Hinge rigidity design method for flexible guide mechanism of energy balance based on machine learning

Abstract

The invention discloses an energy balance flexible guide mechanism hinge rigidity design method based on machine learning, and relates to the technical field of energy balance design. Constructing a mechanism parameter library by taking the hinge thickness as a structural parameter; converting the three-dimensional modeling software design scheme into a three-dimensional model library; performing three-dimensional grid division and finite element analysis to establish a grid library; constructing a three-dimensional simulation result database through finite element simulation results; establishing a mapping relation between the structure parameters and the performance characteristics by utilizing a machine learning method, and taking the mapping relation as a parameter library of machine learning; optimizing the design scheme through a machine learning method; and flexible guide mechanism scheme libraries under different index systems are constructed, and the flexible guide mechanism hinge meeting the requirements is obtained. An optimal flexible guide mechanism hinge design scheme is predicted through a machine learning method, mechanical properties of a hinge three-dimensional model are simulated in combination with finite element analysis, experiment and calculation requirements are reduced, and design efficiency and economic benefits are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of energy balance design, and in particular to a method for designing hinge stiffness of a flexible guide mechanism of an energy balance based on machine learning. Background Art

[0002] As a mass standard measuring device, the flexible guide mechanism of the energy balance is crucial to the E-8 level measurement accuracy and the improvement of the motion performance of the mechanical system. However, due to the long processing cycle, time-consuming assembly process and heavy load, the experimental conditions are difficult to strictly control, resulting in great challenges in the design and application of the flexible guide mechanism hinge.

[0003] Since the flexible guide mechanism hinge operates under complex stress conditions, different combinations of design parameters (such as the thickness of the hinge) will significantly affect its performance. In addition, the flexible guide mechanism has a high risk during assembly, and a slight deviation may lead to performance failure or require rework. Therefore, it is very important to study the design scheme of the flexible guide mechanism hinge and its corresponding mechanical properties under specific working conditions. In addition, the processing cycle of the flexible guide mechanism hinge is long, and the risk and time-consuming assembly process will further increase the cost of experimental optimization, and the simulation calculation is large. Summary of the invention

[0004] In order to improve the performance stability of the hinge of the flexible guide mechanism of the energy balance under complex mechanical conditions, reduce experimental costs and improve computing efficiency, the present invention starts from the relationship between different design parameters and the performance characteristics of the hinge of the flexible guide mechanism, and provides an energy balance flexible guide mechanism hinge stiffness design method based on machine learning. It predicts the optimal flexible guide mechanism hinge design scheme through a machine learning method, simulates the mechanical properties of the hinge three-dimensional model in combination with finite element analysis, and verifies its maximum load-bearing capacity, maximum stroke and motion straightness through simulation, thereby reducing expensive physical experiments and heavy computing requirements, and significantly improving design efficiency and economic benefits.

[0005] To achieve the above object, the present invention adopts the following technical solution: a method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning, comprising the following steps:

[0006] Step 1: Establish the mechanism parameter library of the flexible guide mechanism

[0007] The thickness of the hinge is used as a structural parameter, and the mechanism parameter library of the flexible guide mechanism is constructed by selecting different hinge thicknesses t1-t13;

[0008] Step 2: Create a 3D model library for the flexible guide mechanism

[0009] According to the thickness t1-t13 data of the hinge in the mechanism parameter library of the flexible guide mechanism, a three-dimensional model of the flexible guide mechanism is established by three-dimensional modeling software to obtain different design schemes, and the mechanism parameter library is converted into a three-dimensional model library of the flexible guide mechanism that can be used for numerical simulation;

[0010] Step 3: Create a mesh library corresponding to the 3D model of the flexible guide mechanism

[0011] The three-dimensional models of the flexible guide mechanism in different design schemes are simulated and analyzed under different working environments using three-dimensional meshing software and finite element analysis software. The mechanical properties of the flexible guide mechanism are analyzed through meshing of the three-dimensional model and finite element meshing. The mesh model of the flexible guide mechanism is generated according to different structural parameters, and the mesh library corresponding to the three-dimensional model is constructed using the generated mesh data.

[0012] Step 4: Establish a 3D simulation result database of the flexible guide mechanism

[0013] Through the finite element simulation results, a three-dimensional simulation result database covering the performance characteristics of maximum load capacity F, maximum stroke ΔZ and motion straightness δL is constructed;

[0014] Step 5: Use machine learning methods to establish a library of mechanism parameters and mechanism characteristic parameters

[0015] By using machine learning methods and based on the parameters generated by finite element simulation results, a mapping relationship between the structural parameters of the flexible guide mechanism and its performance characteristics under specific load conditions is established as a mechanism parameter and mechanism characteristic parameter library for machine learning;

[0016] Step 6: Optimize the design scheme through machine learning methods

[0017] 6.1. Machine Learning Model Optimization

[0018] Use machine learning algorithms to comprehensively evaluate and select multiple models, and gradually select the best performing machine learning models based on structural parameter input and performance characteristic output;

[0019] 6.2. Preventing Overfitting and Data Preprocessing

[0020] Divide the mechanism parameters and mechanism characteristic parameter library into training set and validation set in proportion, perform data preprocessing to ensure the quality and consistency of input data, perform model selection and hyperparameter tuning, and use cross-validation to evaluate the performance of the model on different data subsets to avoid overfitting;

[0021] 6.3 Model Evaluation Metrics

[0022] Use evaluation indicators to comprehensively evaluate the prediction effect of the model;

[0023] 6.4. Output Fitting and Model Optimization

[0024] Comprehensively compare the fitting effects of each output under different inputs, and use evaluation indicators to select the best machine learning algorithm corresponding to each structural parameter;

[0025] Step 7: Construct a library of flexible guide mechanism solutions under different indicator systems

[0026] 7.1. Machine Learning Methods to Predict the Optimal Design Solution

[0027] According to the thickness t1-t13 of the input hinge, the design scheme of the flexible guide mechanism hinge with the best maximum load capacity F, maximum stroke ΔZ and motion linearity δL is predicted;

[0028] 7.2. Build a 3D model

[0029] Use 3D modeling software to build a 3D model of the flexible guide mechanism, ensuring that the thickness t1-t13 of all hinges is included in the model;

[0030] 7.3 Simulation and performance analysis

[0031] The 3D model of the flexible guide mechanism was simulated using finite element analysis software to analyze the mechanical behavior of the model under different loads, and the maximum load capacity F, maximum stroke ΔZ, and motion straightness δL were calculated through the simulation results;

[0032] 7.4 Iterative Optimization of Design Parameters

[0033] Based on comprehensive consideration of the performance characteristics of the maximum load capacity F, maximum stroke ΔZ and motion straightness δL, the structural parameters of the flexible guide mechanism hinge are optimized, and the mechanism parameters and mechanism characteristic parameter library are updated until a flexible guide mechanism hinge structure that meets the requirements is obtained.

[0034] Furthermore, the step three specifically includes:

[0035] 3.1 Model division and mesh refinement

[0036] According to the geometric characteristics of the flexible guide mechanism, different types of mesh units and mesh encryption strategies are used to ensure the mesh accuracy of the hinge area where stress is concentrated, while reducing the calculation amount in other areas;

[0037] 3.2 Node numbering and unit division

[0038] According to the structural characteristics of the flexible guide mechanism, the node numbering rules are defined to ensure the continuity and consistency of the node numbers in different areas. The unit types are selected to divide different areas to ensure the reasonable grid distribution of the model.

[0039] 3.3 Calculation and solution of stiffness matrix

[0040] Based on the finite element mesh, the stiffness matrix of each unit is calculated and integrated into the overall stiffness matrix. The system equations are solved using the direct method or iterative method to calculate the displacement, stress and strain distribution of each node and analyze the mechanical properties of the flexible guide mechanism.

[0041] 3.4 Grid independence analysis

[0042] Mesh density test and gradual refinement ensure the convergence of results on stress and displacement distribution. The influence of mesh density on calculation results is evaluated through mesh accuracy test, and mesh independence analysis is performed to find a balance between calculation accuracy and efficiency.

[0043] 3.5. Automatically generate grid database

[0044] Write script code to automatically generate mesh models of flexible guide mechanisms under different structural parameters, evaluate and adjust the quality of the mesh model to ensure the accuracy and quality of the mesh, and store the generated mesh data in the mesh library.

[0045] Furthermore, the step 4 specifically includes:

[0046] 4.1. Define the physical model

[0047] The three-dimensional solid mechanics model is used to solve the elastic deformation of the flexible guide mechanism through the finite element method, the stress-strain relationship is defined in combination with the material properties, and the deformation and stress distribution of the flexible guide mechanism are simulated in combination with different loading conditions;

[0048] 4.2. Establishing Boundary Conditions

[0049] Apply loads to simulate the vertical loads borne by the flexible guide mechanism under actual working conditions, set constraints at both ends of the flexible guide mechanism, and ensure that the model can accurately reflect the force and deformation characteristics of the flexible guide mechanism according to the actual working load path and boundary conditions;

[0050] 4.3 Numerical Solution

[0051] The multi-grid solver is used to analyze the displacement, stress and strain distribution of each hinge node in combination with the reduced-order model and considering the nonlinear deformation behavior of the material.

[0052] 4.4 Post-processing analysis

[0053] The performance characteristics of maximum load capacity F, maximum travel ΔZ and motion straightness δL are calculated by automated scripts and output to the result database;

[0054] 4.5 Parameter optimization and result analysis

[0055] Use a machine learning-based proxy model to perform sensitivity analysis, optimize structural parameter design, and reduce the number of finite element simulations by fitting the relationship between structural parameters and performance characteristics;

[0056] 4.6. Establishing a 3D simulation results database

[0057] The simulation results and solution data are constructed as a three-dimensional simulation result database.

[0058] Furthermore, the machine learning algorithms in step six include gradient boosting tree, multi-layer perceptron, decision tree regression, Lasso regression and radial basis function network.

[0059] Furthermore, the cross validation in step 6 selects K-fold cross validation, with K=5 or K=10.

[0060] Furthermore, the evaluation indicators in step six include mean relative error, root mean square error and explained variance score.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention ensures the effectiveness of the optimization of the flexible guide mechanism hinge design scheme through finite element simulation verification, and the machine learning method can effectively save experimental costs and improve calculation efficiency. At the same time, it has the ability of self-iteration optimization, which provides the possibility for further improving the design performance of the flexible guide mechanism hinge. In addition, considering the requirements of key performance evaluation indicators such as maximum load capacity, maximum stroke and motion straightness, the machine learning model can provide the optimal solution under various design conditions, significantly improving the design accuracy and reliability of the flexible guide mechanism hinge system. It helps to shorten the design and development cycle, and accelerate the development and market launch of products by quickly generating and evaluating multiple design schemes. At the same time, the machine learning model has multi-objective optimization capabilities, can accurately adjust between multiple design parameters, ensure the best comprehensive performance, reduce dependence on physical experiments, reduce material waste and production costs, improve overall economic benefits, and provide strong support for product development and performance optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a flow chart of the method of the present invention;

[0063] Figure 2 It is a schematic diagram of the hinge position of the energy balance flexible guide mechanism;

[0064] Figure 3 It is the schematic diagram of the hinge system of the flexible guide mechanism of the energy balance;

[0065] Figure 4 is a schematic diagram of thickness parameters of hinge 1 in the flexible guide mechanism;

[0066] Figure 5 is a schematic diagram of thickness parameters of hinge 2 and hinge 3 in the flexible guide mechanism;

[0067] Figure 6 is a schematic diagram of thickness parameters of hinges 4 and 5 in the flexible guide mechanism;

[0068] Figure 7 is a schematic diagram of thickness parameters of hinges 6 to 13 in the flexible guide mechanism;

[0069] Figure 8 is a grid schematic diagram of the flexible guide mechanism;

[0070] Fig. 9 It is a schematic diagram of the finite element simulation results of the flexible guiding mechanism. DETAILED DESCRIPTION

[0071] The technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0072] like Figures 1 to 9 As shown in the figure, the hinge stiffness design method of the energy balance flexible guide mechanism based on machine learning is combined with Figure 1 As shown, the following steps are included:

[0073] Step 1: Establish the mechanism parameter library of the flexible guide mechanism

[0074] The hinge position and system principle of the energy balance flexible guide mechanism are combined Figure 2 , Figure 3 As shown, the thickness of hinge 1 to hinge 13 is used as the structural parameter. Figure 4 The thickness parameter t1 of hinge 1 is shown, Figure 5 The thickness parameters t2 and t3 of hinge 2 and hinge 3 are shown. Figure 6 The thickness parameters t4 and t5 of hinge 4 and hinge 5 are shown. Figure 7 The thickness parameters t6-t13 of hinges 6 to 13 are displayed. By selecting the thicknesses t1-t13 of different hinges, a mechanism parameter library of the flexible guiding mechanism is constructed for subsequent three-dimensional modeling, meshing and finite element performance analysis.

[0075] Step 2: Create a 3D model library for the flexible guide mechanism

[0076] According to the thickness t1-t13 data of the hinge in the mechanism parameter library of the flexible guiding mechanism, a three-dimensional model of the flexible guiding mechanism is established through three-dimensional modeling software such as SolidWorks / ProE to obtain different design schemes, and the mechanism parameter library is converted into a three-dimensional model library of the flexible guiding mechanism that can be used for numerical simulation.

[0077] Step 3: Create a mesh library corresponding to the 3D model of the flexible guide mechanism

[0078] The 3D meshing software HyperMesh and the finite element analysis software ANSYS Workbench are used to simulate and analyze the 3D models of the flexible guide mechanism in different design schemes under different working environments. The mechanical properties of the flexible guide mechanism are analyzed by meshing the 3D model and finite element meshing. The mesh model of the flexible guide mechanism is generated according to different structural parameters. Figure 8 As shown in the figure, a mesh schematic diagram of the flexible guide mechanism is shown. The generated mesh data is used to build a mesh library corresponding to the three-dimensional model, including:

[0079] 3.1 Model division and mesh refinement

[0080] Selection of mesh unit type: Different types of mesh units and mesh encryption strategies are used according to the geometric shape and stress distribution characteristics of the flexible guide mechanism. Tetrahedron or hexahedron units are used in large-size areas of the flexible guide mechanism except for the hinges, while quadrilateral or triangular meshes are used in the weak areas of each hinge, which effectively improves the mesh accuracy of the hinge area where stress is concentrated and reduces the calculation amount in other areas;

[0081] Boundary condition processing: When performing meshing, focus on meshing the boundaries of the flexible guide mechanism to ensure accurate transmission of boundary conditions such as loads and displacements. In particular, the mesh of key contact surfaces should be kept at a high level of fineness to avoid stress concentration problems.

[0082] Mesh smoothness and transition: For large-size areas of the flexible guide mechanism except for the hinges, a relatively coarse mesh division is used, and a larger unit size is used to reduce the amount of calculation. The weak areas of each hinge are locally meshed and a smaller unit size is used to ensure that the accurate stress and strain distribution is captured. In order to ensure smooth mesh transition between large-size areas and weak areas, a smoothing algorithm should be used to reduce mesh discontinuity and avoid calculation errors caused by mesh quality problems.

[0083] The local grid is divided into multiple blocks according to the stress concentration in the weak area. Different grid division strategies are applied to different blocks to improve the accuracy of the stress concentration area and reduce the calculation difficulty.

[0084] 3.2 Node numbering and unit division

[0085] Node numbering rules and continuity processing: number each node according to the custom rules to ensure the accuracy and continuity of the node numbering in the weak area. When dividing the unit, the consistency of nodes between different areas must be ensured, especially at the junction of large-size areas and weak areas, to avoid calculation errors caused by node discontinuity;

[0086] Unit type selection and division: According to the structural characteristics of the flexible guide mechanism, select the appropriate unit type (such as SOLID187, SOLID95, etc.) and divide it into different areas to ensure that the grid distribution of the entire model is reasonable and the consistency and accuracy of the calculation results are guaranteed;

[0087] 3.3 Calculation and solution of stiffness matrix

[0088] Calculation of the stiffness matrix: Calculate the stiffness matrix of each unit based on the finite element mesh and integrate it into the overall stiffness matrix to ensure the calculation accuracy of different areas. Especially in the weak area, due to the high-precision mesh brought by mesh encryption, the accuracy of the local stiffness matrix should be ensured;

[0089] Solve the system equations using direct or iterative methods, calculate the displacement, stress and strain distribution of each node, and analyze the mechanical properties of the flexible guide mechanism to evaluate the rationality of the mechanism design;

[0090] 3.4 Grid independence analysis

[0091] Mesh density test and gradual refinement: Gradually refine the mesh and test the influence of different mesh densities on the calculation results to ensure the convergence of the results in stress and displacement distribution. Especially for weak areas, gradually increase the mesh density to ensure that accurate stress concentration phenomena are captured.

[0092] Grid accuracy and convergence evaluation: Through grid accuracy test, evaluate the impact of grid density on calculation results, and perform grid independence analysis to ensure a balance between calculation accuracy and efficiency;

[0093] Automatic mesh refinement process, introducing automated mesh refinement algorithms or adaptive meshing methods to ensure that the mesh in key areas can be adaptively adjusted to the appropriate accuracy, rather than relying on manual adjustment, to improve analysis efficiency and accuracy;

[0094] 3.5. Automatically generate grid database

[0095] Script automation processing, writing script code, automatically generating mesh models of flexible guide mechanisms under different structural parameters. Through the setting of automated scripts, meshes can be automatically divided according to the input parameters, and different mesh densities can be set for large-size areas and weak areas, reducing manual intervention;

[0096] Mesh quality assessment and optimization: when generating meshes, mesh quality assessment criteria (such as unit shape quality, distortion, etc.) are used to assess the quality of each generated mesh model, and automatic adjustments are made when necessary to improve the accuracy and quality of the mesh;

[0097] Mesh database storage and call: Store the generated mesh data in the mesh library to improve the automation processing capability of subsequent simulation analysis, and ensure that the previously generated mesh models can be quickly called in multiple iterative designs to reduce duplication of work.

[0098] Step 4: Establish a 3D simulation result database of the flexible guide mechanism

[0099] Through finite element simulation results, combined with Fig. 9 As shown in the figure, a schematic diagram of the finite element simulation results of the flexible guide mechanism is shown, and a three-dimensional simulation result database covering the performance characteristics of the maximum load capacity F, maximum stroke ΔZ and motion straightness δL is constructed, including:

[0100] 4.1. Define the physical model

[0101] A three-dimensional solid mechanics model is used to solve the elastic deformation of the flexible guide mechanism through the finite element method (FEM). In order to improve the simulation efficiency, a simplified finite element model, such as a beam unit model or a shell unit model, can be selected according to the actual situation, especially when the geometric complexity is low, which can reduce the computing resources and maintain the accuracy.

[0102] Define boundary conditions. According to the actual physical situation, set fixed support and free boundary conditions, and simulate in combination with the nonlinearity of the material. The hinge material is usually 7075 aluminum alloy or beryllium bronze. Titanium alloy can be used when necessary to meet special environmental requirements.

[0103] Material parameter definition, using the material's elastic modulus, Poisson's ratio and other physical properties to define the stress-strain relationship, combined with different loading conditions, to simulate the deformation and stress distribution of the flexible guide mechanism. To improve simulation efficiency, if the deformation is small, a small strain nonlinear model can be used to simplify the calculation;

[0104] 4.2. Establishing Boundary Conditions

[0105] Apply load to simulate the vertical load borne by the flexible guide mechanism under actual working conditions as the input variable in the simulation. To improve the simulation efficiency, a simplified load model can be used to avoid complex load paths.

[0106] Constraint setting: Constraints are set at both ends of the flexible guide mechanism to ensure that its deformation conforms to the actual working conditions and maintain the accuracy of the simulation results. If the structure and load are symmetrical, the symmetric boundary conditions can be used to reduce the model size and thus the amount of calculation;

[0107] Load path and boundary conditions: according to the actual working load path and boundary conditions, ensure that the model can accurately reflect the force and deformation characteristics of the flexible guide mechanism. If contact problems are involved, a penalty contact model can be used to reduce the number of contact iterations and improve the simulation speed.

[0108] 4.3 Numerical Solution

[0109] Solution method selection: use multi-grid solver to replace traditional solver to reduce the number of iterations and improve calculation efficiency;

[0110] Stress and strain distribution calculation, calculate the displacement, stress and strain distribution of each hinge node, pay special attention to the stress concentration phenomenon in the weak area, combine the reduced-order model (ROM) to reduce the calculation degrees of freedom by simplifying the system mode to improve efficiency;

[0111] Material nonlinear solution, considering the nonlinear deformation behavior of the material, obtains the key performance characteristics of stress distribution, maximum stroke ΔZ and motion straightness δL, introduces an iterative optimization algorithm, and optimizes the solution process according to the nonlinear characteristics of the material;

[0112] 4.4 Post-processing analysis

[0113] The maximum load-bearing capacity F is calculated by integrating the stress of each node to obtain the maximum load-bearing capacity F of the flexible guide mechanism. An automated post-processing tool can be introduced to automatically extract the key data of this performance characteristic through scripts to accelerate the analysis of the results.

[0114] The maximum stroke ΔZ calculation, through the relationship between displacement and force, the maximum stroke ΔZ of the flexible guide mechanism is obtained. This performance characteristic can be automatically calculated using an automated post-processing script to reduce manual processing time;

[0115] Motion straightness δL calculation: by analyzing the straightness of the flexible guide mechanism in vertical motion, the motion straightness δL is calculated and the performance characteristics are automatically output to the result database;

[0116] 4.5 Parameter optimization and result analysis

[0117] Analysis of the influence of structural parameters, by changing the thickness of the hinge t1-t13, the influence on the maximum load capacity F, maximum stroke ΔZ and motion straightness δL of the performance characteristics of the flexible guide mechanism is studied. In order to speed up the optimization process, a machine learning-based agent model is used. The agent model is trained through simulation results to quickly predict the performance in the parameter space;

[0118] Sensitivity analysis and optimization: sensitivity analysis is performed on the hinge thickness t1-t13 as input variables to optimize the structural parameter design. By fitting the relationship between structural parameters and performance characteristics, the number of finite element simulations is reduced and the performance prediction accuracy is improved.

[0119] 4.6. Establishing a 3D simulation results database

[0120] The simulation results and solution data are constructed into a three-dimensional simulation result database, which contains the performance characteristic data of maximum load capacity F, maximum stroke ΔZ and motion straightness δL.

[0121] Step 5: Use machine learning methods to establish a library of mechanism parameters and mechanism characteristic parameters

[0122] Using machine learning methods, according to the parameters generated by the finite element simulation results of ANSYS Workbench, a mapping relationship between the structural parameters of the flexible guiding mechanism and its performance characteristics under specific load conditions is established as the mechanism parameters and mechanism characteristic parameter library for machine learning. The input parameters include the thickness t1-t13 of the hinge, and the output parameters include the maximum load capacity F, the maximum stroke ΔZ and the motion straightness δL.

[0123] Step 6: Optimize the design scheme through machine learning methods

[0124] 6.1. Machine Learning Model Optimization

[0125] In view of the multiple input conditions of 13 input parameters, namely the thickness of the hinge t1-t13, a machine learning algorithm is used to comprehensively evaluate and select multiple models to ensure the accuracy and stability of the hinge design of the flexible guide mechanism. The best performing machine learning model is gradually selected based on the input and output parameters. Each machine learning algorithm is evaluated based on its performance under high-dimensional input data. The machine learning algorithms include:

[0126] Gradient Boosted Trees (GBT), which performs well in regression problems and can effectively handle complex nonlinear relationships. It is suitable for processing high-dimensional input data and improves prediction accuracy by gradually building a decision tree model.

[0127] Multilayer Perceptron (MLP), as a deep learning model, is particularly suitable for nonlinear problems. It can capture complex input-output relationships through the nonlinear activation function of the hidden layer and is suitable for modeling the multi-input characteristics of the flexible hinge system.

[0128] Decision Tree Regression: This method can quickly segment the data space when there are many input variables by constructing a series of decision nodes. It is suitable for finding the direct impact of each input parameter on the performance output.

[0129] Lasso Regression: This method can effectively reduce the complexity of the model in the case of multiple input variables by introducing L1 regularization, and screen out the most significant features for the results. It is suitable for optimizing the model performance under multiple input conditions.

[0130] Radial Basis Function Network (RBFN): This method is a neural network model that uses radial basis function as the hidden layer activation function. It can better handle local nonlinear problems and is suitable for nonlinear mapping of flexible guide mechanisms under different parameter conditions.

[0131] 6.2. Preventing Overfitting and Data Preprocessing

[0132] In order to prevent overfitting during the training of machine learning algorithms and improve the generalization ability of the model, the mechanism parameters and mechanism characteristic parameter libraries are divided into training sets and validation sets in an appropriate ratio. The recommended ratio is 8:2 or 7:3. At the same time, necessary data preprocessing is performed, including feature scaling (such as standardization or normalization), missing value processing, and abnormal data cleaning to ensure the quality and consistency of input data. To cope with multiple input situations, feature selection techniques such as principal component analysis (PCA) or Lasso Regression can be further applied to reduce feature redundancy and improve model efficiency. Model selection and hyperparameter optimization are performed using grid search or Bayesian optimization methods to ensure that each machine learning algorithm performs optimally under high-dimensional input data. In addition, cross-validation (such as K-fold cross-validation, usually K=5 or K=10) is used to evaluate the performance of the model on different data subsets to further verify the robustness of the model and avoid overfitting.

[0133] 6.3 Model Evaluation Metrics

[0134] The machine learning model evaluation indicators suitable for high-dimensional input data are used to evaluate the prediction effect of the flexible guide mechanism hinge design. The evaluation indicators include mean relative error (MRE), root mean square error (RMSE) and explained variance score (EVS). The above evaluation indicators can better reflect the prediction accuracy and generalization ability of the model under high-dimensional input. The specific formula of the evaluation indicators is as follows:

[0135]

[0136] By using the above evaluation indicators to comprehensively evaluate the prediction effect of the model, the performance of the model under multi-input conditions can be more accurately captured. Finally, the best corresponding machine learning algorithm for different structural parameter combinations and performance characteristics is selected to obtain more efficient and stable model prediction results. The selection of each machine learning algorithm is based on minimizing the average relative error, reducing the root mean square error, and improving the explained variance score to ensure the prediction accuracy and robustness of the model under multi-dimensional input;

[0137] 6.4. Output Fitting and Model Optimization

[0138] Comprehensively compare the fitting effects of various output parameters under different input parameters, use evaluation indicators to select the best machine learning algorithm corresponding to each output parameter, and select the optimal model based on the following evaluation indicators:

[0139] MRE minimization: By reducing the average relative error of the model under all input parameters, the relative error of the model to the real data is ensured to be as small as possible, thus improving the overall prediction accuracy;

[0140] RMSE minimization: By reducing the root mean square error between the model's predicted values ​​and the actual values, it ensures that the model can effectively handle large errors and avoid excessive deviations for certain input parameters;

[0141] EVS maximization: By improving the model's ability to explain data changes, it ensures its robustness and adaptability under different input parameters.

[0142] Finally, the optimal machine learning model combination for each output parameter is selected based on the optimization of the evaluation indicators to ensure the accuracy, stability and generalization ability of the model prediction under multiple input dimensions.

[0143] Step 7: Construct a library of flexible guide mechanism solutions under different indicator systems

[0144] 7.1. Machine Learning Methods to Predict the Optimal Design Solution

[0145] According to the input hinge thickness t1-t13, the flexible guide mechanism hinge design with the best maximum load capacity F, maximum stroke ΔZ and motion linearity δL is predicted;

[0146] 7.2. Build a 3D model

[0147] Use 3D modeling software such as SolidWorks / ProE to build a 3D model of the flexible guide mechanism, ensuring that the model includes the thickness t1-t13 of all hinges to accurately simulate the real structure;

[0148] 7.3 Simulation and performance analysis

[0149] The finite element analysis software ANSYS Workbench was used to simulate the three-dimensional model of the flexible guide mechanism, analyze the mechanical behavior of the model under different loads, and calculate the maximum load capacity F, maximum stroke ΔZ and motion straightness δL through the simulation results to evaluate the overall performance of the hinge of the flexible guide mechanism;

[0150] 7.4 Iterative Optimization of Design Parameters

[0151] Based on comprehensive consideration of the performance characteristics of the maximum load capacity F, maximum stroke ΔZ and motion straightness δL, the structural parameters of the flexible guide mechanism hinge are optimized, and the mechanism parameters and mechanism characteristic parameter library are updated until a flexible guide mechanism hinge structure that meets the requirements is obtained.

[0152] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other forms of assembly without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations within the meaning and range of equivalents of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0153] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning, characterized in that: The following steps are involved: Step 1: Establish the mechanism parameter library of the flexible guide mechanism The thickness of the hinge is used as a structural parameter, and the mechanism parameter library of the flexible guide mechanism is constructed by selecting different hinge thicknesses t1-t13; Step 2: Create a 3D model library for the flexible guide mechanism According to the thickness t1-t13 data of the hinge in the mechanism parameter library of the flexible guide mechanism, a three-dimensional model of the flexible guide mechanism is established by three-dimensional modeling software to obtain different design schemes, and the mechanism parameter library is converted into a three-dimensional model library of the flexible guide mechanism that can be used for numerical simulation; Step 3: Create a mesh library corresponding to the 3D model of the flexible guide mechanism The three-dimensional models of the flexible guide mechanism in different design schemes are simulated and analyzed under different working environments using three-dimensional meshing software and finite element analysis software. The mechanical properties of the flexible guide mechanism are analyzed through meshing of the three-dimensional model and finite element meshing. The mesh model of the flexible guide mechanism is generated according to different structural parameters, and the mesh library corresponding to the three-dimensional model is constructed using the generated mesh data. Step 4: Establish a 3D simulation result database of the flexible guide mechanism Through the finite element simulation results, a three-dimensional simulation result database covering the performance characteristics of maximum load capacity F, maximum stroke ΔZ and motion straightness δL is constructed; Step 5: Use machine learning methods to establish a library of mechanism parameters and mechanism characteristic parameters By using machine learning methods and based on the parameters generated by finite element simulation results, a mapping relationship between the structural parameters of the flexible guide mechanism and its performance characteristics under specific load conditions is established as a mechanism parameter and mechanism characteristic parameter library for machine learning; Step 6: Optimize the design scheme through machine learning methods 6.

1. Machine Learning Model Optimization Use machine learning algorithms to comprehensively evaluate and select multiple models, and gradually select the best performing machine learning models based on structural parameter input and performance characteristic output; 6.

2. Preventing Overfitting and Data Preprocessing Divide the mechanism parameters and mechanism characteristic parameter library into training set and validation set in proportion, perform data preprocessing to ensure the quality and consistency of input data, perform model selection and hyperparameter tuning, and use cross-validation to evaluate the performance of the model on different data subsets to avoid overfitting; 6.3 Model Evaluation Metrics Use evaluation indicators to comprehensively evaluate the prediction effect of the model; 6.

4. Output Fitting and Model Optimization Comprehensively compare the fitting effects of each output under different inputs, and use evaluation indicators to select the best machine learning algorithm corresponding to each structural parameter; Step 7: Construct a library of flexible guide mechanism solutions under different indicator systems 7.

1. Machine Learning Methods to Predict the Optimal Design Solution According to the thickness t1-t13 of the input hinge, the design scheme of the flexible guide mechanism hinge with the best maximum load capacity F, maximum stroke ΔZ and motion linearity δL is predicted; 7.

2. Build a 3D model Use 3D modeling software to build a 3D model of the flexible guide mechanism, ensuring that the thickness t1-t13 of all hinges is included in the model; 7.3 Simulation and performance analysis The 3D model of the flexible guide mechanism was simulated using finite element analysis software to analyze the mechanical behavior of the model under different loads, and the maximum load capacity F, maximum stroke ΔZ, and motion straightness δL were calculated through the simulation results; 7.4 Iterative Optimization of Design Parameters Based on comprehensive consideration of the performance characteristics of the maximum load capacity F, maximum stroke ΔZ and motion straightness δL, the structural parameters of the flexible guide mechanism hinge are optimized, and the mechanism parameters and mechanism characteristic parameter library are updated until a flexible guide mechanism hinge structure that meets the requirements is obtained.

2. The method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning according to claim 1, characterized in that: the step three specifically comprises: 3.1 Model division and mesh refinement According to the geometric characteristics of the flexible guide mechanism, different types of mesh units and mesh encryption strategies are used to ensure the mesh accuracy of the hinge area where stress is concentrated, while reducing the calculation amount in other areas; 3.2 Node numbering and unit division According to the structural characteristics of the flexible guide mechanism, the node numbering rules are defined to ensure the continuity and consistency of the node numbers in different areas. The unit types are selected to divide different areas to ensure the reasonable grid distribution of the model. 3.3 Calculation and solution of stiffness matrix Based on the finite element mesh, the stiffness matrix of each unit is calculated and integrated into the overall stiffness matrix. The system equations are solved using the direct method or iterative method to calculate the displacement, stress and strain distribution of each node and analyze the mechanical properties of the flexible guide mechanism. 3.4 Grid independence analysis Mesh density test and gradual refinement ensure the convergence of results on stress and displacement distribution. The influence of mesh density on calculation results is evaluated through mesh accuracy test, and mesh independence analysis is performed to find a balance between calculation accuracy and efficiency. 3.

5. Automatically generate grid database Write script code to automatically generate mesh models of flexible guide mechanisms under different structural parameters, evaluate and adjust the quality of the mesh model to ensure the accuracy and quality of the mesh, and store the generated mesh data in the mesh library.

3. The method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning according to claim 1, characterized in that: The step 4 specifically includes: 4.

1. Define the physical model The three-dimensional solid mechanics model is used to solve the elastic deformation of the flexible guide mechanism through the finite element method, the stress-strain relationship is defined in combination with the material properties, and the deformation and stress distribution of the flexible guide mechanism are simulated in combination with different loading conditions; 4.

2. Establishing Boundary Conditions Apply loads to simulate the vertical loads borne by the flexible guide mechanism under actual working conditions, set constraints at both ends of the flexible guide mechanism, and ensure that the model can accurately reflect the force and deformation characteristics of the flexible guide mechanism according to the actual working load path and boundary conditions; 4.3 Numerical Solution The multi-grid solver is used to analyze the displacement, stress and strain distribution of each hinge node in combination with the reduced-order model and considering the nonlinear deformation behavior of the material. 4.4 Post-processing analysis The performance characteristics of maximum load capacity F, maximum travel ΔZ and motion straightness δL are calculated by automated scripts and output to the result database; 4.5 Parameter optimization and result analysis Use a machine learning-based proxy model to perform sensitivity analysis, optimize structural parameter design, and reduce the number of finite element simulations by fitting the relationship between structural parameters and performance characteristics; 4.

6. Establishing a 3D simulation results database The simulation results and solution data are constructed as a three-dimensional simulation result database.

4. The method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning according to claim 1, characterized in that: The machine learning algorithms in step six include gradient boosting tree, multi-layer perceptron, decision tree regression, Lasso regression and radial basis function network.

5. The method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning according to claim 1, characterized in that: The cross validation in step 6 selects K-fold cross validation, with K=5 or K=10.

6. The method for designing hinge stiffness of an energy balance flexible guide mechanism based on machine learning according to claim 1, characterized in that: The evaluation indicators in step six include mean relative error, root mean square error and explained variance score.

Images