A double-reduced-order numerical control machine tool digital twin physical field construction method

Abstract

The application discloses a double-order-reduced numerical control machine tool digital twin physical field construction method. The physical field finite element simulation sample data set under the working state sampling space is initially reduced to a relatively sparse grid node data through an improved KNN algorithm, so that the number of point sets is kept within the real-time rendering capability range of graphics rendering software; an RBF radial basis function is used to train a digital twin physical field agent model, so that the physical field is reconstructed; a genetic algorithm is used to iteratively optimize the digital twin physical field agent model, so that the deviation between the agent physical field model and the real physical model is solved; and finally, the digital twin physical field agent model is deployed to a digital twin system and used to solve the physical field distribution in real time. The application reflects the spatial distribution of the structure field and the temperature field in the digital twin system, so that the real-time operation data driven virtual-real image mapping physical field simulation is realized.

Description

Technical Field

[0001] This invention relates to a method for constructing the finite element physical field of a CNC machine tool, and more particularly to a method for constructing the digital twin physical field of a CNC machine tool with dual order reduction. Background Technology

[0002] In industrial production operation and maintenance support, digital twin technology fully utilizes physical models to integrate online real-time data with historical data from tests and experiments, integrating multi-disciplinary and multi-performance simulation processes to complete physical performance mapping in virtual space and provide feedback on the equipment's working status. To fully reflect the spatial distribution of multiple physical attributes of industrial equipment, the physical attribute status of the equipment is typically simulated and analyzed using numerical methods such as the finite element method.

[0003] Due to the complexity of industrial production environments, only limited data can be obtained from sensors during industrial production processes. In operational status monitoring, the computationally intensive finite element analysis method struggles to meet real-time requirements and cannot synchronously present key physical properties during machine tool operation. Therefore, there is an urgent need for a digital twin system capable of real-time fitting of the physical field to achieve visualized monitoring of equipment physical properties.

[0004] The computational speed of digital twin systems is closely related to hardware computing power, algorithm complexity, and model complexity. For solving physics problems in scenarios with large datasets, appropriate surrogate models can be used for physics field fitting and solving, thus meeting the requirements of fast and accurate physics solutions in digital twin systems. As a real-time fitting method for physics field states, surrogate models can present the spatial distribution of key physical attributes of a product, driven by time-series and environmental data. Using physical parameters as input, surrogate models can solve for the three-dimensional field within seconds, thus replacing time-consuming finite element method calculations.

[0005] Model order reduction methods based on radial basis functions can meet the requirements of surrogate model solving to a certain extent. However, for commonly used virtual simulation software, the constructed surrogate model needs to consider the real-time rendering capability of the twin system. Both the communication interface between the surrogate model and the software's real-time rendering capability limit the number of model nodes. Furthermore, low-quality mesh surrogate models often result in poor accuracy. Therefore, current common surrogate models suffer from a trade-off between mesh quantity and accuracy. Summary of the Invention

[0006] To address the problems existing in the background technology, this invention proposes a dual-order reduction method for constructing the physical field of a digital twin for CNC machine tools. This method overcomes the shortcomings of existing methods, achieving real-time stable communication between the digital twin system and the proxy model, triangular mesh rendering of virtual-real mirror mapping, and real-time simulation of the physical field. It also reduces the deviation between the physical field model and the real physical model.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows:

[0008] S1: Based on the working state parameters of the CNC machine tool, establish the finite element analysis mesh model of the CNC machine tool, as well as establish the working state sampling space of the CNC machine tool and construct the physical field finite element simulation sample dataset under the working state sampling space;

[0009] S2: Based on the finite element simulation sample dataset of the physical field, the improved K-nearest neighbor algorithm is used to perform preliminary order reduction of the simulation data of the finite element analysis mesh model, so as to obtain the sparse mesh model and the simulation data corresponding to its mesh nodes.

[0010] S3: Based on the sparse mesh model and the simulation data corresponding to its mesh nodes, a digital twin physical field proxy model is established using radial basis functions to establish the mapping relationship between the mesh nodes and the sampling data collected by the sensors at key sampling points;

[0011] S4: Verify the consistency between the digital twin physical field proxy model and the real sample data of the CNC machine tool. If the verification fails, optimize the digital twin physical field proxy model until a verified digital twin physical field proxy model is obtained and deployed in the digital twin system.

[0012] S5: Input real-time sensing data into the digital twin system, and combine the mapping relationship between the sampling data collected by the sensors of the grid nodes and key sampling points. The digital twin system obtains the physical field distribution in real time through the digital twin physical field proxy model.

[0013] In step S1, a working state sampling space for the CNC machine tool is first established based on the working state parameters of the CNC machine tool. Then, the working state of the CNC machine tool is sampled in the working state sampling space to obtain machine tool sampling points. Finally, parametric simulation analysis is performed on the machine tool under the working state corresponding to each machine tool sampling point to obtain a physical field finite element simulation sample dataset.

[0014] In S1, for each triangular facet of the finite element analysis mesh model of the CNC machine tool, each edge of the triangular facet consists of two directed half-edges.

[0015] Specifically, S2 is:

[0016] S21: Based on the finite element simulation sample dataset of the physical field, after mapping all the triangular patches in the finite element analysis mesh model to three-dimensional space, it is determined whether each triangular patch is on the surface of the mesh or inside the mesh, thereby determining whether each mesh node is an internal node of the mesh or a surface node of the mesh, and obtaining a sparse mesh model.

[0017] S22: Use the K-nearest neighbor algorithm to calculate and update the physical property values ​​corresponding to the internal nodes and surface nodes of each mesh, respectively, to obtain the simulation data corresponding to each mesh node in the sparse mesh model. The weights of the internal nodes and the surface nodes of the mesh are different.

[0018] Specifically, S4 is:

[0019] S41: Perform virtual-real consistency verification on the digital twin physical field proxy model and the CNC machine tool real model. If the verification passes, the current digital twin physical field proxy model is used as the final digital twin physical field proxy model; otherwise, execute S42.

[0020] S42: Use a genetic algorithm to iteratively optimize the digital twin physical field proxy model to obtain an optimized digital twin physical field proxy model. Perform virtual-real consistency verification on the optimized digital twin physical field proxy model. If the verification passes, the current digital twin physical field proxy model is used as the final digital twin physical field proxy model. If the verification fails, proceed to S43.

[0021] S43: The working state of the CNC machine tool is resampled in the working state sampling space using the Latin hypersquare sampling method to obtain the machine tool sampling points, thereby obtaining the physical field finite element simulation sample dataset. S2, S3 and S41 are repeated until the final digital twin physical field proxy model is obtained and deployed in the digital twin system.

[0022] In S1, the physical field finite element simulation sample data set includes structural field and temperature field, and the object to which the finite element simulation sample data belongs is the key component in CNC machine tool.

[0023] The interpolation kernel function for the radial basis function is a Gaussian kernel function.

[0024] This invention employs a dual-order reduction surrogate model construction method. The finite element method (FEM) solution results are initially reduced to sparser grid node data using the KNN algorithm, keeping the number of points within the real-time rendering capabilities of the graphics rendering software. A surrogate model is trained using the RBF radial basis function to reconstruct the physical field. A genetic algorithm is used for iterative optimization of the surrogate model. For surrogate models that still exhibit significant deviations after convergence, a scheme to reconstruct the surrogate model is adopted. This enables real-time data-driven virtual-real mirror mapping physical field simulation.

[0025] Compared with existing technologies and methods, the present invention has the following beneficial effects:

[0026] This invention employs the K-nearest neighbor algorithm with fused boundary conditions to initially reduce the order of the physics model, avoiding the shift of the physics boundary fitting results towards the center. It reduces the dense grid point cloud features to the physical information values ​​of each point of the sparse grid triangular facet, keeping the number of grid points within the real-time rendering capability of the graphics rendering software.

[0027] After initially reducing the order of the physical field mesh model, this invention establishes a mapping relationship between all mesh elements and sensor sampling data at key sampling points through radial basis functions, thereby inferring the distribution of the physical properties of the entire physical field based on different boundary condition information.

[0028] This invention uses a genetic algorithm to iteratively optimize the surrogate model. For surrogate models that still have significant deviations after convergence, a new surrogate model is constructed. This reduces the deviation between the surrogate model results and the actual physical model, and improves the optimization efficiency of the surrogate model.

[0029] In summary, this invention achieves the construction of a physical field proxy model suitable for digital twin systems, thereby enabling real-time solution and simulation of the physical fields of key components in industrial production processes. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the overall process for the grinding machine spindle in an example of the present invention.

[0031] Figure 2 This is a schematic diagram illustrating the initial reduction of the physical field model using the K-nearest neighbor algorithm with fused boundary conditions during the implementation of this invention.

[0032] Figure 3 This is a schematic diagram illustrating the process of determining the consistency between the virtual and real digital twin physical field model during the implementation of this invention.

[0033] Figure 4 This is a flowchart of the algorithm used in the implementation of the present invention to iteratively optimize the digital twin physical field proxy model using a genetic algorithm.

[0034] Figure 5 This is a schematic diagram illustrating the process of using a genetic algorithm to optimize the optimal solution of a digital twin physical field proxy model during the implementation of this invention. Detailed Implementation

[0035] The present invention will be further described below with reference to the accompanying drawings and specific examples:

[0036] like Figure 1 As shown, the embodiments of the present invention are illustrated using a surface forming grinder as an example:

[0037] The geometric structure of CNC grinding machine models is complex, making it difficult to directly construct a physical field model of the entire grinding machine, and failing to highlight the stress and deformation of key components during operation. In actual machining, the main components that directly affect machining accuracy are the grinding head and the spindle. Therefore, we extract the spindle components from the CNC grinding machine model for boundary condition analysis and physical field model construction.

[0038] From the perspective of physical field analysis, the physical field of the CNC grinding machine spindle is divided into two parts: the structural field and the temperature field. The main parameter variables include the grinding depth 'a'. p Grinding head speed V s and feed rate V w The grinding force acting on the grinding wheel and the workpiece consists of two components: the normal component F of the grinding force. n and tangential component F t Grinding force can be divided into two parts: cutting deformation force and sliding force. Cutting deformation force can be further divided into chip forming force and plowing force.

[0039] The boundary conditions of the temperature field mainly consist of the grinding wheel spindle with rolling bearings at both the front and rear ends, and the grinding wheel itself. The heating of the rolling bearings is primarily affected by three factors: the type of rolling bearing, the type of lubricant, and the stress on the rolling bearing. The heat flux of the grinding wheel mainly includes the heat flux q entering the workpiece. w The heat flux q removed by the chips ch The heat flow q transferred to the grinding wheel s The heat flow q transferred to the fluid through convection f .

[0040] like Figure 1 As shown, the method includes the following steps:

[0041] S1: In the offline stage, based on the working state parameters of the CNC machine tool, a finite element analysis mesh model of the precision CNC machine tool is established, as well as the working state sampling space of the CNC machine tool and the physical field finite element simulation sample dataset constructed in the working state sampling space are established.

[0042] Among them, the physical field finite element simulation sample data includes structural field and temperature field, and the objects to which the finite element simulation sample data belongs are key components in CNC machine tools.

[0043] First, a working state sampling space for the CNC machine tool is established based on its working state parameters (such as working time, spindle speed, feed rate, vibration, ambient temperature, etc.). Then, the working state of the CNC machine tool is sampled in the working state sampling space to obtain machine tool sampling points. Finally, parametric simulation analysis is performed on the machine tool under the working state corresponding to each machine tool sampling point to obtain a physical field finite element simulation sample dataset.

[0044] In practical implementation, the parameters affecting the physical properties of the target during operation are divided into two aspects: time-series parameters and environmental parameters. Based on these two types of parameters, operational state parameters are constructed, and parametric finite element analysis is performed. Time-series parameters are variables such as execution time and motion speed that can be extracted from time-series data; environmental parameters are variables such as temperature and vibration that are sensed by sensors. The operational state parameter p can be represented as a vector: p = [a1, ..., a m ,c1,…,c n ] T , where a1,…,a m There are m timing parameters, c1, ..., c n There are n environmental parameters. Furthermore, it is necessary to determine the value range of each parameter in the working state, thereby defining the sampling space p∈A1×…A m ×C1×…×C n Subsequently, based on the determined operating state parameters, a parametric finite element model is established, and the operating state parameters are loaded onto the finite element model as boundary conditions. Simultaneously, the mesh generation of the finite element model remains consistent under all boundary conditions. Maintaining a uniform finite element mesh generation provides a constant spatial distribution of element nodes for the surrogate model construction, thereby simplifying the selection of key sampling points. Several key values ​​are selected from the sampling space of the operating state parameters, and the parametric finite element model under multiple boundary conditions is solved using a finite element analysis tool. In this embodiment, the force analysis and thermal boundary condition analysis of the spindle system are performed, and the temperature field of the spindle system is solved using finite element analysis.

[0045] For each triangular patch in the finite element analysis mesh model of a CNC machine tool, each edge of the triangular patch consists of two directed half-edges, which are used to represent its planar structure.

[0046] S2: Based on the finite element simulation sample dataset of the physical field, the improved K-nearest neighbor algorithm (i.e., the K-nearest neighbor algorithm that integrates the boundary features of the mesh model) is used to perform preliminary order reduction of the simulation data of the finite element analysis mesh model, so as to obtain the sparse mesh model and the simulation data corresponding to its mesh nodes.

[0047] S2 specifically refers to:

[0048] S21: Based on the finite element simulation sample dataset of the physical field, after mapping all triangular patches in the finite element analysis mesh model to three-dimensional space, it is determined whether each triangular patch is on the surface or inside the mesh, thereby determining whether each mesh node is an internal node or a surface node of the mesh, and obtaining a sparse mesh model; In specific implementation, a half-face S with vertex data is constructed using triangular patches. The number of times half-face S is used in all mesh elements E is used to determine whether the triangular patch is on the surface of the mesh, satisfying t=∑E∩S; when the number of times it is used t is equal to 1, the node corresponding to the half-face is a surface node of the mesh, and if the number of times it is used t is greater than 1, it is an internal node of the mesh.

[0049] S22: Use the K-nearest neighbor algorithm to calculate and update the physical property values ​​corresponding to the internal nodes and surface nodes of each mesh, respectively, to obtain the simulation data corresponding to each mesh node in the sparse mesh model. The weights of the internal nodes and the surface nodes of the mesh are different.

[0050] Specifically:

[0051] For the spindle portion of the grinding system, Ansys software was used to perform a relatively sparse mesh generation on the spindle model. To keep the number of mesh points within the real-time rendering capabilities of the digital twin system, the number of mesh points was 6750, while the original number of mesh points was 24964. The vertex set of all surface mesh triangles after deduplication is denoted as D. o Assuming the grid points are uniformly distributed, for each sparse grid node Q... i In this case, the n nearest neighbors are calculated using Euclidean distance, denoted as D. q .

[0052] For nodes inside the mesh, i.e. Then Q i Let Q be an internal node. i The physical information values ​​corresponding to the physical fields of the n nodes are {p1, p2, ..., p...} n In this embodiment, n = 15, which is related to node Q. i Distances are {d1, d2, ..., d} n}, then Q i The calculation of the point values ​​corresponding to the physical field follows the following formula:

[0053]

[0054] For nodes on the surface of the mesh, if D q ∩D o If the value is not empty, and the quantity is denoted as u, then Q i These are edge nodes in a sparse mesh. According to Q... i D, the neighboro The influence weights of n nearest neighbors are assigned based on the number of points u. Then Q i The calculation of the point values ​​corresponding to the physical field follows the following formula:

[0055]

[0056] Where, p out For Q i The physical information value of the surface nodes of the mesh in the nearest neighbor, d out For Q i The mesh surface nodes in the nearest neighbor points and Q i The distance, d k For Q i The k nearest neighbor and Q i The distance, p in For Q i The physical information value of the internal nodes of the mesh volume in the nearest neighbor points, d in For Q i The nearest neighbor nodes of the mesh and Q i The distance.

[0057] S3: In the online phase, based on the sparse mesh model and the simulation data corresponding to its mesh nodes, a digital twin physics surrogate model is established using radial basis functions (RBFs). The interpolation kernel function for the RDFs is a Gaussian kernel function. This completes a second-order reduction of the simulation data and establishes a mapping relationship between the mesh nodes and the sampling data collected by sensors at key sampling points. Through this mapping relationship, the distribution of physical properties of the entire physics field can be inferred from the discrete key sampling point sensor data.

[0058] Specifically, suppose there are N input grid nodes in the multidimensional space, each containing physical field analysis results, and the corresponding physical attribute value of each grid node at the spatial coordinates is d. i Find a mapping function f(x) = d using RBF interpolation that satisfies the following condition. i Let i = 1, 2, ..., N. Digital twin systems drive model updates through sensor data. The input variables in the training sample dataset are typically a subset of the control variables. Using partial node response values ​​from the physical field as input, a mapping relationship f(x) from these discrete measurement points to all spatial nodes is constructed to reconstruct the physical field. The radial basis interpolation function t(x) is defined as a linear combination of basis functions following the formula:

[0059]

[0060] Among them, w i These are the interpolation coefficients. Let x be the kernel function, and x be the interpolation point. i As the interpolation center, ||xxi || represents the distance from the interpolation point to the interpolation center.

[0061] The RBF interpolation coefficient matrix W is obtained from the training sample dataset, where W = Φ. -1 Let y and Φ be the radial basis function matrices, and y be the training sample matrix. A physics surrogate model is trained using the spatial coordinates of sparse grid nodes and physical parameter values ​​obtained from multiple samplings as input. By inputting variable parameters at different times, the model outputs the corresponding physical values ​​for each node.

[0062] Since the available space for direct sensor placement on the spindle is limited, and the main heat flow originates from the two areas connected to the bearings, the boundary condition update process samples from outside the bearings at both ends of the spindle. This is achieved by using the positions x = [x1, x2] at the beginning and end of the spindle. T The corresponding temperature t = [t(x1), t(x2)] T Input the surrogate model, where x1 and x2 represent the positions of the bearings at both ends of the spindle, and solve for the temperature field of the entire spindle system, such as... Figure 2 As shown.

[0063] S4: Verify the consistency between the digital twin physical field proxy model and the real sample data of the CNC machine tool (signal data collected by the sensors corresponding to the real parts). If the verification fails, optimize the digital twin physical field proxy model until a verified digital twin physical field proxy model is obtained and deployed in the digital twin system.

[0064] S4 specifically refers to:

[0065] S41: Simulation data of physical quantities collected in the digital twin physical field proxy model s and the actual model of the CNC machine tool (i.e., the signal data collected by the sensors corresponding to the actual parts). e To verify the consistency between the virtual and real physical fields, the smaller the deviation, the stronger the consistency between the virtual and real physical field mirror images, and the more accurate the surrogate model results. Figure 3 As shown. If the verification passes, the current digital twin physical field proxy model is used as the final digital twin physical field proxy model; otherwise, S42 is executed.

[0066] S42: As Figure 4 and Figure 5As shown, a genetic algorithm with global optimization capability is used to iteratively optimize the digital twin physical field proxy model. The population size is set to 50, the algorithm's termination generation is 100, and the crossover probability is 0.5. The optimized digital twin physical field proxy model is obtained. During the iterative optimization process, the relative error between the real objective function value corresponding to the possible optimal solution in the k-th iteration and the (k-1)-th iteration is used to determine whether convergence has occurred. The optimized digital twin physical field proxy model undergoes virtual-real consistency verification. If the verification passes, the current digital twin physical field proxy model is adopted as the final digital twin physical field proxy model; if the verification fails, step S43 is executed. Specifically, when the relative error between the objective function value of the real model and the proxy model corresponding to the possible optimal solution in the k-th iteration exceeds a preset threshold, the verification is considered to have failed.

[0067] S43: The working state of the CNC machine tool is resampled in the working state sampling space using the Latin hypersquare sampling method to obtain the machine tool sampling points, thereby obtaining the physical field finite element simulation sample dataset. S2, S3 and S41 are repeated until the final digital twin physical field proxy model is obtained and deployed in the digital twin system.

[0068] S5: Real-time sensor data is input into the digital twin system via socket communication. Combining the mapping relationship between the sampling data collected by the sensors of the grid nodes and key sampling points, the digital twin system obtains the physical field distribution in real time through the digital twin physical field proxy model. The digital twin system records the spatial coordinates of each node of the grid model. Based on the solved physical field distribution and the spatial coordinates of the corresponding nodes, the physical field is fitted in real time to achieve visualization output.

[0069] In the offline phase, boundary condition analysis is performed on the spindle system of the CNC grinding machine, and sample points are collected using the Latin hypersquare sampling method. Node data is then exported through physical analysis in finite element simulation software. Subsequently, the K-nearest neighbor algorithm is used to initially reduce the physical field sample data to a mesh with a specified number of nodes. A surrogate physical field model is solved using radial basis functions based on the initially reduced data. After obtaining the surrogate model, the parameter values ​​of the test nodes are input into the surrogate model and transmitted to the digital twin platform via socket communication. The twin software then fits the received mesh node parameters to the physical field. In the online phase, sensor data at specified locations is input into the surrogate model to deduce the entire physical field data for the spindle. The surrogate model's solution values ​​at the sampling point locations are compared with the actual values ​​measured by the sensors. A genetic algorithm is used to iteratively optimize the surrogate model, thereby reducing the deviation between the physical property values ​​solved by the twin system's surrogate model and the actual values. Compared to existing methods, the surrogate model solution method of this invention can meet the needs of different rendering capabilities of the digital twin system while ensuring the accuracy of the physical field solution, achieving real-time simulation of the physical field through virtual-real mirror mapping, and providing a model for the physical field data information of the digital twin.

[0070] Finally, it should be noted that the above embodiments and descriptions are only used to illustrate the technical solutions of the present invention and not to limit it. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the disclosure of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for constructing a digital twin physical field for a CNC machine tool with dual order reduction, characterized in that, Includes the following steps: S1: Based on the working state parameters of the CNC machine tool, establish the finite element analysis mesh model of the CNC machine tool, as well as establish the working state sampling space of the CNC machine tool and construct the physical field finite element simulation sample dataset under the working state sampling space; S2: Based on the finite element simulation sample dataset of the physical field, the improved K-nearest neighbor algorithm is used to perform preliminary order reduction of the simulation data of the finite element analysis mesh model, so as to obtain the sparse mesh model and the simulation data corresponding to its mesh nodes. S3: Based on the sparse mesh model and the simulation data corresponding to its mesh nodes, a digital twin physical field proxy model is established using radial basis functions to complete the second order reduction of the simulation data and establish the mapping relationship between the mesh nodes and the sampling data collected by the sensors at key sampling points. S4: Verify the consistency between the digital twin physical field proxy model and the real sample data of the CNC machine tool. If the verification fails, optimize the digital twin physical field proxy model until a verified digital twin physical field proxy model is obtained and deployed in the digital twin system. S5: Input real-time sensing data into the digital twin system, and combine the mapping relationship between the sampling data collected by the sensors of the grid nodes and key sampling points. The digital twin system obtains the physical field distribution in real time through the digital twin physical field proxy model.

2. The method for constructing a digital twin physical field for a CNC machine tool with dual order reduction according to claim 1, characterized in that, In step S1, a working state sampling space for the CNC machine tool is first established based on the working state parameters of the CNC machine tool. Then, the working state of the CNC machine tool is sampled in the working state sampling space to obtain machine tool sampling points. Finally, parametric simulation analysis is performed on the machine tool under the working state corresponding to each machine tool sampling point to obtain a physical field finite element simulation sample dataset.

3. The method for constructing a digital twin physical field for a CNC machine tool with dual order reduction according to claim 1, characterized in that, In S1, for each triangular facet of the finite element analysis mesh model of the CNC machine tool, each edge of the triangular facet consists of two directed half-edges.

4. The method for constructing a digital twin physical field for a CNC machine tool with dual order reduction according to claim 1, characterized in that, Specifically, S2 is: S21: Based on the finite element simulation sample dataset of the physical field, after mapping all the triangular patches in the finite element analysis mesh model to three-dimensional space, it is determined whether each triangular patch is on the surface of the mesh or inside the mesh, thereby determining whether each mesh node is an internal node of the mesh or a surface node of the mesh, and obtaining a sparse mesh model. S22: Use the K-nearest neighbor algorithm to calculate and update the physical property values ​​corresponding to the internal nodes and surface nodes of each mesh, respectively, to obtain the simulation data corresponding to each mesh node in the sparse mesh model. The weights of the internal nodes and the surface nodes of the mesh are different.

5. The method for constructing a digital twin physical field for a CNC machine tool with dual order reduction according to claim 1, characterized in that, Specifically, S4 is: S41: Perform virtual-real consistency verification on the digital twin physical field proxy model and the CNC machine tool real model. If the verification passes, the current digital twin physical field proxy model is used as the final digital twin physical field proxy model; otherwise, execute S42. S42: Use a genetic algorithm to iteratively optimize the digital twin physical field proxy model to obtain an optimized digital twin physical field proxy model. Perform virtual-real consistency verification on the optimized digital twin physical field proxy model. If the verification passes, the current digital twin physical field proxy model is used as the final digital twin physical field proxy model. If the verification fails, proceed to S43. S43: The working state of the CNC machine tool is resampled in the working state sampling space using the Latin hypersquare sampling method to obtain the machine tool sampling points, thereby obtaining the physical field finite element simulation sample dataset. S2, S3 and S41 are repeated until the final digital twin physical field proxy model is obtained and deployed in the digital twin system.

6. The method for constructing a digital twin physical field for a CNC machine tool with dual order reduction according to claim 1, characterized in that, In S1, the physical field finite element simulation sample data set includes structural field and temperature field, and the object to which the finite element simulation sample data belongs is the key component in CNC machine tool.

7. The method for constructing a digital twin physical field for a CNC machine tool with dual order reduction according to claim 1, characterized in that, The interpolation kernel function for the radial basis function is a Gaussian kernel function.

Images