurately predict the second physical morphological characteristics of the fine wire diameter, high coil count spring under the influence of uncertainty, thereby ensuring that the subsequent robust performance evaluation results are more reliable and avoiding potential design risks caused by insufficient model extrapolation ability.

[0109] In one possible design, Figure 4 This is a flowchart illustrating step S314 according to an exemplary embodiment. (Refer to the attached diagram.) Figure 4 Step S314 includes:

[0110] S3141. Based on the first stiffness characteristic curve data and the service history information of the spring, determine the current service stage of the spring;

[0111] In this embodiment, the service history information refers to data on various environmental and load conditions experienced by the spring during actual use, such as the cumulative load cycle count, high-temperature exposure duration, and corrosive environment exposure time. By analyzing this historical data, the current service stage of the spring can be estimated, such as whether it is in the initial break-in period, stable operation period, fatigue damage accumulation period, or nearing failure.

[0112] S3142. For the current service stage, select the corresponding feature extraction strategy from the preset feature extraction strategy library, wherein the feature extraction strategy library contains feature extraction methods optimized for different service stages or material cumulative damage states.

[0113] In this embodiment, the feature extraction strategy library is a pre-established knowledge base that stores various feature extraction algorithms or models optimized for different service stages or material cumulative damage states. For example, for the initial break-in period, the focus may be on identifying stiffness changes caused by manufacturing defects or initial stress relaxation; for the fatigue damage accumulation period, the focus may be on capturing nonlinear stiffness changes caused by microcrack initiation and propagation.

[0114] S3143. Based on the selected feature extraction strategy, analyze the data of the first stiffness characteristic curve, identify and quantify the stiffness change characteristics that reflect the evolution of the microstructure of the spring material, wherein the stiffness change characteristics include nonlinear inflection point offset, hysteresis loop area change rate and stiffness decay factor.

[0115] In this embodiment, stiffness variation characteristics are key indicators reflecting the microstructural evolution of spring materials. Specifically, the nonlinear inflection point offset refers to the change in the position of the inflection point in the stiffness characteristic curve from the linear region to the nonlinear region, which can indicate the change in the initiation point of plastic deformation or damage in the material; the hysteresis loop area change rate refers to the change in the area of ​​the hysteresis loop formed by the stiffness characteristic curve under cyclic loading, which can reflect the changes in energy dissipation and damping characteristics inside the material and is closely related to the accumulation of damage inside the material; the stiffness decay factor directly quantifies the degree of decrease in the elastic modulus or overall stiffness of the spring during service. These characteristics can reveal the microstructural changes of spring materials from different dimensions, such as dislocation motion, grain boundary slip, micropore formation, or crack propagation.

[0116] S3144. Combine the stiffness variation characteristics with other physical morphological characteristics of the first stiffness characteristic curve data to obtain the first physical morphological characteristics of the spring.

[0117] In this embodiment, the stiffness change characteristics in step S3143 are combined with other physical morphological characteristics (such as geometric dimensions, material type, surface treatment, etc.) of the first stiffness characteristic curve data of the spring to form a comprehensive first physical morphological feature vector for subsequent prediction models.

[0118] The technical solution of the above embodiments, by introducing the concepts of service history information and current service stage, and combining them with a feature extraction strategy library, achieves adaptive and refined extraction of the first physical morphological features of the spring, significantly improving the accuracy and pertinence of the extraction of the first physical morphological features of the spring. Especially when the spring has experienced complex service history and material degradation, the prediction model can more accurately capture the real behavior of the spring, thereby improving the reliability and efficiency of robust design optimization iteration.

[0119] In one possible design, Figure 5 This is a flowchart illustrating step S311 according to an exemplary embodiment. (Refer to the attached document.) Figure 5 Step S311 includes:

[0120] S3111. Based on the expected service history information, perform consistency checks and missing data completion to obtain preprocessed service history information;

[0121] In this embodiment, the expected service history information refers to various operating condition data that the spring may experience during actual use, such as load spectrum, temperature changes, and environmental corrosion. Consistency checks are performed on this information to identify and correct logical conflicts or outliers in the data, such as sudden unreasonable peaks or troughs in load data. Missing data completion refers to the use of interpolation, regression, or other statistical methods to fill in missing data when some service history information is not fully recorded, ensuring data integrity and continuity. Through these preprocessing steps, more accurate and reliable preprocessed service history information can be obtained.

[0122] S3112. Based on the geometric design parameters and pre-processed service history information, dynamically select the damage evolution model and calculate the material cumulative damage factor of the spring based on the damage evolution model.

[0123] In this embodiment, the damage evolution model is a mathematical model describing how the degree of damage to a material changes over time or the number of cycles under specific load, temperature, and environmental conditions. This model can be based on physical mechanisms (such as fatigue damage models or creep damage models) or on empirical statistics. Dynamically selecting the damage evolution model involves the system intelligently selecting the most suitable damage evolution model from a pre-set model library based on the spring's geometric design parameters (such as material type and surface treatment) and pre-processed service history information (such as main failure modes and service environment characteristics). For example, a fatigue damage model might be selected for a spring primarily subjected to cyclic loads; a creep damage model might be selected for a spring subjected to prolonged high-temperature environments. Calculations based on the selected model quantify the cumulative material damage factor of the spring, which reflects the degree of damage the spring experiences during service.

[0124] S3113. Based on the material cumulative damage factor, evaluate the material cumulative damage state of the spring.

[0125] In this embodiment, the material cumulative damage factor is a quantitative indicator, such as a value between 0 and 1, where 0 represents no damage and 1 represents complete failure. Based on this material cumulative damage factor, the material cumulative damage state of the spring can be evaluated. For example, the damage factor can be compared with a preset damage threshold to classify the damage state into different levels, such as "minor damage," "moderate damage," or "severe damage."

[0126] The technical solution described above effectively improves the quality and reliability of input data by rigorously checking the consistency of expected service history information and completing missing data, thus avoiding estimation deviations caused by data defects. Based on this, according to the spring's geometric design parameters and pre-processed service history information, the damage evolution model that best matches the actual damage mechanism is dynamically selected, ensuring the scientific rigor and relevance of the damage calculation. Subsequently, a damage evolution model highly matched to actual working conditions is adopted, enabling the calculation results of the material cumulative damage factor to accurately reflect the true damage degree of the spring. Finally, based on this accurate damage factor, the material cumulative damage state of the spring can be precisely assessed.

[0127] In one example, suppose we need to perform robust design optimization for a coil spring used in an automotive suspension system. First, we collect the expected service history information of the spring under different road conditions, loads, and temperatures, including the number of load cycles, maximum stress, ambient temperature, etc.

[0128] Specifically, during consistency checks, if abnormal fluctuations in load data are identified within a certain time period, such as a sudden recording of extremely high impact loads under stable driving conditions, this could be due to sensor malfunction or data transmission errors. In such cases, the system flags the abnormal data and corrects or removes it based on the trend of preceding and following data. For missing data completion, if temperature data for a certain time period is missing, the system can interpolate and complete the data based on temperature data from adjacent time points, combined with an environmental temperature model, ensuring the continuity of temperature data. After these preprocessing steps, a complete and reliable preprocessed service history information is obtained.

[0129] Subsequently, based on the material of the helical spring (e.g., high-strength alloy steel) and the service history information from pretreatment, a damage evolution model is dynamically selected. For example, if the service history information shows that the spring mainly bears high-frequency cyclic loads, the system may select a fatigue damage accumulation model based on Miner's law. If the spring is also exposed to a high-temperature environment for a long time, the system may further consider a creep damage model and couple it with the fatigue damage model. Based on the selected model, the system calculates the material cumulative damage factor of the spring over the entire expected service life, for example, a result of 0.65.

[0130] Finally, based on the material's cumulative damage factor of 0.65, the system evaluates and obtains the spring's material cumulative damage state. For example, the system can preset a damage factor below 0.3 as minor damage, between 0.3 and 0.7 as moderate damage, and above 0.7 as severe damage. Therefore, the spring's material cumulative damage state is evaluated as "moderate damage." This precise damage state information will be used to subsequently modify the spring's original physical characteristics, ensuring that the impact of material damage on spring performance is fully considered in robust design optimization iterations.

[0131] In one possible design, step S3141 includes:

[0132] S31411. Standardize the service history information data and preprocess the service history information based on the standardized data. The preprocessing includes timestamp alignment, interpolation completion and data cleaning.

[0133] In this embodiment, service history information data is standardized to eliminate dimensional differences caused by different data sources or measurement units, ensuring data consistency. Based on this, preprocessing is performed on the standardized service history information to improve data quality and usability. Specifically, timestamp alignment synchronizes data collected by different sensors or recording systems according to a unified time base to ensure the accuracy of the data's time series; interpolation completion calculates missing values ​​from known data points using mathematical methods (e.g., linear interpolation, spline interpolation) to maintain data continuity when missing values ​​exist in the data series; and data cleaning identifies, corrects, or removes errors, outliers, or inconsistencies in the data to improve its accuracy and reliability.

[0134] S31412. Extract key parameters from the preprocessed service history information, including the cumulative load cycle count, high temperature exposure duration, and corrosive environment exposure time.

[0135] In this embodiment, the key parameters are physical quantities or events that have a decisive impact on the service condition of the spring. Specifically, the cumulative load cycle count is an important indicator for measuring the degree of fatigue damage to the spring, reflecting the periodic loads the spring endures during its service life; high-temperature exposure time refers to the cumulative time the spring spends outside its design operating temperature range, which may lead to material creep, oxidation, or changes in microstructure, thereby affecting its mechanical properties; corrosive environment exposure time refers to the cumulative time the spring spends in corrosive media (such as acids, alkalis, salt spray, etc.), which may lead to damage such as material surface corrosion and stress corrosion cracking. By quantifying these key parameters, the actual service condition and potential damage risks of the spring can be assessed more accurately.

[0136] S31413. Compare the key parameters with the preset material damage threshold and performance degradation curve, and determine the current service stage of the spring based on the comparison results.

[0137] In this embodiment, the material damage threshold is the critical value at which the material properties begin to significantly decline or fail under different damage mechanisms, such as the fatigue crack initiation threshold and creep strain threshold. The performance degradation curve describes the trend of the spring's key properties, such as stiffness and load-bearing capacity, changing with time or load accumulation under different service conditions. For example, when the cumulative load cycle exceeds a certain fatigue damage threshold, the spring may enter the fatigue damage accumulation stage; when the high-temperature exposure time reaches a certain level, it may enter the material aging stage. Through this comparison, the current service stage of the spring can be objectively and quantitatively determined.

[0138] The technical solution described above effectively addresses potential quality issues with raw data by systematically standardizing and preprocessing service history information, ensuring the accuracy of subsequent analysis. Subsequently, by extracting key parameters directly related to spring material damage and comparing them with preset damage thresholds and performance degradation curves, the assessment of the spring's current service stage becomes more scientific and quantitative. This allows for a more precise identification of the spring's damage state, such as whether it is in the initial wear, fatigue accumulation, material aging, or critical failure stage.

[0139] In one possible design, Figure 6 This is a flowchart illustrating step S313 according to an exemplary embodiment. (Refer to the attached diagram.) Figure 6 Step S313 includes:

[0140] S3131. Obtain the estimated confidence level information of the cumulative damage state of the material;

[0141] In this embodiment, the evaluation results are quantified by comprehensively assessing the completeness, accuracy, timeliness of the data source (e.g., service history information) used to estimate the cumulative damage state of materials, as well as the applicability of the damage evolution model employed. This confidence level information can be a percentage value, a probability value, or a confidence interval to reflect the reliability of the current estimate of the cumulative damage state of the materials. For example, when there is a large amount of missing or noisy service history data, the estimation confidence level will be low.

[0142] S3132. When the estimated confidence level is lower than the preset confidence level threshold, determine the range of change of each original physical morphological feature under the current damage state based on the cumulative damage state of the material.

[0143] In this embodiment, when the estimated confidence level is lower than a preset confidence threshold, it means that there is a high degree of uncertainty in the estimation of the current cumulative damage state of the material. The preset confidence threshold can be set according to the actual application scenario and the requirements for the accuracy of the results, for example, it can be set to 70% or 80%. When it is lower than this threshold, the system will initiate a more conservative or more refined correction strategy. Subsequently, based on known material damage mechanisms, empirical data, or more detailed physical models, the reasonable fluctuation range of the original physical morphological characteristics that may occur under a given cumulative damage state of the material is predicted. For example, material fatigue damage may cause a decrease in spring stiffness, and the magnitude of the decrease will have an expected range. This range of variation can be a symmetrical range or an asymmetrical range, the purpose of which is to provide a reasonable boundary for subsequent corrections and avoid over-correction or under-correction.

[0144] S3133. Within the range of variation, the original physical morphological features are modified to obtain the modified second physical morphological features.

[0145] In this embodiment, within the variation range determined in step S3132, the original physical morphological features are adjusted using methods such as interpolation, weighted averaging, sampling based on probability distribution, or optimization algorithms. For example, weighted corrections can be performed within the variation range based on estimated confidence information; the lower the confidence level, the greater the correction magnitude may be, or the correction result may tend towards a more conservative estimate.

[0146] S3134. Compare the second physical morphological features with the actual simulation results of a small number of key points, and correct the second physical morphological features based on the comparison results.

[0147] In this embodiment, to further verify and improve the accuracy of the correction results, a small number of key points of the spring under specific load, temperature, or deformation conditions are selected for high-precision multiphysics simulation to obtain the actual physical characteristics of these key points. Subsequently, the second physical characteristics obtained through the above steps are compared with these actual simulation results. If there is a deviation, the second physical characteristics are fine-tuned or corrected according to the magnitude and direction of the deviation. This allows for local verification and optimization of the correction process using high-precision simulation results, ensuring that the corrected features are closer to reality.

[0148] The technical solution of the above embodiments, by introducing an assessment of the confidence level of the cumulative damage state estimation of materials and using high-precision simulation data to locally verify and refine the correction results, compensates for the potential errors caused by the uncertainty of damage state estimation and further improves the accuracy of the correction results.

[0149] In one example, suppose a spring, during its service life, has some service history information missing due to sensor failure or incomplete data recording. As a result, the confidence level of the material's cumulative damage state estimated by existing methods is only 60%, which is lower than the preset 75% confidence threshold.

[0150] In this scenario, the proposed solution first obtains a 60% confidence level estimate. Since this confidence level is below a threshold, based on the currently estimated cumulative material damage state (e.g., mild fatigue damage), and combined with a material property database and empirical models, the possible range of variation of the original physical characteristics of the spring (e.g., wire diameter, number of coils, pitch, etc.) under mild fatigue damage is determined. For example, the wire diameter may fluctuate within ±0.5%, the number of coils may remain unchanged, and the pitch may fluctuate within ±0.2%. Subsequently, the original physical characteristics are corrected within these determined ranges of variation. For example, the original wire diameter is adjusted to its expected average value under mild fatigue damage using a weighted average method to obtain the corrected second physical characteristics. Finally, to further verify the correction results, two key points of the spring under maximum and minimum compressive loads are selected for high-precision finite element simulation to obtain the actual physical characteristics corresponding to these two key points. The corrected second physical morphological feature is compared with the actual simulation results of these two key points. If a slight deviation is found in the corrected wire diameter under the maximum compressive load, the second physical morphological feature is fine-tuned based on the comparison results until it highly matches the simulation results of the key points. Through this series of steps, even when the uncertainty in damage state estimation is high, a more accurate and reliable second physical morphological feature can be obtained.

[0151] In one possible design, Figure 7 This is a partial flowchart illustrating step S1 according to an exemplary embodiment. (Refer to the attached diagram.) Figure 7 Step S1 includes:

[0152] S11. Perform a global analysis on the first stiffness characteristic curve data, extract features that describe the overall trend and macroscopic shape of the first stiffness characteristic curve data, and obtain global smoothing features;

[0153] In this embodiment, global analysis refers to a macroscopic examination of the entire first stiffness characteristic curve data, aiming to capture its overall trend, average level, and fluctuation range. For example, methods such as curve fitting (e.g., polynomial fitting, spline fitting), statistical moments (e.g., mean, variance, skewness, kurtosis), Fourier transform, or wavelet transform can be used to extract parameters reflecting the overall characteristics of the curve. Therefore, global smoothing features refer to the characteristics obtained through global analysis that describe the overall trend and macroscopic properties of the first stiffness characteristic curve data. These features typically reflect the average stiffness, linearity, and overall nonlinearity of the spring within different load or displacement ranges. For example, they can be the coefficients of the fitted curve, the values ​​of statistical moments, or the amplitude of specific frequency components.

[0154] S12. Run the local event detection algorithm to identify and quantify the non-smooth locality features on the first stiffness characteristic curve data to obtain the local event features;

[0155] In this embodiment, the local event detection algorithm refers to an algorithm used to identify events occurring in a specific local region of the first stiffness characteristic curve data that are inconsistent with the overall trend or have significant changes. These algorithms may include, but are not limited to, peak and valley detection, inflection point detection, outlier detection, and edge detection, thereby accurately capturing abrupt changes or discontinuities in the curve that may be caused by material phase transformation, microcrack initiation, local plastic deformation, etc.

[0156] Non-smooth locality features refer to the characteristics of local regions that exhibit non-smoothness, abrupt changes, or significant variations on the first stiffness characteristic curve data, as identified by local event detection algorithms. These features are usually closely related to microstructural changes in the spring material, damage accumulation, or local responses under specific operating conditions. For example, they may be nonlinear inflection points of the stiffness curve, the start or end points of hysteresis loops, or the magnitude and location of sudden drops or increases in local stiffness.

[0157] Local event characteristics refer to the results of quantitative descriptions of non-smooth local characteristics. Examples include nonlinear inflection point offset, hysteresis loop area change rate, and local stiffness attenuation factor. These quantitative indicators can more accurately reflect the performance changes of a spring in specific local regions.

[0158] S13. Combine the global smoothing features with the local event features into a hybrid feature vector, which serves as the first physical morphological feature of the spring.

[0159] In this embodiment, the hybrid feature vector refers to a comprehensive feature representation formed by effectively combining global smooth features and local event features. This combination can be a simple concatenation or a weighted fusion, aiming to comprehensively and multidimensionally characterize the primary physical features of the spring.

[0160] The technical solution of the above embodiment, by performing global analysis and local event detection on the first stiffness characteristic curve data, can not only grasp the macroscopic stiffness characteristics of the spring, but also accurately capture the non-smooth local features caused by material damage, phase transformation, etc. Thus, the obtained hybrid feature vector can more completely characterize the actual physical state of the spring.

[0161] In summary, the spring stiffness characteristic simulation optimization method provided by this invention comprehensively considers various complex factors such as material dynamic response, temperature influence, manufacturing tolerances, and assembly deviations. By constructing a predictive model and introducing robust design optimization iterations, it significantly reduces the enormous computational resource requirements of traditional statistical analysis methods such as Monte Carlo simulations, avoiding the problem of optimization processes being impractical due to excessive computation time. This efficiently finds a spring design scheme that satisfies the target stiffness characteristics while exhibiting good resistance to uncertainties in actual production and assembly, thus improving the efficiency and reliability of spring design.

[0162] Example 2

[0163] Embodiment 2 of this application provides a simulation optimization system for spring stiffness characteristics. Figure 8 This is a block diagram illustrating a spring stiffness characteristic simulation and optimization system according to an exemplary embodiment. (See attached diagram.) Figure 8 The system includes:

[0164] The feature acquisition module 01 is used to acquire the first stiffness characteristic curve data of the spring through multiphysics simulation, and extract the first physical morphological features of the spring based on the first stiffness characteristic curve data. The multiphysics simulation is based on the material dynamic response and temperature influence of the spring.

[0165] Model building module 02 is used to establish a predictive model between the spring input parameters and the first physical morphological features. The input parameters include geometric design parameters, operating temperature range, manufacturing tolerances, and assembly deviations.

[0166] The characteristic curve reconstruction module 03 is used to obtain the corresponding second physical morphological features based on geometric design parameters and uncertainty parameters and based on the prediction model in the robust design optimization iteration, and to reconstruct the second stiffness characteristic curve data of the spring based on the second physical morphological features.

[0167] The robust performance evaluation module 04 is used to obtain the robust performance evaluation results of the design scheme under the influence of uncertainty based on the statistical analysis of the second stiffness characteristic curve data, and to iteratively adjust the geometric design parameters according to the robust performance evaluation results until a spring design scheme that meets the robustness requirements is obtained.

[0168] In summary, the spring stiffness characteristic simulation and optimization system provided in this invention comprehensively considers various complex factors such as material dynamic response, temperature influence, manufacturing tolerances, and assembly deviations. By constructing a predictive model and introducing robust design optimization iterations, it significantly reduces the enormous computational resource requirements of traditional statistical analysis methods such as Monte Carlo simulations, avoiding the problem of optimization processes being impractical due to excessive computation time. This efficiently finds a spring design scheme that satisfies the target stiffness characteristics while exhibiting good resistance to uncertainties in actual production and assembly, thus improving the efficiency and reliability of spring design.

[0169] Example 3

[0170] Embodiment 3 of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method provided in Embodiment 1.

[0171] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0172] In a possible implementation, the present invention can also be implemented as a program product comprising program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps of implementing the method provided in Embodiment 1.

[0173] The program code for executing the present invention can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0174] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0175] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A spring rate characteristic emulation optimization method, characterized by, The method comprises: obtaining first stiffness characteristic curve data of the spring through multi-physics field simulation, and extracting first physical morphology features of the spring based on the first stiffness characteristic curve data, wherein the multi-physics field simulation is based on material dynamic response and temperature influence of the spring; establishing a prediction model between spring input parameters and the first physical morphology features, wherein the input parameters include geometric design parameters, working temperature range, manufacturing tolerance, and assembly deviation; in a robust design optimization iteration, obtaining corresponding second physical morphology features according to the geometric design parameters and uncertainty parameters and based on the prediction model, and reconstructing second stiffness characteristic curve data of the spring based on the second physical morphology features; based on statistical analysis of the second stiffness characteristic curve data, obtaining robust performance evaluation results of a design scheme under the influence of uncertainty, and iteratively adjusting the geometric design parameters according to the robust performance evaluation results until a spring design scheme meeting the robustness requirement is obtained; wherein the prediction model adopts a support vector machine or a Gaussian process regression model, and the training step comprises: collecting spring simulation data under different geometric design parameter combinations, and recording the corresponding stiffness characteristic curves and extracted physical morphology features; using the spring simulation data under different geometric design parameter combinations, the corresponding stiffness characteristic curves and the extracted physical morphology features to train the prediction model, so that the prediction model learns the complex nonlinear relationship between the input parameters and the first physical morphology features.

2. The spring rate characteristic emulation optimization method according to claim 1, characterized by, In the robust design optimization iteration, the corresponding second physical morphology features are obtained according to the geometric design parameters and uncertainty parameters and based on the prediction model, and the second stiffness characteristic curve data of the spring is reconstructed based on the second physical morphology features, comprising: estimating the material cumulative damage state of the spring according to the geometric design parameters and expected service history information; obtaining the original physical morphology features of the spring according to the geometric design parameters and the uncertainty parameters and based on the prediction model; correcting the original physical morphology features according to the material cumulative damage state to obtain the corrected second physical morphology features; reconstructing the second stiffness characteristic curve data of the spring based on the second physical morphology features.

3. The spring rate characteristic emulation optimization method according to claim 1, characterized by, In the robust design optimization iteration, the corresponding second physical morphology features are obtained according to the geometric design parameters and uncertainty parameters and based on the prediction model, comprising: evaluating the spatial distribution density of the combination of the geometric design parameters and the uncertainty parameters in the training data of the prediction model; when the spatial distribution density is lower than a preset density threshold, generating a new parameter sub-combination according to the combination of the geometric design parameters and the uncertainty parameters; supplementing the parameter sub-combination to the training data set of the prediction model, and locally retraining the prediction model based on the supplemented training data set to obtain an updated prediction model; Based on the updated prediction model, a second physical morphology feature corresponding to the geometric design parameter and the uncertainty parameter is obtained.

4. The spring rate characteristic emulation optimization method according to claim 2, characterized by, The first physical morphology feature of the spring is extracted based on the first stiffness characteristic curve data, including: Based on the first stiffness characteristic curve data and the service history information of the spring, the current service stage of the spring is determined; For the current service stage, a corresponding feature extraction strategy is selected from a preset feature extraction strategy library, wherein the feature extraction strategy library contains feature extraction methods optimized for different service stages or material cumulative damage states; According to the selected feature extraction strategy, the first stiffness characteristic curve data is analyzed to identify and quantify the stiffness change feature reflecting the evolution of the spring material microstructure, wherein the stiffness change feature includes nonlinear inflection point offset, hysteresis loop area change rate and stiffness attenuation factor; The stiffness change feature and other physical morphology features of the first stiffness characteristic curve data are combined to obtain the first physical morphology feature of the spring.

5. The spring rate characteristic emulation optimization method according to claim 2, characterized by, The material cumulative damage state of the spring is estimated according to the geometric design parameter and the expected service history information, including: Based on the expected service history information, consistency check and missing data completion are performed to obtain preprocessed service history information; According to the geometric design parameter and the preprocessed service history information, a damage evolution model is dynamically selected and calculated based on the damage evolution model to obtain the material cumulative damage factor of the spring; Based on the material cumulative damage factor, the material cumulative damage state of the spring is evaluated.

6. The spring rate characteristic emulation optimization method of claim 4, wherein The current service stage of the spring is determined based on the first stiffness characteristic curve data and the service history information of the spring, including: Standardize the service history information data and preprocess based on the standardized service history information, wherein the preprocessing includes timestamp alignment, interpolation completion and data cleaning; Extract the key parameters in the preprocessed service history information, wherein the key parameters include cumulative load cycle times, high temperature exposure time and corrosion environment exposure time; Compare the key parameters with the preset material damage threshold and performance attenuation curve, and determine the current service stage of the spring according to the comparison result.

7. The spring rate characteristic emulation optimization method of claim 2, wherein The original physical morphology feature is corrected according to the material cumulative damage state to obtain the corrected second physical morphology feature, including: Obtain the estimation confidence information of the material cumulative damage state; When the estimation confidence is lower than the preset confidence threshold, determine the change range of each original physical morphology feature under the current damage state according to the material cumulative damage state; In the change range, the original physical morphology feature is corrected to obtain the corrected second physical morphology feature; Compare the second physical morphology feature with the actual simulation result of a small number of key points, and correct the second physical morphology feature according to the comparison result.

8. The method of claim 1, wherein, The first physical morphology feature of the spring is extracted based on the first stiffness characteristic curve data, including: The first stiffness characteristic curve data is globally analyzed to extract a global smooth feature describing the overall trend and macroscopic shape of the first stiffness characteristic curve data; A local event detection algorithm is run to identify and quantify non-smooth local features on the first stiffness characteristic curve data, obtaining a local event feature; The global smooth feature and the local event feature are combined into a hybrid feature vector as the first physical morphology feature of the spring.

9. A spring rate characteristic emulation optimization system, comprising: The system comprises: A feature acquisition module configured to obtain first stiffness characteristic curve data of a spring through multi-physics simulation, and extract a first physical morphology feature of the spring based on the first stiffness characteristic curve data, wherein the multi-physics simulation is based on material dynamic response and temperature influence of the spring; A model construction module configured to establish a prediction model between input parameters of the spring and the first physical morphology feature, wherein the input parameters include geometric design parameters, working temperature range, manufacturing tolerance, and assembly deviation; A characteristic curve reconstruction module configured to obtain a corresponding second physical morphology feature according to the geometric design parameters and uncertainty parameters and based on the prediction model in a robust design optimization iteration, and reconstruct second stiffness characteristic curve data of the spring based on the second physical morphology feature; A robust performance evaluation module configured to obtain a robust performance evaluation result of a design scheme under the influence of uncertainty based on statistical analysis of the second stiffness characteristic curve data, and iteratively adjust the geometric design parameters according to the robust performance evaluation result until a spring design scheme meeting the robustness requirement is obtained; wherein the prediction model adopts a support vector machine or a Gaussian process regression model, and the training steps include: Taking the geometric design parameters, working temperature range, manufacturing tolerance, and assembly deviation as inputs, and taking the first physical morphology feature as output, collecting spring simulation data under different combinations of geometric design parameters, and recording the corresponding stiffness characteristic curves and extracted physical morphology features; Using the spring simulation data under different combinations of geometric design parameters, the corresponding stiffness characteristic curves, and the extracted physical morphology features to train the prediction model, so that the prediction model learns the complex nonlinear relationship between the input parameters and the first physical morphology feature.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.

Images