A method for selecting deformation constraints for a linear motor module, electronic equipment, and computer-readable storage medium.

By constructing an equivalent finite element model of the guide rail slider and a proxy model library, the motion condition is decomposed into multi-stage multi-dimensional loads, which solves the problem of insufficient accuracy caused by neglecting the deformation of the slide table in the selection of linear motor modules, and realizes an efficient and accurate selection method.

CN123116722APending Publication Date: 2026-11-17SHENZHEN DYNAMIKWELL TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611553759.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-09
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing linear motor module selection methods ignore slide deformation, resulting in insufficient accuracy. Traditional finite element simulation is time-consuming and has a high barrier to entry, making it impossible to embed into conventional selection processes and failing to meet the accuracy requirements of precision applications.

Method used

An equivalent finite element model of the guide rail slider with nonlinear contact characteristics is constructed. A surrogate model library is established by sampling. The motion condition is decomposed into multi-stage multi-dimensional loads. The deformation prediction value is obtained by querying the surrogate model library. The linear motor module is selected based on the deformation evaluation index.

Benefits of technology

This method enables the rapid acquisition of slide deformation during the selection phase, transforming it into calculable constraints. This improves the accuracy and efficiency of the selection process, balances computational efficiency and accuracy, and solves the problem of insufficient accuracy caused by slide deformation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116722A_ABST
    Figure CN123116722A_ABST
Patent Text Reader

Abstract

This invention discloses a method for selecting linear motor modules based on deformation constraints, an electronic device, and a computer-readable storage medium. The method constructs an equivalent finite element model of a guide rail slider incorporating nonlinear contact characteristics. It uses multidimensional generalized loads as input and multidimensional deformation of the slide corners as output, establishing a proxy model library through sampling. During selection, the motion condition is decomposed into multi-stage multidimensional loads, and the deformation prediction value is obtained by querying the proxy model library. The linear motor module is then screened based on deformation evaluation indicators. This invention pre-simulates complex finite element deformation and parameterizes it into a proxy model library. During the selection stage, the slide deformation is quickly obtained through data querying, transforming deformation accuracy into calculable conventional constraints. This achieves multidimensional and accurate selection, balancing computational efficiency and selection accuracy, and solving the problem of insufficient accuracy caused by neglecting the elastic deformation of the slide in existing selection methods.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of linear motor technology, and in particular to a method for selecting deformation constraints for linear motor modules, electronic devices, and computer-readable storage media. Background Technology

[0002] As a core component of automated equipment, the selection of linear motor modules directly determines the final performance of the equipment. Currently, the online selection software or engineering manuals provided by mainstream suppliers typically follow this calculation logic: First, the user-provided motion task parameters are confirmed, mainly including motion curves (trapezoidal or S-shaped velocity curves, defining maximum speed, acceleration, and total stroke), load characteristics (load mass, load eccentricity), installation and use (installation direction, friction coefficient, reverse resistance), etc. The core calculation includes motor calculation and guide rail calculation: the motor is calculated based on Newton's second law to determine the required thrust, continuous thrust, and corresponding current and voltage at each stage, ensuring the selected motor has sufficient power to drive the load; the guide rail calculation is based on the load mass and possible eccentricity, calculating the equivalent load acting on the guide rail, and then combining this with the module's rated dynamic load, according to... The lifespan formula estimates the travel life of the guide rail. Finally, based on the evaluation of the above calculation results, a list of module models that meet the thrust and lifespan requirements is provided.

[0003] This selection method ignores the positional accuracy errors caused by slide deformation, relying solely on thrust and lifespan indicators to screen modules. This fails to guarantee that the selected linear motor module will meet the accuracy requirements of real-world applications. In other words, current linear motor module selection focuses on two lines: "sufficient thrust" and "sufficient lifespan," neglecting the structural elastic deformation of the slide plate under actual loads. In precision applications, even if the module's lifespan and thrust meet requirements, the instantaneous pitch, yaw, or roll deformation of the slide plate caused by inertial forces and their torques during high-speed start-stop can reach tens of micrometers. This is fatal for tasks requiring micrometer-level or higher positioning. For example, in wafer inspection, changes in probe height in the Z-axis can lead to measurement failure or wafer damage; in laser cutting, defocusing of the laser spot can degrade cutting quality. Therefore, neglecting the deformation changes generated during slide operation can result in the selected module failing to meet design accuracy requirements, impacting the overall performance of the equipment.

[0004] Conventional deformation assessment methods involve conducting finite element deformation simulations for specific working conditions and module models. However, this traditional approach has significant limitations: firstly, it requires building a model, meshing, and performing calculations separately for each candidate model, a cumbersome and time-consuming process that makes it difficult to complete batch assessments of multiple models in a short time; secondly, finite element analysis demands a high level of expertise from operators, and ordinary selection personnel often lack the corresponding simulation skills. Currently, the industry lacks a simplified deformation assessment scheme that can be directly embedded into the selection process and adapted to most application scenarios, preventing deformation factors from becoming a regular consideration in the selection phase. Summary of the Invention

[0005] This invention provides a linear motor module deformation constraint selection method, electronic device, and computer-readable storage medium, aiming to improve the problems of insufficient accuracy caused by neglecting the elastic deformation of the slide table in existing selection methods, and the long time consumption and high threshold of traditional finite element simulation, which cannot be embedded in the conventional selection process.

[0006] To achieve the above-mentioned objectives, the following technical solution is provided: Provides a method for selecting linear motor module deformation constraints, including: An equivalent finite element model of the guide rail slider with nonlinear contact characteristics is constructed. Multidimensional generalized load is used as input and multidimensional deformation of the slide corner is used as output. A proxy model library is established through sampling. During the selection process, the motion conditions are decomposed into multi-stage, multi-dimensional loads. The deformation prediction values ​​are obtained by querying the proxy model library, and the linear motor modules are screened based on the deformation evaluation index.

[0007] Furthermore, the construction of the equivalent finite element model of the guide rail slider incorporating nonlinear contact characteristics includes: The contact stiffness coefficient is calculated based on Hertzian point contact theory, and nonlinear spring elements are generated along the contact normal to obtain the equivalent spring finite element model of the guide rail slider. The preload state is simulated by the overall offset of the force and displacement curves, and the preload parameters are adjusted by replacing the spring curve; the calculation of the contact stiffness coefficient based on Hertz point contact theory includes: identifying the contact points between the ball and the guide rail groove and the slider groove, and defining each contact as a ball contact pair; The contact stiffness coefficient is calculated based on the material's elastic modulus, Poisson's ratio, ball diameter, groove curvature, fit, and contact angle. The contact force and elastic approximation satisfy... ,in For contact force, For contact stiffness coefficient, This is the elastic approximation quantity.

[0008] Furthermore, the step of establishing a proxy model library through sampling includes: The multidimensional load sampling range is determined based on the rated static load and static safety factor of the guide rail pair. Latin hypercube sampling is performed within the multidimensional load sampling range to obtain load sample sets and deformation sample sets, and a Kriging surrogate model is established to obtain a Kriging surrogate model library. The load sampling intervals of each dimension of the multidimensional load sampling range are determined based on the total number of sliders, the basic rated static load of the slider radial direction, the basic rated static load of the slider in reverse radial direction, the basic rated static load of the slider in the side direction, the front and rear span of the slider, and the span of the guide rail.

[0009] Furthermore, the decomposition of the motion condition into multi-stage, multi-dimensional loads includes: Based on d'Alembert's principle, inertial force is superimposed on the slide as a virtual external force, transforming the multi-stage dynamics problem into a static equilibrium problem; The static force of the dispersed external load is equivalent to the preset reference center point of the slide table to obtain a multi-stage five-dimensional load vector acting on the reference center point. The multi-stage process includes an acceleration stage, a constant speed stage, a deceleration stage, and a dwell stage. The corresponding thrust is calculated for each of the acceleration phase, the constant speed phase, the deceleration phase, and the dwell phase. The five-dimensional generalized load vector for each phase is calculated by combining the load gravity, motor magnetic attraction force, friction force, and process force.

[0010] Furthermore, the screening of linear motor modules based on deformation evaluation indicators includes: Obtain the predicted deformation and prediction variance output by the proxy model library; The absolute deformation margin and deformation change are calculated based on the predicted deformation. When all directions in the evaluation direction set satisfy the condition that the absolute deformation margin is less than or equal to the preset deformation threshold, and the deformation change is less than or equal to the preset change threshold, the deformation evaluation is deemed to have passed. If the maximum prediction variance exceeds the preset confidence threshold, a confidence alarm will be triggered. The absolute deformation margin is the ratio of the maximum deformation amplitude of each corner point in each motion stage to a preset direction threshold. The deformation change is the difference between the maximum and minimum values ​​of the deformation extrema (with positive and negative signs) at each corner point during each motion stage.

[0011] Furthermore, after establishing the proxy model library, the following is also included: Leave-one-out cross-validation is used to evaluate the accuracy of each output channel of the proxy models in the proxy model library, and the coefficient of determination is calculated. ; When the determination coefficient is greater than or equal to a preset threshold, the output channel is deemed to have successfully built its library.

[0012] Furthermore, the multidimensional generalized load includes the resultant force along the equivalent Y-axis and Z-axis, and the resultant moment around the equivalent X-axis, Y-axis and Z-axis. The multidimensional deformation of the slide corners refers to the deformation response of the four corners of the slide in three dimensions.

[0013] Furthermore, when all candidate linear motor modules fail the deformation evaluation, a model prompt indicating that no deformation constraint is met is output, and a modification suggestion is generated. The modification suggestion includes at least one of improving module specifications, adjusting the installation method, adjusting the motion cycle, or reducing the load.

[0014] An electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described linear motor module deformation constraint selection method.

[0015] A computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the above-described method for selecting linear motor module deformation constraints. Beneficial effects:

[0016] This invention pre-parameterizes complex finite element deformation simulation into a proxy model library, and quickly obtains the deformation of the slide table through data query during the selection stage. It transforms deformation accuracy into calculable conventional constraints, achieving multi-dimensional accurate selection while balancing computational efficiency and selection accuracy. Attached Figure Description

[0017] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the linear motor module deformation constraint selection method based on response surface model of the present invention; Figure 2 This is a three-dimensional schematic diagram of the linear motor module structure (constant module) of the present invention; Figure 3 This is a cross-sectional schematic diagram of the linear motor module structure (conventional module) of the present invention; Figure 4 This is a three-dimensional schematic diagram of the linear motor module structure (embedded module) of the present invention; Figure 5This is a cross-sectional schematic diagram of the linear motor module structure (embedded module) of the present invention; Figure 6 This is a cross-sectional schematic diagram of the nonlinear spring arrangement (Gothic groove structure of embedded module) of the present invention and a schematic diagram of its equivalent spring arrangement structure.

[0019] Figure 7 This is a schematic diagram of the nonlinear spring arrangement along the guide rail direction and a schematic diagram of its equivalent spring arrangement structure.

[0020] Figure label: Figures 1-2 In the middle: 1. Slider; 2. Standard assembly slide; 3. Guide rail; Figures 3-4 In the middle: 4. Embedded modular slide table; 5. Slide table rail; 6. Base rail; Figures 5-6 7. Ball bearing; 8. Equivalent spring. Detailed Implementation

[0021] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings of the embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the invention.

[0022] As a core component of automated equipment, the selection of linear motor modules directly determines the final performance of the equipment. Currently, mainstream selection calculations include motor and guide rail calculations, ultimately providing a list of module models that meet thrust and lifespan requirements. However, this selection method ignores the positional accuracy errors caused by slide deformation, selecting modules solely based on thrust and lifespan indicators. In precision applications, the instantaneous deformation of the slide caused by inertial forces and their torques during high-speed start-stop processes can reach tens of micrometers, leading to measurement failures or reduced cutting quality. Conventional deformation assessment methods involve conducting finite element deformation simulations for specific working conditions and models, but this method is cumbersome, time-consuming, and requires a high level of operator expertise, making it difficult to integrate into standard selection processes.

[0023] The present invention will be further described below with reference to specific embodiments and accompanying drawings, including the linear motor module deformation constraint selection method, electronic equipment, and computer-readable storage medium.

[0024] Please see Figures 2-3 It is a linear motor module structure with a constant scale. Its constant scale slider 2 slides on the guide rail 3 through the slider 1. At this time, deformation will occur between the slider and the guide rail.

[0025] Please see Figures 4-5It is an embedded linear motor module structure, in which the embedded module slide 4 and base are respectively embedded in the slide rail 5 and the base rail 6, and deformation will occur between the embedded module slide 4 and slide rail 5 and the base rail 6.

[0026] Please see Figures 6-7 It employs Gothic-style grooves, meaning each groove on both sides comprises upper and lower sections, forming a pointed arch structure (with an enlarged radius). The contact between the ball bearing 7 and the pointed arch structure is equivalent to the effect of an equivalent spring 8, thus causing deformation. To overcome this deformation problem, please refer to... Figure 1 This invention discloses a method for selecting linear motor modules based on deformation constraints, comprising: constructing an equivalent finite element model of a guide rail slider that includes nonlinear contact characteristics; taking multidimensional generalized loads as inputs and multidimensional deformation of the slide corners as outputs; establishing a proxy model library through sampling; during selection, decomposing the motion condition into multi-stage multidimensional loads; querying the proxy model library to obtain deformation prediction values; and screening linear motor modules based on deformation evaluation indicators.

[0027] In this embodiment of the invention, the "equivalent finite element model of the guide rail slider" can be the "equivalent spring finite element model of the guide rail slider," etc., and this embodiment of the invention does not limit this. The "multidimensional generalized load" can be the "five-dimensional generalized load," etc., and this embodiment of the invention does not limit this. The "surrogate model library" can be the "Kriging surrogate model library," etc., and this embodiment of the invention does not limit this. The "deformation evaluation index" can be the "absolute deformation margin and deformation change," etc., and this embodiment of the invention does not limit this.

[0028] By constructing an equivalent finite element model of the guide rail slider incorporating nonlinear contact characteristics and establishing a surrogate model library, complex finite element deformation simulations are pre-simulated and parameterized. During the selection phase, by decomposing the motion condition into multi-stage, multi-dimensional loads and querying the surrogate model library, the deformation of the slide table can be quickly obtained, transforming deformation accuracy into calculable conventional constraints. This approach achieves multi-dimensional, accurate selection, balancing computational efficiency and selection accuracy. It effectively solves the problems of existing selection methods neglecting the elastic deformation of the slide table, resulting in insufficient accuracy, and the drawbacks of traditional finite element simulations being time-consuming, having high barriers to entry, and being unable to be integrated into conventional selection processes.

[0029] In some embodiments, the contact stiffness coefficient is calculated based on Hertzian point contact theory, and a nonlinear spring element is generated along the contact normal to obtain an equivalent spring finite element model of the guide rail and slider; the preload state is simulated by the overall offset of the force and displacement curves, and the preload parameters are adjusted by replacing the spring curves; the contact stiffness coefficient is calculated based on Hertzian point contact theory, including: identifying the contact points between the ball and the guide rail groove and the slider groove as ball contact pairs; calculating the contact stiffness coefficient according to the material elastic modulus, Poisson's ratio, ball diameter, groove curvature and fit, and contact angle, and the contact force and elastic convergence satisfying ,in For contact force, For contact stiffness coefficient, This is the elastic approximation quantity.

[0030] Specifically, the accuracy retention, vibration resistance, and dynamic characteristics of rolling linear guide pairs are all closely related to the stiffness of their joints. The finite element modeling steps are as follows: First, determine the contact pairs, given the guide model, determine the ball arrangement: the number of effective load-bearing balls per row Z, the rolling element pitch, the number of grooves i, and the contact angle. 1. Groove curvature radius coefficient. For each ball, identify all contact points with the guide rail groove and slider groove (4-point contact for Gothic grooves, 2-point contact for arc-shaped grooves), and define one "ball-groove contact point" as a ball contact pair. Next, calculate the Hertzian contact stiffness coefficient. For each contact pair, calculate its contact stiffness coefficient according to Hertzian point contact theory. Then, a nonlinear spring element is inserted along the contact normal (load transfer direction) of each contact pair; the nodes at both ends of the spring are respectively bound to the corresponding nodes on the slider side and the guide rail side. The ball body acts as a zero-mass contact medium, and its stiffness is reflected by the series connection of Hertz springs at both ends. Finally, the preload and initial contact state are achieved, with the preload level corresponding to an initial normal contact force. The initial compression is obtained by inversely solving the Hertzian relation: Since the preload of a nonlinear spring cannot be set independently, preload is achieved by offsetting the entire force-displacement curve. The force-displacement relationship on the compression side is rewritten as follows: This ensures that all contact pairs are in a pre-compression state under zero external load, and the relative displacement on the unloading side is greater than [the specified value]. When the contact pair disengages, adjustments to parameters such as preload level are needed; simply replacing the spring curve and regenerating the model will achieve "quick parameter adjustment."

[0031] By calculating the contact stiffness coefficient based on Hertzian point contact theory and generating nonlinear spring elements, the actual physical characteristics such as preload-related stiffness, load distribution, and unloading separation are preserved, avoiding the problem of excessive computational load in solid contact models. At the same time, the simulation of the preload state is achieved by using the overall offset of force and displacement curves, so that preload switching can be achieved by simply replacing the spring curve to quickly adjust parameters. This supports batch library creation for multiple models and multiple preload levels, improving modeling efficiency and model accuracy.

[0032] In some embodiments, the multidimensional load sampling range is determined based on the guide rail pair's rated static load and static safety factor; Latin hypercube sampling is performed within the multidimensional load sampling range to obtain load sample sets and deformation sample sets, and a Kriging surrogate model is established to obtain a Kriging surrogate model library; the load sampling intervals of each dimension of the multidimensional load sampling range are determined based on the total number of sliders, the slider's radial basic rated static load, the anti-radial basic rated static load, the lateral basic rated static load, the slider's front and rear spans, and the guide rail spans.

[0033] Specifically, Latin hypercube sampling requires defining intervals for each dimension of the five-dimensional space. The sampling range should be bounded by the actual load-bearing capacity of the guide rail, ensuring that the modeling envelope equals the physically accessible and effective working domain of the model. The range of five-dimensional load values ​​should be based on the rated static load of the guide rail pair. Based on the reference values ​​provided by various guide rail manufacturers, the static safety factor under typical operating conditions is... The value should be between 1.0 and 3.0. Considering the deformation threshold of the platform during actual use, a value of 1.0 should be used. Therefore, the calculation boundary for the loads in all directions of the guide rail is taken as... Let the total number of sliders be n. These are the basic rated static loads for the slider's radial, reverse radial, and lateral directions, respectively, and the slider's front-to-back span is... The guide rail span is Based on the principle of distributed force systems in theoretical mechanics, the sampling intervals for loads in each dimension are shown in the table below:

[0034] This formula is based on slide types with more than two guide rails and two sliders, i.e., n≥4.

[0035] To construct the overall nonlinear mapping relationship, an experimental design method using Latin hypercube sampling (LHS) was employed for the five-dimensional effective independent load space, ensuring that N samples uniformly cover the boundaries of each dimension of the five-dimensional space. Finite element analysis was applied to each of the N samples to obtain the corresponding deformations, forming the load sample set and the deformation sample set. Due to the contact nonlinearity of the linear rolling guide pair, a Kriging surrogate model was adopted.

[0036] By determining the multidimensional load sampling range based on the guide rail pair's rated static load and static safety factor, the modeling envelope is equal to the physically reachable and effective working domain of the model. This avoids the problem of physical inaccessibility or insufficient coverage caused by arbitrarily given sampling intervals, ensuring that the surrogate model interpolates accurately within the intervals of greatest concern in the design, and improving the prediction accuracy and reliability of the surrogate model.

[0037] In some embodiments, based on d'Alembert's principle, inertial force is superimposed on the slide as a virtual external force, transforming the multi-stage dynamics problem into a static equilibrium problem; the dispersed external load is statically equivalent to a preset reference center point of the slide, resulting in a multi-stage five-dimensional load vector acting on the reference center point; the multi-stage includes an acceleration stage, a constant speed stage, a deceleration stage, and a dwell stage; the corresponding thrust is calculated in each of the acceleration stage, constant speed stage, deceleration stage, and dwell stage, and the five-dimensional generalized load vector of each stage is calculated in combination with the load gravity, motor magnetic attraction force, friction force, and process force.

[0038] Specifically, taking the guide rail stiffness center O as the coordinate origin, the multiple dispersed external loads acting on the slide table and load are statically equivalent to the preset reference center point O of the slide table, resulting in a five-dimensional generalized load vector acting at point O: In the direction of the slide's forward movement (X direction), the d'Alembert method is introduced to address the inertial force. The virtual external force superimposed on the slide table transforms the original dynamics problem into a static equilibrium problem. The external loads in three directions of the linear motor module's slide table during movement are analyzed in the table below:

[0039] Taking a side-mounted application as an example, the load is m, and the center of gravity is... During the dwell phase, there is a force acting in the z-direction. The point of application is A force in the x-direction The point of application is Friction Let the thrust center of each module be denoted as . Magnetic attraction force is Thrust is The equivalent load calculations are performed in stages, as shown in the table below:

[0040] Thrust at each stage (k=1…4, representing acceleration, constant speed, deceleration, and dwell stages), related to the external load in this example: The equivalent load table is as follows:

[0041] By superimposing inertial force as a virtual external force onto the slide based on d'Alembert's principle and statically equivalentizing the dispersed external load to the reference center point, a general method for equivalent handling of external loads is provided. This transforms multi-stage dynamic problems into static equilibrium problems, covering different installation methods and load variations. This enables the proxy model to serve the loads decomposed from any motion task, improving the versatility and computational efficiency of the selection method.

[0042] In some embodiments, the predicted deformation and prediction variance output by the proxy model library are obtained; the absolute deformation margin and deformation change are calculated based on the predicted deformation; when all directions in the evaluation direction set satisfy that the absolute deformation margin is less than or equal to a preset deformation threshold and the deformation change is less than or equal to a preset change threshold, the deformation evaluation is deemed to pass; if the maximum prediction variance exceeds a preset confidence threshold, a confidence alarm is triggered; the absolute deformation margin is the ratio of the maximum deformation amplitude of each corner point in each motion stage to the preset direction threshold; the deformation change is the difference between the maximum and minimum values ​​of the signed deformation extreme values ​​of each corner point in each motion stage.

[0043] Specifically, the proxy model outputs 4×3×4 predicted values ​​for each model. and prediction variance Absolute deformation margin The calculation method is as follows: For direction j, first select the one with the largest deformation amplitude among the four corner points and four stages in that direction. Then, with the direction threshold compared to, . This indicates that the direction is evaluated using absolute deformation margin. To prevent the surrogate model's smoothing effect on deformation peaks from leading to an underestimation of extreme values, Conservative envelope value can be adopted. Deformation change The calculation method is as follows: for direction j, take the signed extreme value. and The difference between the two is the amount of deformation in that direction. . This indicates that the deformation change in this direction is assessed. A set D of assessment directions is defined; the model passes the deformation assessment only if all directions in D meet the conditions. If the prediction variance of the most unfavorable channel in this direction exceeds a preset confidence threshold, it indicates insufficient accuracy of the surrogate model and is downgraded to "pending review".

[0044] By acquiring the predicted deformation and prediction variance, and evaluating them based on the dual indicators of absolute deformation margin and deformation change, the actual accuracy risk can be reflected more accurately, preventing positioning failures caused by dynamic fluctuations. At the same time, by triggering confidence alarms through prediction variance, the risk of insufficient model accuracy can be effectively indicated, avoiding the underestimation of extreme values ​​caused by the smoothing effect of proxy models, further improving the reliability and security of the selection results.

[0045] In some embodiments, leave-one-out cross-validation is used to evaluate the accuracy of each output channel of the proxy models in the proxy model library and to calculate the coefficient of determination. When the determination coefficient is greater than or equal to the preset threshold, the output channel is deemed to have passed the database construction test.

[0046] Specifically, the model, trained on the i-th sample and the remaining N-1 samples, is used to predict the cross-validation prediction value. ; with the coefficient of determination For indicators: Take the threshold. The channels that meet the criteria are deemed qualified for inclusion in the database and retained. Only when all 12 channels meet the criteria will the surrogate model of that model be included in the displacement surrogate model library. By employing leave-one-out cross-validation to evaluate the accuracy of each output channel, the predictive ability of the surrogate model can be objectively and accurately measured, ensuring that the models included in the surrogate model library have sufficient accuracy.

[0047] In some embodiments, the multidimensional generalized load includes the resultant force along the equivalent rear Y-axis and Z-axis, and the resultant moment around the equivalent X-axis, Y-axis and Z-axis; the multidimensional deformation of the slide corners is the deformation response of the four corners of the slide in three-dimensional directions.

[0048] Specifically, the independently given load components are five-dimensional: The output deformation vector is defined as the three-dimensional deformation response of the four corner points of the slide, resulting in a total of 12 dimensions: By clarifying the specific dimensions and physical meaning of multidimensional generalized loads and multidimensional deformation at the corners of the slide, we can comprehensively and accurately characterize the deformation features of the slide under complex stress conditions, providing a complete and accurate data foundation for subsequent deformation assessment.

[0049] In some embodiments, when all candidate linear motor modules fail the deformation evaluation, a model prompt indicating that the deformation constraint is not met is output, and a modification suggestion is generated. The modification suggestion includes at least one of improving module specifications, adjusting the installation method, adjusting the motion cycle, or reducing the load.

[0050] Specifically, if the Z-axis of all candidate models under a certain application All exceeded (For example, in the case of a long cantilever structure), each model is first judged as "non-compliant," and the direction and indicators of exceeding limits are marked. Since all models fail, the selection ranking outputs "No model that meets deformation constraints," and provides modification suggestions—upgrading specifications (increasing preload, adding auxiliary guide rails), adjusting the installation method, adjusting the motion rhythm (reducing acceleration a to reduce inertial torque), or reducing the load (reducing m and eccentricity). By outputting prompts and generating modification suggestions when all candidate modules fail the evaluation, a closed-loop decision-making process for selection ranking is achieved. This guides users to optimize and adjust from multiple dimensions such as structure, installation, and motion control, improving the practicality and intelligence level of the selection system.

[0051] The invention will be further explained below with reference to specific implementation examples.

[0052] Specific Implementation Case 1 (Complete Selection of Standard Modules for Horizontal Installation): Proxy model library construction: For the candidate model set, establish finite element models (light / medium / heavy preload) including nonlinear equivalent springs. The five-dimensional load range was determined by parameters (see table below), and N=120 groups of Latin hypercube sampling were used to build a 12-channel Kriging model.

[0053]

[0054] Input application conditions: horizontal installation, m=12kg; center of gravity (35,0,90)mm; a=20m / s². =2m / s, s=0.6m; =10N; =950N.

[0055] Work condition breakdown: The five-dimensional loads at each stage are as follows (torque unit N·m):

[0056] Proxy model query and evaluation: Outputs 48 predicted displacements. Taking the Z-axis as an example: , , pass; Passed. The same applies to directions X and Y.

[0057] Output selection sorting: Based on thrust, life calculation and economic evaluation, a preliminary selection list is obtained, and then selection is performed through deformation constraints. The multiple candidate models are sorted and output in the original list order.

[0058] Specific Implementation Case 2 (Side-mounted Installation Variant): The installation method in Specific Implementation Case 1 was changed to side-mounted, and the z-axis process force and x-axis process force were superimposed during the dwell phase. The five-dimensional load was decomposed using the same method. The load is no longer always zero. This proves that the external load equivalent method is applicable to different installation methods; the evaluation and ranking process is completely consistent with specific implementation case 1. This case verifies that the present invention does not depend on a specific installation direction and can cover horizontal / side-mounted / inverted working conditions simply by using the equivalent load vector.

[0059] This invention also provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the linear motor module deformation constraint selection method described in any of the foregoing embodiments. This electronic device enables the rapid and accurate execution of the deformation constraint selection method, improving the selection efficiency and accuracy of automated equipment.

[0060] This invention also provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the linear motor module deformation constraint selection method described in any of the foregoing embodiments. This computer-readable storage medium facilitates the storage and distribution of the computer program, enabling the deployment and application of the selection method on different devices.

[0061] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for selecting linear motor module deformation constraints, characterized in that, include: An equivalent finite element model of the guide rail slider with nonlinear contact characteristics is constructed. Multidimensional generalized load is used as input and multidimensional deformation of the slide corner is used as output. A proxy model library is established through sampling. During the selection process, the motion conditions are decomposed into multi-stage, multi-dimensional loads. The deformation prediction values ​​are obtained by querying the proxy model library, and the linear motor modules are screened based on the deformation evaluation index.

2. The linear motor module deformation constraint selection method according to claim 1, characterized in that, The construction of the equivalent finite element model of the guide rail slider, which includes nonlinear contact characteristics, includes: The contact stiffness coefficient is calculated based on Hertzian point contact theory, and nonlinear spring elements are generated along the contact normal to obtain the equivalent spring finite element model of the guide rail slider. The preload state is simulated by the overall offset of the force and displacement curves, and the preload parameters are adjusted by replacing the spring curve; the calculation of the contact stiffness coefficient based on Hertz point contact theory includes: identifying the contact points between the ball and the guide rail groove and the slider groove, and defining each contact as a ball contact pair; The contact stiffness coefficient is calculated based on the material's elastic modulus, Poisson's ratio, ball diameter, groove curvature, fit, and contact angle. The contact force and elastic approximation satisfy... ,in For contact force, For contact stiffness coefficient, This is the elastic approximation quantity.

3. The linear motor module deformation constraint selection method according to claim 1, characterized in that, The process of establishing a proxy model library through sampling includes: The multidimensional load sampling range is determined based on the rated static load and static safety factor of the guide rail pair. Latin hypercube sampling is performed within the multidimensional load sampling range to obtain load sample sets and deformation sample sets, and a Kriging surrogate model is established to obtain a Kriging surrogate model library. The load sampling intervals of each dimension of the multidimensional load sampling range are determined based on the total number of sliders, the basic rated static load of the slider radial direction, the basic rated static load of the slider in reverse radial direction, the basic rated static load of the slider in the side direction, the front and rear span of the slider, and the span of the guide rail.

4. The linear motor module deformation constraint selection method according to claim 1, characterized in that, The decomposition of motion conditions into multi-stage, multi-dimensional loads includes: Based on d'Alembert's principle, inertial force is superimposed on the slide as a virtual external force, transforming the multi-stage dynamics problem into a static equilibrium problem; The static force of the dispersed external load is equivalent to the preset reference center point of the slide table to obtain a multi-stage five-dimensional load vector acting on the reference center point. The multi-stage process includes an acceleration stage, a constant speed stage, a deceleration stage, and a dwell stage. The corresponding thrust is calculated for each of the acceleration phase, the constant speed phase, the deceleration phase, and the dwell phase. The five-dimensional generalized load vector for each phase is calculated by combining the load gravity, motor magnetic attraction force, friction force, and process force.

5. The linear motor module deformation constraint selection method according to claim 1, characterized in that, The screening of linear motor modules based on deformation evaluation indicators includes: Obtain the predicted deformation and prediction variance output by the proxy model library; The absolute deformation margin and deformation change are calculated based on the predicted deformation. When all directions in the evaluation direction set satisfy the condition that the absolute deformation margin is less than or equal to the preset deformation threshold, and the deformation change is less than or equal to the preset change threshold, the deformation evaluation is deemed to have passed. If the maximum prediction variance exceeds the preset confidence threshold, a confidence alarm will be triggered. The absolute deformation margin is the ratio of the maximum deformation amplitude of each corner point in each motion stage to a preset direction threshold. The deformation change is the difference between the maximum and minimum values ​​of the deformation extrema (with positive and negative signs) at each corner point during each motion stage.

6. The linear motor module deformation constraint selection method according to claim 1, characterized in that, After establishing the proxy model library, the following is also included: Leave-one-out cross-validation is used to evaluate the accuracy of each output channel of the proxy models in the proxy model library, and the coefficient of determination is calculated. ; When the determination coefficient is greater than or equal to a preset threshold, the output channel is deemed to have successfully built its library.

7. The linear motor module deformation constraint selection method according to claim 1, characterized in that, The multidimensional generalized load includes the resultant force along the equivalent Y-axis and Z-axis, and the resultant moment around the equivalent X-axis, Y-axis and Z-axis. The multidimensional deformation of the slide corners refers to the deformation response of the four corners of the slide in three dimensions.

8. The linear motor module deformation constraint selection method according to claim 1, characterized in that, When all candidate linear motor modules fail the deformation evaluation, a model prompt indicating that no deformation constraint is met is output, and a modification suggestion is generated. The modification suggestion includes at least one of improving module specifications, adjusting the installation method, adjusting the motion cycle, or reducing the load.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the linear motor module deformation constraint selection method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the linear motor module deformation constraint selection method as described in any one of claims 1 to 8.