A method for identifying dynamic characteristic parameters of rolling joint of ball screw feed system

Abstract

The present invention discloses a method for identifying the dynamic characteristic parameters of a rolling joint of a ball screw feed system, comprising: taking the ball screw feed system as the analysis object and establishing its twin mechanical model; selecting a series of elastic joint equivalent rolling joints and establishing a twin finite element model of the ball screw feed system that can simultaneously consider multiple rolling joints and their stiffness and damping in all directions; taking the dynamic characteristic parameters to be identified as input and the first six natural frequencies of the system as output, designing a deep neural network (DNN) model, and training it, optimizing the optimal number of network layers, determining the twin data model and finally completing the establishment of a dynamic digital twin model of the ball screw feed system; constructing an optimized identification model in combination with modal test data, and using a particle swarm algorithm to solve and identify various stiffness and damping parameters. According to the present invention, the ball screw feed system dynamic modeling accuracy is improved, and the dynamic characteristic parameters of each rolling joint are efficiently and accurately identified.

Description

Technical Field

[0001] The invention relates to the technical field of dynamic characteristic parameter identification of a machine tool joint, and in particular to a method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system. Background Art

[0002] The ball screw feed system has been widely used in CNC machine tools due to its advantages of high positioning accuracy and transmission efficiency, strong reliability and relatively low cost. The system is composed of screws, nuts, rolling guides, bearings, workbenches, bed bodies and other components connected by different types of joints. Among them, the rolling joints are a kind of movable joints. They are not only numerous but also relatively dispersed in distribution. Their stiffness values ​​are relatively small, which has a significant impact on the dynamic performance of the system. Therefore, it is of great significance to accurately establish the dynamic model of the ball screw feed system and accurately identify the dynamic characteristic parameters of its rolling joints for evaluating the dynamic performance of the CNC machine tool feed system and improving the design level of the machine tool. Establishing an accurate dynamic model is a prerequisite for identifying the dynamic characteristic parameters of the joints.

[0003] From the current research, there are still the following problems in the dynamic modeling of ball screw feed system and the identification of dynamic characteristic parameters of its rolling joint. First, many literatures assume that the dynamic performance in different directions is completely decoupled, and then the dynamic characteristic parameters of the rolling joint in a single direction are independently identified. Although multiple rolling joints can be considered simultaneously in dynamic modeling, it ignores the non-negligible coupling effect of the dynamic characteristics of the feed system in different directions, and the multiple rolling joints and their dynamic characteristic parameters in different directions jointly affect the overall dynamic performance of the system. Therefore, dynamic modeling has limitations, resulting in modeling accuracy and parameter identification accuracy being restricted. Second, a large number of literatures consider the joint effect of dynamic characteristic parameters in various directions of space, perform dynamic modeling for a single joint, and identify various stiffness and damping parameters. Although the joint effect of dynamic characteristic parameters in multiple directions is considered in dynamic modeling, the rolling joint under study is extracted from the ball screw feed system, which changes the assembly state of the research object, which is inconsistent with the actual situation, and also results in modeling accuracy and parameter identification accuracy being restricted. Summary of the invention

[0004] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a method for identifying the dynamic characteristic parameters of the rolling joint of a ball screw feed system, which improves the accuracy of dynamic modeling of the ball screw feed system and efficiently and accurately identifies the dynamic characteristic parameters of each rolling joint. In order to achieve the above-mentioned purpose and other advantages according to the present invention, a method for identifying the dynamic characteristic parameters of the rolling joint of a ball screw feed system is provided, comprising:

[0005] S1. Taking the physical entity of the ball screw feed system as the analysis object, construct its twin mechanical model;

[0006] S2, importing the twin mechanical model constructed in step S1 into the finite element software to construct a twin finite element model of the ball screw feed system dynamics;

[0007] S3. Generate a large amount of modal analysis data through twin finite element model and parametric analysis tools to generate dynamic twin data of ball screw feed system;

[0008] S4. Using twin data as a data set, design and train a DNN model, determine the optimal network depth, construct a ball screw feed system dynamics twin data model, and complete the establishment of a ball screw feed system dynamics digital twin model;

[0009] S5. Perform modal testing on the physical entity of the ball screw feed system to obtain experimental modal data;

[0010] S6. Establish an optimization model for identifying dynamic characteristic parameters of the rolling joint of the ball screw feed system, and simultaneously identify all stiffness and damping parameters;

[0011] S7. Substitute the identification result in step S6 into the twin finite element model to solve the modal data, and compare it with the experimental results to verify the correctness and accuracy.

[0012] Preferably, in step S1, the mechanical structure, geometric dimensions and other attribute information of the physical entity of the ball screw feed system are digitally mirrored, and a twin mechanical model of the system is established using three-dimensional software.

[0013] Preferably, in step S2, the twin mechanical model is imported into the multi-body dynamics module of the finite element software COMSOL Multiphysics, and a series of matching elastic joints are selected to be equivalent to each rolling joint according to actual conditions, and the total stiffness and total damping of each joint in the x, y, and z directions are defined respectively, and then the material properties, constraints and grid division are set, so as to establish a ball screw feed system dynamic twin finite element model that can simultaneously consider multiple rolling joints and their dynamic characteristic parameters in different directions.

[0014] Preferably, in step S3, the value ranges of the stiffness and damping parameters of each rolling joint in different directions are determined according to Hertz contact theory and literature research, and random, uniform and disordered sampling is performed in the value space to obtain 6000 groups of dynamic characteristic parameter sample points to be identified, and the dynamic characteristic parameters are used as input items for finite element analysis; the parametric scanning function of the finite element software COMSOL Multiphysics is used to automatically extract the input parameters group by group and substituted them into the finite element model for modal analysis, and the first six natural frequencies of the ball screw feed system under 6000 corresponding input conditions are solved, which are used as the output item of the finite element analysis, thereby generating twin data of the digital twin dynamic model of the ball screw feed system.

[0015] Preferably, in step S4, the dynamic characteristic parameters of the rolling joint to be identified are used as input and the first six natural frequencies of the ball screw feed system are used as output, and DNN models with 2 to 10 hidden layers are designed respectively; the twin data generated in step S3 are used as the data set to train the constructed DNN model to determine the optimal number of hidden layers, thereby completing the establishment of the ball screw feed system dynamics twin data model; by integrating the twin mechanical model, the twin finite element model and the twin data model, the ball screw feed system dynamics digital twin model is established.

[0016] Preferably, in step S5, the LMS Test.Lab vibration and noise testing system is used to perform a modal test on the ball screw feed system by a single-point excitation and multi-point vibration pickup method to obtain the experimental modal results of the system.

[0017] Preferably, in step S6, the stiffness and damping parameters to be identified are used as design variables, and the value range of each dynamic characteristic parameter is used as a constraint condition. The optimization objective function is constructed by combining the experimental value of the natural frequency measured by the modal test and the predicted value of the natural frequency driven by the digital twin dynamic model, and an optimization model for identifying the dynamic characteristic parameters of the rolling joint of the ball screw feed system is established. The model is solved using the particle swarm optimization algorithm, and the stiffness and damping parameters of all rolling joints in different directions are identified at the same time.

[0018] Preferably, in step S7, the optimally identified stiffness and damping parameters are substituted into the twin finite element model for modal analysis to obtain the first six natural frequencies of the ball screw feed system based on the identified parameters, and compared with the experimental results of the corresponding orders, and the relative errors are calculated to verify the correctness and accuracy of the method of the present invention.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] (1) A series of contact characteristics of the equivalent rolling joints of elastic joints that can simultaneously consider the stiffness and damping in different directions are used to establish a finite element model of the ball screw feed system in the assembled state, which realizes the comprehensive consideration of the influence of the multi-joint and multi-directional dynamic characteristic parameters on the dynamic characteristics of the feed system.

[0021] (2) A digital twin dynamic model that can faithfully mirror the ball screw feed system is established by using three-dimensional mechanical modeling, parametric design, finite element simulation, and DNN. This allows the digital twin to be applied to the dynamic modeling of machine tool joints and the identification of their dynamic characteristic parameters, thereby improving the accuracy of dynamic modeling and parameter identification.

[0022] (3) This method is not only applicable to the rolling joint of ball screws, but also to solving the dynamic problems of complex mechanical systems with multiple joints and multi-directional dynamic characteristic parameters, and has a high value for technical promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flowchart of a method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to the present invention;

[0024] Figure 2 A twin mechanical model of a method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to the present invention;

[0025] Figure 3 A method for identifying dynamic characteristic parameters of rolling joints of a ball screw feed system according to the present invention adopts a series of elastic joints that can simultaneously consider stiffness and damping in different directions to be equivalent to each rolling joint;

[0026] Figure 4 A twin finite element model of a method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to the present invention;

[0027] Figure 5 It is a performance diagram of different depth DNN models of a method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to the present invention;

[0028] Figure 6 It is a modal test equivalent three-dimensional model and a measuring point distribution diagram of a method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to the present invention;

[0029] Figure 7 It is a convergence curve diagram of particle swarm optimization algorithm optimization of a method for identifying dynamic characteristic parameters of rolling joint of a ball screw feed system according to the present invention; DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] Reference Figure 1-7 , a method for identifying dynamic characteristic parameters of rolling joints of a ball screw feed system, comprising: S1, taking the physical entity of the ball screw feed system as an analysis object, and constructing a twin mechanical model thereof;

[0032] S2, importing the twin mechanical model constructed in step S1 into the finite element software to construct a twin finite element model of the ball screw feed system dynamics;

[0033] S3. Generate a large amount of modal analysis data through twin finite element model and parametric analysis tools to generate dynamic twin data of ball screw feed system;

[0034] S4. Using twin data as a data set, design and train a DNN model, determine the optimal network depth, construct a ball screw feed system dynamics twin data model, and complete the establishment of a ball screw feed system dynamics digital twin model;

[0035] S5. Perform modal testing on the physical entity of the ball screw feed system to obtain experimental modal data;

[0036] S6. Establish an optimization model for identifying dynamic characteristic parameters of the rolling joint of the ball screw feed system, and simultaneously identify all stiffness and damping parameters;

[0037] S7. Substitute the identification result in step S6 into the twin finite element model to solve the modal data, and compare it with the experimental results to verify the correctness and accuracy.

[0038] Furthermore, in step S1, the mechanical structure, geometric dimensions and other attribute information of the physical entity of the ball screw feed system are digitally mirrored, and a twin mechanical model of the system is established using three-dimensional software.

[0039] Furthermore, in step S2, the twin mechanical model is imported into the multi-body dynamics module of the finite element software COMSOL Multiphysics, and a series of matching elastic joints are selected to be equivalent to each rolling joint according to actual conditions, and the total stiffness and total damping of each joint in the x, y, and z directions are defined respectively, and then the material properties, constraints and grid division are set, so as to establish a ball screw feed system dynamic twin finite element model that can simultaneously consider multiple rolling joints and their dynamic characteristic parameters in different directions.

[0040] Furthermore, in step S3, according to Hertz contact theory and literature research, the value ranges of the stiffness and damping parameters of each rolling joint in different directions are determined, and random, uniform and disordered sampling is performed in the value space to obtain 6000 groups of dynamic characteristic parameter sample points to be identified, and the dynamic characteristic parameters are used as input items for finite element analysis; the parametric scanning function of the finite element software COMSOL Multiphysics is used to automatically extract the input parameters group by group and substituted them into the finite element model for modal analysis, and the first six natural frequencies of the ball screw feed system under 6000 corresponding input conditions are solved, which are used as the output item of the finite element analysis, thereby generating twin data of the digital twin dynamic model of the ball screw feed system.

[0041] Furthermore, in the step S4, the dynamic characteristic parameters of the rolling joint to be identified are used as input and the first six natural frequencies of the ball screw feed system are used as output, and DNN models with 2 to 10 hidden layers are designed respectively; the twin data generated in step S3 is used as a data set to train the constructed DNN model to determine the optimal number of hidden layers, thereby completing the establishment of the ball screw feed system dynamics twin data model; by integrating the twin mechanical model, the twin finite element model and the twin data model, the ball screw feed system dynamics digital twin model is established.

[0042] Furthermore, in step S5, the LMS Test.Lab vibration and noise testing system is used to perform a modal test on the ball screw feed system by using a single-point excitation and multi-point vibration pickup method to obtain an experimental modal result of the system.

[0043] Furthermore, in step S6, the stiffness and damping parameters to be identified are used as design variables, and the value range of each dynamic characteristic parameter is used as a constraint condition. The optimization objective function is constructed by combining the experimental value of the natural frequency measured by the modal test and the predicted value of the natural frequency driven by the digital twin dynamic model. An optimization model for identifying the dynamic characteristic parameters of the rolling joint of the ball screw feed system is established, and the particle swarm optimization algorithm is used to solve it, and the stiffness and damping parameters of all rolling joints in different directions are identified at the same time.

[0044] Furthermore, in step S7, the optimally identified stiffness and damping parameters are substituted into the twin finite element model for modal analysis to obtain the first six natural frequencies of the ball screw feed system based on the identified parameters, and compared with the experimental results of the corresponding orders, and the relative errors are calculated to verify the correctness and accuracy of the method of the present invention.

[0045] Example 1

[0046] S1: Taking the self-designed and manufactured ball screw feed system as the analysis object, mirroring the mechanical structure, geometric dimensions and other attribute information of its physical entity, and using SolidWorks to establish its twin mechanical model, such as Figure 2 shown.

[0047] S2: Import the twin mechanical model into the multi-body dynamics module of the finite element software COMSOL Multiphysics, use a series of predefined elastic joints to equal each rolling joint, and define the total stiffness and total damping of each elastic joint in each direction in the Cartesian coordinate system. Define the transmission direction of the ball screw feed system as the x direction, the support direction as the y direction, and the direction perpendicular to both x and y as the z direction. Figure 3 As shown, select the threaded joint equivalent screw-nut joint in the software and define its dynamic characteristic parameters in each direction as follows: and Select the equivalent fixed end bearing joint of the hinge joint and define its dynamic characteristic parameters in each direction as follows: and Select the equivalent simply supported end bearing joint of the cylindrical joint and define its dynamic characteristic parameters in each direction as follows: and Select the prismatic joint equivalent rolling guide joint and define its dynamic characteristic parameters in each direction as follows: and Among them, the screw is almost not loaded in the x direction at the simply supported end, and the bearing can move slightly in this direction, thereby releasing the screw deformation and thermal deformation caused by temperature changes. Then the stiffness and damping of the joint in the x direction can be regarded as 0, that is, The displacement of the rolling guide pair in the x direction is mainly the linear feed motion of the worktable driven by the nut, so the rolling guide joint is almost not subjected to load in this direction, so the stiffness and damping of the joint in the x direction can be regarded as 0, that is, As the lead screw is a rotating body, it rotates around the x-axis and is mainly affected by the axial preload of the nut (x-direction). The lead screw-nut joint, the fixed end joint and the simply supported end bearing joint are almost unaffected by the gravity of the worktable. It can be considered that the rolling joints of the three rotary motion pairs have equivalent contact characteristics in the two radial directions (y-direction and z-direction) of the feed system, that is, Then define the material properties, set the constraints, and divide the appropriate mesh to establish the twin finite element model of the ball screw feed system, such as Figure 4 shown.

[0048] S3: According to Hertz contact theory and literature research, the range of values ​​of stiffness and damping parameters in different directions of each rolling joint of the ball screw feed system is determined as shown in Table 1.

[0049] Table 1 Value ranges of dynamic characteristic parameters of each rolling joint in each direction

[0050]

[0051]

[0052] Random, uniform and disordered sampling is performed within the above range to generate 6000 groups of dynamic characteristic parameter sample points to be identified as shown in formula (1), and the dynamic characteristic parameters are used as input items of finite element analysis.

[0053]

[0054] The parametric scanning function of the finite element software COMSOL Multiphysics is used to automatically extract input parameters group by group and substitute them into the finite element model for modal analysis. The first six natural frequencies of the ball screw feed system corresponding to different input conditions shown in 6000 sets of equations (2) are solved and used as the output of the finite element analysis. Thus, the twin data of the digital twin dynamic model of the ball screw feed system is generated.

[0055] (f' 1i ,f' 2i ,f' 3i ,f' 4i ,f' 5i ,f' 6i )i=1,2,…,6000 (2)

[0056] S4: Dynamic characteristic parameter variables of the rolling joint to be identified As input, the first six natural frequencies of the ball screw feed system (f1, f2, f3, f4, f5, f6) are outputs. DNN models with 2 to 10 hidden layers are designed. The input layer of the DNN model contains 14 neurons and the output layer contains 6 neurons. The number of neurons in each layer is set to 11, "Sigmod" is selected as the activation function of the hidden layer, "Linear" is selected as the activation function of the output layer, the learning rate is set to 0.01, and the minimum gradient is 1×10 -20 The global mean square error (MSE) is used as the loss function, and its calculation formula is shown in formula (3):

[0057]

[0058] Among them, ([k,c]) is the input layer variable, N is the number of sample groups, f j is the true value during the network training process, that is, the j-th natural frequency of the ball screw feed system solved by finite element analysis, is the predicted value in network training, that is, the j-th natural frequency of the ball screw feed system predicted by the DNN model. The smaller the value of MSE, the closer the predicted value is to the true value, and the more convergent the model is.

[0059] The twin data generated in step S3 is used as the data set to train the constructed DNN model, and the training cutoff condition is set to the minimum global mean square error of 1×10 -8 Or the maximum number of iterations is 2000. When one of the two conditions is met and the loss function converges, the DNN training is considered complete and effective.

[0060] Another 500 sets of non-training sample points are taken as the test set. The trained DNN model with 2 to 10 hidden layers is used to calculate the mean absolute percentage error (MAPE) of the test set output results. The calculation formula is shown in formula (4):

[0061]

[0062] It is used as an evaluation index to reflect the closeness between the predicted natural frequency value and the finite element value in the test set. The smaller the value of this index, the closer the predicted natural frequency is. Compared with the finite element calculation results f ij The closer it is, the better the model performance and the higher the prediction accuracy. The test set MAPE results obtained for different depth DNN models are as follows: Figure 5 As shown in the figure, from the analysis results, we can see that the optimal number of hidden layers is 6, and the final number of neurons in each layer of the DNN model is determined to be 14-11-11-11-11-11-11-6.

[0063] S5: Use LMS to perform modal testing on the ball screw feed system using the single-point excitation and multi-point vibration pickup method to obtain the experimental modal results of the system. First, establish the following in the LMS modal analysis software: Figure 6 The equivalent three-dimensional model shown in the figure is constructed, and 33 measuring points are arranged, which correspond to the positions in the real object one by one. The measuring point No. 33 at the center of the workbench surface is set as the excitation point. The point is hit by a Kistler 9724A200 hammer to excite the system. The vibration signals of the point in the x, y, and z directions are picked up by the BK4525B three-axis acceleration sensor arranged at the corresponding measuring point, and transmitted to the LMS data acquisition device. The experimental data are then analyzed and processed by the LMS modal analysis software. After collecting the vibration signals of all measuring points, the LMS analysis calculates the natural frequency of the ball screw feed system, as shown in Table 2.

[0064] Table 2 Experimental values ​​of the first six natural frequencies of the ball screw feed system

[0065] Order Natural frequency experimental value (Hz) 1 59.63 2 88.47 3 178.11 4 237.17 5 279.95 6 383.59

[0066] S6: Take the stiffness and damping parameters to be identified as design variables, take the value range of each dynamic characteristic parameter as the constraint condition, and combine the experimental values ​​of the first six natural frequencies f measured by modal test j Natural frequency predictions driven by digital twin dynamics models The optimization objective function is constructed, and thus the optimization model for identifying the dynamic characteristic parameters of the rolling joint of the ball screw feed system is established as shown in formula (5).

[0067]

[0068] The particle swarm optimization algorithm is used to solve the above equation to identify the dynamic characteristic parameters of each rolling joint in each direction. The convergence curve is as follows: Figure 7 The recognition results are shown in Table 3.

[0069] Table 3 Optimized recognition results

[0070]

[0071]

[0072] S7: Substitute the optimally identified stiffness and damping parameters into the twin finite element model in the digital twin dynamics model, perform modal analysis on it, and obtain the first six natural frequencies of the ball screw feed system based on the identified parameters, and compare them with the experimental values ​​one by one. The comparison results are shown in Table 4.

[0073] Table 4 Comparison of the finite element calculated values ​​and experimental values ​​of the natural frequency of the ball screw feed system based on the identified dynamic characteristic parameters

[0074] Order Experimental value (Hz) Finite element values ​​based on identified parameters (Hz) Relative error (%) 1 59.63 59.55 -0.13 2 88.47 89.49 1.15 3 178.11 179.30 0.67 4 237.17 237.06 -0.05 5 279.95 279.46 -0.18 6 383.59 394.72 2.90

[0075] It can be seen from the above table that the relative error between the natural frequency solved based on the identified dynamic characteristic parameters and the natural frequency measured by modal testing does not exceed 3%, achieving a very high accuracy and verifying the correctness and effectiveness of this method.

[0076] The number of devices and processing scales described here are used to simplify the description of the present invention, and the application, modification and variation of the present invention will be obvious to those skilled in the art.

[0077] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and implementation modes. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system, characterized in that: The following steps are involved: S1. Taking the physical entity of the ball screw feed system as the analysis object, construct its twin mechanical model; S2, importing the twin mechanical model constructed in step S1 into the finite element software to construct a twin finite element model of the ball screw feed system dynamics; S3. Generate a large amount of modal analysis data through twin finite element model and parametric analysis tools to generate dynamic twin data of ball screw feed system; S4. Using twin data as a data set, design and train a DNN model, determine the optimal network depth, construct a ball screw feed system dynamics twin data model, and complete the establishment of a ball screw feed system dynamics digital twin model; S5. Perform modal testing on the physical entity of the ball screw feed system to obtain experimental modal data; S6. Establish an optimization model for identifying dynamic characteristic parameters of the rolling joint of the ball screw feed system, and simultaneously identify all stiffness and damping parameters; S7. Substitute the identification result in step S6 into the twin finite element model to solve the modal data, and compare it with the experimental results to verify the correctness and accuracy.

2. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In the step S1, the mechanical structure, geometric dimensions and other attribute information of the physical entity of the ball screw feed system are digitally mirrored, and a twin mechanical model of the system is established using three-dimensional software.

3. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In step S2, the twin mechanical model is imported into the multi-body dynamics module of the finite element software COMSOL Multiphysics, and a series of matching elastic joints are selected to be equivalent to each rolling joint according to the actual situation, and the total stiffness and total damping of each joint in the x, y, and z directions are defined respectively, and then the material properties, constraints and grid division are set, so as to establish a ball screw feed system dynamic twin finite element model that can simultaneously consider multiple rolling joints and their dynamic characteristic parameters in different directions.

4. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In step S3, according to Hertz contact theory and literature research, the value ranges of the stiffness and damping parameters of each rolling joint in different directions are determined, and random, uniform and disordered sampling is performed in the value space to obtain 6000 groups of dynamic characteristic parameter sample points to be identified, and the dynamic characteristic parameters are used as input items for finite element analysis; the parametric scanning function of the finite element software COMSOL Multiphysics is used to automatically extract the input parameters group by group and substitute them into the finite element model for modal analysis, and the first six natural frequencies of the ball screw feed system under 6000 corresponding input conditions are solved, which are used as the output items of the finite element analysis, thereby generating twin data of the digital twin dynamic model of the ball screw feed system.

5. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In the step S4, the dynamic characteristic parameters of the rolling joint to be identified are used as input and the first six natural frequencies of the ball screw feed system are used as output, and DNN models with 2 to 10 hidden layers are designed respectively; the twin data generated in step S3 is used as a data set to train the constructed DNN model to determine the optimal number of hidden layers, thereby completing the establishment of the ball screw feed system dynamics twin data model; by integrating the twin mechanical model, the twin finite element model and the twin data model, the ball screw feed system dynamics digital twin model is established.

6. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In the step S5, the LMS Test.Lab vibration and noise test system is used to perform a modal test on the ball screw feed system by using a single-point excitation and multi-point vibration pickup method to obtain the experimental modal results of the system.

7. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In step S6, the stiffness and damping parameters to be identified are used as design variables, and the value range of each dynamic characteristic parameter is used as a constraint condition. The optimization objective function is constructed by combining the experimental value of the natural frequency measured by the modal test and the predicted value of the natural frequency driven by the digital twin dynamic model. An optimization model for identifying the dynamic characteristic parameters of the rolling joint of the ball screw feed system is established, and the particle swarm optimization algorithm is used to solve it, and the stiffness and damping parameters of all rolling joints in different directions are identified at the same time.

8. A method for identifying dynamic characteristic parameters of a rolling joint of a ball screw feed system according to claim 1, characterized in that: In step S7, the optimally identified stiffness and damping parameters are substituted into the twin finite element model for modal analysis to obtain the first six natural frequencies of the ball screw feed system based on the identified parameters, and compared with the experimental results of the corresponding orders, and the relative errors are calculated to verify the correctness and accuracy of the method of the present invention.

Images