Roll finishing processing method combined with digital twinning simulation technology

Abstract

The invention belongs to the technical field of barreling finishing machining, particularly relates to a barreling finishing machining method combined with a digital twinborn simulation technology, and solves the problems that an existing barreling finishing machining technology is difficult in real-time state monitoring, debugging of technological parameters needs to consume a large amount of time, existing analog simulation can only obtain the surface stress condition of a machined workpiece, and the machining precision is poor. The method comprises the following steps: establishing a digital twinborn model of a machined workpiece and a digital twinborn model of barreling finishing machining equipment, and carrying out simulation analysis on the barreling finishing machining process of the machined workpiece by utilizing an EDEM discrete element method and an ANSYS finite element method. Constructing a surface roughness cloud picture of the processed workpiece according to the coordinates of the surface roughness points obtained by simulation, and updating a digital twin model of the processed workpiece in real time by utilizing the surface roughness cloud picture; and the Bayesian algorithm is utilized to continuously reduce errors between analog simulation data and actual experimental data, and the surface quality of the workpiece in the machining process is predicted in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of tumbling and finishing processing, and specifically to a tumbling and finishing processing method combined with digital twin simulation technology. Background Art

[0002] The tumbling and finishing processing technology aims to improve the surface integrity of parts. Through the relative movement between the grinding tool and the part surface, the part surface is finish-machined. The tumbling and finishing processing technology has the characteristics of low cost, easy operation, and high efficiency, and is one of the most widely used surface finishing processing technologies in high-tech fields. The tumbling and finishing processing technology can effectively improve the surface performance of parts on the basis of ensuring the shape accuracy.

[0003] In the actual production process, most of the tumbling and finishing processing equipment is closed processing, and the tumbling and finishing processing technology is wet processing. The grinding tool is the combined action of a solid processing medium (tumbling and polishing block) and a liquid processing medium (grinding fluid). This makes it difficult to monitor the real-time state of the tumbling and finishing processing technology. Usually, various experimental parameters need to be continuously adjusted, such as the equipment rotation speed, the filling rate of the grinding block, the grinding block material, and the properties of the grinding fluid. Experimental equipment is used to debug and experiment with various parameters. A large amount of time and experimental consumables will be spent in this process, and the experimental results also need to be further processed. Therefore, the experimental cycle is too long. Therefore, it is particularly important to use virtual methods to simulate and simulate the processing technology. However, the common method for simulating the tumbling and finishing processing technology at present is to directly simulate and simulate the process with software. For example, the discrete element method is used to simulate the collision effect of the grinding block on the workpiece. This simulation can only obtain the force condition on the surface of the workpiece, and cannot predict the processing process of the workpiece in real time. Summary of the Invention

[0004] To overcome the technical defects that it is difficult to monitor the real-time state of the existing tumbling and finishing processing technology, and the debugging of process parameters requires a lot of time, and the existing simulation can only obtain the force condition on the surface of the workpiece and cannot predict the processing process in real time, the present invention provides a tumbling and finishing processing method combined with digital twin simulation technology.

[0005] The present invention provides a tumbling and finishing processing method combined with digital twin simulation technology, and the steps are as follows:

[0006] Step 1: Conduct reverse modeling of the surface topography of the workpiece to establish an initial digital twin model of the workpiece;

[0007] Step 2: Use the EDEM discrete element method and the ANSYS finite element method to simulate and analyze the tumbling and finishing process of the workpiece. Construct a real-time surface roughness contour map of the workpiece based on the surface point coordinates obtained from the simulation, and use the real-time surface roughness contour map of the workpiece to update the digital twin model of the workpiece in real time;

[0008] Step 3: Use the Bayesian algorithm to optimize the fitness between the simulation data of the real-time surface roughness of the workpiece in the digital twin process and the experimental data of the surface roughness of the workpiece obtained in actual processing. Continuously reduce the error between the simulation data and the actual experimental data through continuous learning and optimization until the optimal digital twin model of the workpiece is obtained, and use the optimal digital twin model of the workpiece to predict the surface quality of the workpiece in the processing process in real time.

[0009] Preferably, in Step 1, the sub-steps of establishing the digital twin model of the workpiece are:

[0010] Step 1.1: Use structured light scanning technology to obtain the scanned image of the workpiece, and obtain the surface information of the workpiece through the scanned image, where the surface information includes the size and initial surface roughness of the workpiece; the data of the material, hardness, plasticity, brightness, and cleanliness of the workpiece are the known data of the workpiece; input the size, material, hardness, plasticity, initial surface roughness, brightness, and cleanliness data of the workpiece into the initial digital twin model of the workpiece;

[0011] Step 1.2: Use Gaussian filtering to denoise and enhance the surface of the scanned image of the workpiece, and then use Canny edge detection to extract the edge information of the workpiece on the scanned image, thereby improving the clarity of the initial digital twin model of the workpiece.

[0012] Preferably, in Step 2, the sub-steps of constructing the real-time surface roughness contour map of the workpiece are:

[0013] Step 2.1: Set the properties, contact model, and boundary conditions of the tumbling and polishing blocks according to the processing parameters of the tumbling and finishing equipment, conduct EDEM simulation, and generate the motion trajectory and collision force data of the tumbling and polishing blocks;

[0014] Step 2.2: Further process the initial digital twin model of the workpiece processed in Step S1 using the ANSYS finite element method. Import the motion trajectory and collision force data of the tumbling abrasive block obtained in Step 2.1 as boundary conditions into ANSYS for finite element calculation, extract the wear amount and surface morphology data of the workpiece, simulate the surface point coordinates of the workpiece through the wear amount and surface morphology data of the workpiece, calculate the real-time surface roughness of the workpiece during the simulation process using the surface point coordinates of the workpiece, and obtain the real-time surface roughness contour map of the workpiece through the real-time surface roughness of the workpiece.

[0015] Preferably, in Step 3, the sub-steps for optimizing the degree of fit between the simulated simulation data and the actual experimental data are as follows:

[0016] Step 3.1: Define the mean square error MSE as the objective function, and measure the degree of fit between the simulated simulation data and the actual experimental data through this objective function. The formula for the mean square error MSE is:

[0017] ;

[0018] where n is the total number of control samples of the simulated simulation data and the actual experimental data, is the actual experimental data, is the simulated simulation data. The mean square error is used to reflect the difference between the simulated simulation data and the actual experimental data. The smaller the mean square error value, the better the prediction effect of the digital twin model of the workpiece;

[0019] Step 3.2: Take the rotational speed of the tumbling and finishing equipment, the filling rate of the tumbling abrasive block, the material of the tumbling abrasive block, and the properties of the grinding fluid as parameters x1, x2, x3, x4 respectively. Based on this set of parameters, perform simulation operations, and the obtained simulated simulation data is The expression relationship between the simulated simulation data and each parameter is:

[0020] ;

[0021] where β0, β1, β2, β3, β4 are different regression coefficients respectively;

[0022] Step 3.3: Use the Gaussian process model to estimate the mean square error MSE. In the initial stage, use Latin hypercube sampling to select a set of parameters x1, x2, x3, x4 within the parameter range of the four parameters as the current data points, solve the corresponding simulated simulation data through the current data points, and fit the Gaussian process model using the current data points. During this process, the Gaussian process model will learn the trends and variances in the current data points;

[0023] Step 3.4: Under the current data point, conduct actual processing experiments to obtain actual experimental data corresponding to the simulation process;

[0024] Step 3.5, calculating the value of the objective function MSE according to the simulation data of step 3.3 and the actual experimental data of step 3.4, and updating the prior distribution of the Gaussian process model based on the current data point;

[0025] Step 3.6: Based on the updated Gaussian process model in step 3.5, optimize and obtain the next parameter combination to be tested, perform simulation operation on the selected next parameter combination to be tested, and obtain new simulation data. And calculate the corresponding objective function MSE, and iterate this process continuously until the reduction of the objective function MSE is less than the set threshold or the maximum computing power is reached. At this time, the degree of fit between the simulation data and the actual experimental data is the highest, and the error between the simulation data and the actual experimental data is the smallest, indicating that the current simulation data corresponds to the optimal digital twin model of the processed workpiece.

[0026] Preferably, the method further includes step 4, performing three-dimensional modeling of the barrel finishing equipment by a CAD method to construct a digital twin model of the barrel finishing equipment; establishing an InfluxDB database to realize real-time maintenance of the barrel finishing equipment, and the sub-steps thereof are:

[0027] Step 4.1, collect the operating status, fault maintenance and repair cost of the barrel finishing equipment during the actual processing as historical operation records, wherein the operating status of the barrel finishing equipment includes the rotating speed r of the barrel finishing equipment, and use sensors to collect real-time operation records of the barrel finishing equipment, and establish an InfluxDB database based on historical operation records and real-time operation records; use InfluxQL to perform data query and analyze the operation trend of the barrel finishing equipment to identify potential faults;

[0028] Step 4.2: Based on the InfluxDB database, import the real-time rotating speed r of the roller finishing equipment and the historical operating speed r ＇ , when r=r ＇ When r≠r ＇ When the error occurs, it indicates that the barrel grinding and finishing equipment is operating abnormally, and the abnormal information is fed back to the control end of the digital twin model of the barrel grinding and finishing equipment to realize real-time monitoring of the operating status of the barrel grinding and finishing equipment. The barrel grinding and finishing equipment can be maintained in real time according to its operating status.

[0029] Compared with the prior art, the technical solution provided by the present invention has the following technical effects:

[0030] The method described in the present invention adopts digital twin technology to simulate and analyze the barrel finishing process, so as to realize the intelligent detection of processing parameters and the prediction of the state of processing equipment, enabling the digital twin models of the processed workpiece and the processing equipment to achieve virtual mapping of the processing technology under the condition of small samples or no samples, further saving experimental costs, improving the experimental accuracy, and enhancing the observability and operability of the processing technology. Among them, the digital twin models of the processed workpiece and the barrel finishing equipment are further optimized through continuous learning to improve the accuracy, and an objective function MSE is constructed to measure the degree of fit between the simulated data and the actual experimental data. The error between the simulated data and the actual experimental data is continuously reduced through continuous learning and optimization to predict the surface quality of the processed workpiece during the processing process in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings herein are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present invention and used together with the specification to explain the principles of the present invention.

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0033] Figure 1 It is a flowchart of a barrel finishing method combining digital twin simulation technology in an embodiment of the present invention;

[0034] Figure 2 It is a learning and updating flowchart of step 3 of the method described in the present invention in an embodiment of the present invention;

[0035] Figure 3 It is an interaction schematic diagram between the barrel finishing process processing table and the digital twin model in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] In order to better understand the above-mentioned objects, features, and advantages of the present invention, the solutions of the present invention will be further described below. It should be noted that, without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0037] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein. Obviously, the embodiments in the specification are only a part of the embodiments of the present invention, rather than all the embodiments.

[0038] The following combines the attached Figures 1 to 3A detailed description of specific embodiments of the present invention is provided.

[0039] Based on the fact that vertical centrifugal barrel finishing equipment has good advantages in terms of material removal ability, processing efficiency, surface brightness of parts, and improvement of surface physical and mechanical properties, and is suitable for small-sample processing of specimen types, it has certain advantages in operation and loading and unloading. Therefore, in one embodiment, the equipment selected for the barrel finishing experiment is the BJL-LL05 type vertical centrifugal barrel finishing equipment. The vertical centrifugal barrel finishing equipment loads the workpiece, barrel polishing blocks, and grinding fluid into a sealed drum in a certain proportion. The base revolves around a fixed axis, and the drum revolves in the reverse direction around its own axis, thus forming a planetary motion. Under the action of centrifugal force, the vertical centrifugal barrel finishing equipment forms a forced slip flow layer, enabling the barrel polishing blocks to collide, roll, and slide on the surface of the workpiece in the drum, thereby achieving the surface finishing of the workpiece.

[0040] As Figure 1 shown, a barrel finishing method combined with digital twin simulation technology has the following steps:

[0041] Step 1: Perform reverse modeling of the surface topography of the workpiece to establish the initial digital twin model of the workpiece.

[0042] Step 2: Use the EDEM discrete element method and the ANSYS finite element method to perform simulation analysis on the barrel finishing process of the workpiece. Based on the surface point coordinates obtained from the simulation, construct the real-time surface roughness cloud map of the workpiece, and use the real-time surface roughness cloud map of the workpiece to update the digital twin model of the workpiece in real time. The sub-steps for constructing the real-time surface roughness cloud map of the workpiece are as follows:

[0043] Step 2.1: Set the properties, contact model, and boundary conditions of the barrel polishing blocks according to the processing parameters of the vertical centrifugal barrel finishing equipment, and perform EDEM simulation to generate the motion trajectory and collision force data of the barrel polishing blocks. Among them, in the EDEM simulation, it is necessary to determine the basic parameters of the barrel polishing blocks, including two categories: the intrinsic parameters and contact parameters of the barrel polishing blocks. The intrinsic parameters of the barrel polishing blocks are the density, shear modulus, and Poisson's ratio of the barrel polishing blocks. The contact parameters of the barrel polishing blocks include the static friction coefficient, rolling friction coefficient, and collision recovery coefficient.

[0044] Step 2.2: Further process the initial digital twin model of the workpiece processed in Step S1 using the ANSYS finite element method. Assign material properties such as density, shear modulus, and Poisson's ratio to the tumbling abrasive blocks. Import the motion trajectory and collision force data of the tumbling abrasive blocks obtained in Step 2.1 as boundary conditions into ANSYS for finite element calculation. Extract the wear amount and surface morphology data of the workpiece. Simulate the surface point coordinates of the workpiece through the wear amount and surface morphology data of the workpiece. Calculate the real-time surface roughness of the workpiece during the simulation process using the surface point coordinates of the workpiece. Obtain the real-time surface roughness contour map of the workpiece through the real-time surface roughness of the workpiece;

[0045] The calculation formula for surface roughness is:

[0046] ,

[0047] where is the height deviation of the surface point, and N is the total number of surface points;

[0048] Step 3: Use the Bayesian algorithm to optimize the fit between the simulation data of the real-time surface roughness of the workpiece in the digital twin process and the experimental data of the surface roughness of the workpiece obtained in actual processing. Continuously reduce the error between the simulation data and the actual experimental data through continuous learning and optimization until the digital twin model of the workpiece is obtained. Real-time predict the surface quality of the workpiece during the processing through the optimal digital twin model of the workpiece;

[0049] As Figure 2 shown, the sub-steps for optimizing the fit between the simulation data and the actual experimental data are:

[0050] Step 3.1: Define the mean square error MSE as the objective function. Measure the fit between the simulation data and the actual experimental data through this objective function. The formula for the mean square error MSE is:

[0051] ;

[0052] where n is the total number of comparison samples of the simulation data and the actual experimental data, is the actual experimental data, is the simulation data. The mean square error is used to reflect the difference between the simulation data and the actual experimental data. The smaller the mean square error value, the better the prediction effect of the digital twin model of the workpiece;

[0053] Step 3.2: Take the rotation speed of the tumbling and finishing equipment, the filling rate of the tumbling abrasive blocks, the material of the tumbling abrasive blocks, and the properties of the grinding fluid as parameters x1, x2, x3, and x4 respectively. Perform simulation operations based on this set of parameters, and the obtained simulation data is , the expression relationship between simulation data and various parameters is:

[0054] ;

[0055] Among them, β0, β1, β2, β3, and β4 are different regression coefficients;

[0056] Step 3.3: Use the Gaussian process model to estimate the mean square error (MSE). In the initial stage, Latin hypercube sampling is used to select a set of parameters x1, x2, x3, and x4 within the parameter range of the four parameters as the current data point, and the corresponding simulation data is solved through the current data point. , use the current data points to fit the Gaussian process model. During this process, the Gaussian process model will learn the trend and variance of the current data points;

[0057] Step 3.4: Under the current data point, conduct actual processing experiments to obtain actual experimental data corresponding to the simulation process;

[0058] Step 3.5, calculating the value of the objective function MSE according to the simulation data of step 3.3 and the actual experimental data of step 3.4, and updating the prior distribution of the Gaussian process model based on the current data point;

[0059] Step 3.6: Based on the updated Gaussian process model in step 3.5, optimize and obtain the next parameter combination to be tested, perform simulation operation on the selected next parameter combination to be tested, and obtain new simulation data. And calculate the corresponding objective function MSE, and iterate this process continuously until the reduction of the objective function MSE is less than the set threshold or the maximum computing power is reached. At this time, the degree of fit between the simulation data and the actual experimental data is the highest, and the error between the simulation data and the actual experimental data is the smallest, indicating that the current simulation data corresponds to the optimal digital twin model of the processed workpiece.

[0060] like Figure 3 As shown, the workpiece, processing equipment and sensors are arranged in the roller finishing process table. The sensors collect data information of the workpiece and processing equipment. The digital twin model has a control end, a display platform and an InfluxDB database. Functions such as simulation calculation are all arranged in the control end. Information is exchanged between the control end, the display platform and the InfluxDB database, and virtual-reality interaction is carried out between the roller finishing process table and the digital twin model.

[0061] The digital twin model can combine virtual reality technology to provide immersive training and operation guidance for operators, improving their skills and work efficiency. By using digital twin technology to construct a virtual reality scene that highly restores the real barrel finishing processing environment, operators can wear VR helmets and conduct various operation trainings in the virtual environment. The system can simulate various parameter conditions, provide interactive guidance, and give real-time feedback on the training effect. During on-site operations, operators can wear augmented reality glasses to overlay the three-dimensional visualization interface of the digital twin model onto the actual equipment. The system can intelligently push maintenance steps, component structure analysis, precautions, etc. according to the operation content, providing intuitive operation guidance for operators. By combining VR / AR technology with digital twins, the limitations of time and space are broken through, and advanced digital technology is maximally utilized to empower on-site operators in barrel finishing, which helps improve the safety, proficiency, and intelligent level of processing operation.

[0062] Further, as a specific implementation manner of this embodiment, in step 1, the sub-steps of establishing the digital twin model of the workpiece to be machined are as follows:

[0063] Step 1.1: Use structured light scanning technology to obtain the scanned image of the workpiece to be machined, and obtain the surface information of the workpiece to be machined through the scanned image, where the surface information includes the size and initial surface roughness of the workpiece to be machined; the data of the material, hardness, plasticity, brightness, and cleanliness of the workpiece to be machined are the known data of the workpiece to be machined; input the data of the size, material, hardness, plasticity, initial surface roughness, brightness, and cleanliness of the workpiece to be machined into the initial digital twin model of the workpiece to be machined;

[0064] Step 1.2: Use Gaussian filtering to perform denoising and surface enhancement processing on the scanned image of the workpiece to be machined, and then use Canny edge detection to extract the edge information of the workpiece to be machined on the scanned image, thereby improving the clarity of the initial digital twin model of the workpiece to be machined;

[0065] The formula for Gaussian filtering is:

[0066] ;

[0067] where x and y are the Gaussian kernel coordinates; represents the standard deviation of the Gaussian filtering formula, which determines the width of the Gaussian function;

[0068] The gradient calculation formula for Canny edge detection is:

[0069] ;

[0070] where, and are the gradients in the horizontal and vertical directions respectively.

[0071] Further, as a specific implementation manner of this embodiment, it further includes step 4 of performing three-dimensional modeling on the barrel finishing equipment by means of CAD method to construct a digital twin model of the barrel finishing equipment; establishing an InfluxDB database to realize the real-time maintenance of the barrel finishing equipment, and its sub-steps are as follows:

[0072] Step 4.1: Collect the operating status, fault maintenance, and maintenance costs of the barrel finishing equipment during the actual processing as historical operation records. Among them, the operating status of the barrel finishing equipment includes the rotation speed r of the barrel finishing equipment. At the same time, use sensors to collect the real-time operation records of the barrel finishing equipment, and establish an InfluxDB database based on the historical operation records and real-time operation records; use InfluxQL to query and analyze the data to analyze the operation trend of the barrel finishing equipment and identify potential faults;

[0073] Step 4.2: Based on the InfluxDB database, transfer the rotation speed r of the barrel finishing equipment running in real time and the rotation speed r of the historical operation records ＇ , when r = r ＇ , it indicates that the barrel finishing equipment is operating normally; when r ≠ r ＇ , it indicates that the barrel finishing equipment is operating abnormally, and the abnormal information is fed back to the control end of the digital twin model of the barrel finishing equipment to realize the real-time monitoring of the operating status of the barrel finishing equipment, and the barrel finishing equipment can be maintained in real time according to the operating status of the barrel finishing equipment.

[0074] The above is only the specific implementation manner of the present invention, enabling those skilled in the art to understand or implement the present invention. Although the above embodiments have been described in detail, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the above embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the above embodiments, and they should all be covered by the protection scope of the claims.

Claims

1. A barrel finishing method combined with digital twin simulation technology, characterized in that, The steps are as follows: Step 1: Conduct reverse modeling of the surface topography of the workpiece to be processed and establish the initial digital twin model of the workpiece to be processed; Step 2: Use the EDEM discrete element method and the ANSYS finite element method to conduct simulation analysis on the tumbling and finishing process of the workpiece to be processed. Based on the surface point coordinates obtained from the simulation, construct the real-time surface roughness cloud map of the workpiece to be processed, and use the real-time surface roughness cloud map of the workpiece to be processed to update the digital twin model of the workpiece to be processed in real time; Step 3: Use the Bayesian algorithm to optimize the fit between the simulation data of the real-time surface roughness of the workpiece to be processed in the digital twin process and the experimental data of the surface roughness of the workpiece to be processed obtained in actual processing. Continuously reduce the error between the simulation data and the actual experimental data through continuous learning and optimization until the optimal digital twin model of the workpiece to be processed is obtained. Predict the surface quality of the workpiece to be processed during the processing process in real time through the optimal digital twin model of the workpiece to be processed.

2. The barrel finishing method combining digital twin simulation technology according to claim 1, characterized in that In Step 1, the sub-steps for establishing the digital twin model of the workpiece to be processed are: Step 1.1: Use structured light scanning technology to obtain the scanned image of the workpiece to be processed, and obtain the surface information of the workpiece to be processed through the scanned image. The surface information includes the size and initial surface roughness of the workpiece to be processed; the data of the material, hardness, plasticity, brightness, and cleanliness of the workpiece to be processed are known data of the workpiece to be processed; input the data of the size, material, hardness, plasticity, initial surface roughness, brightness, and cleanliness of the workpiece to be processed into the initial digital twin model of the workpiece to be processed; Step 1.2: Use Gaussian filtering to denoise and enhance the surface of the scanned image of the workpiece to be processed, and then use Canny edge detection to extract the edge information of the workpiece to be processed on the scanned image, thereby improving the clarity of the initial digital twin model of the workpiece to be processed.

3. The barrel finishing method combining digital twin simulation technology according to claim 1, characterized in that In Step 2, the sub-steps for constructing the real-time surface roughness cloud map of the workpiece to be processed are: Step 2.1: Set the properties, contact model, and boundary conditions of the tumbling and polishing blocks according to the processing parameters of the tumbling and finishing equipment, conduct EDEM simulation, and generate the motion trajectory and collision force data of the tumbling and polishing blocks; Step 2.2: Further process the initial digital twin model of the workpiece to be processed in Step S1 using the ANSYS finite element method. Import the motion trajectory and collision force data of the tumbling and polishing blocks obtained in Step 2.1 as boundary conditions into ANSYS for finite element calculation, extract the wear amount and surface morphology data of the workpiece to be processed, simulate the surface point coordinates of the workpiece to be processed through the wear amount and surface morphology data of the workpiece to be processed, calculate the real-time surface roughness of the workpiece to be processed during the simulation process through the surface point coordinates of the workpiece to be processed, and obtain the real-time surface roughness cloud map of the workpiece to be processed through the real-time surface roughness of the workpiece to be processed.

4. A barrel finishing method combining digital twin simulation technology according to claim 1, characterized in that In Step 3, the sub-steps for optimizing the fit between the simulation data and the actual experimental data are: Step 3.1: Define the mean square error MSE as the objective function, and measure the fit between the simulation data and the actual experimental data through this objective function. The formula for the mean square error MSE is: ； where n is the total number of control samples of the simulation data and the actual experimental data, is the actual experimental data, is the simulation data. The mean square error is used to reflect the difference between the simulation data and the actual experimental data. The smaller the mean square error value, the better the prediction effect of the digital twin model of the machined workpiece; Step 3.2: Take the rotation speed of the rotary finishing equipment, the filling rate of the rotary abrasive blocks, the material of the rotary abrasive blocks, and the properties of the grinding fluid as parameters x1, x2, x3, and x4 respectively. Based on this set of parameters, perform simulation operations, and the obtained simulation data is , and the expression relationship between the simulation data and each parameter is: ； where β0, β1, β2, β3, and β4 are different regression coefficients respectively; Step 3.3: Use the Gaussian process model to estimate the mean squared error MSE. In the initial stage, use Latin hypercube sampling to select a set of parameters x1, x2, x3, x4 within the parameter ranges of the four parameters as the current data points, and solve the corresponding simulation data through the current data points. , and use the current data points to fit the Gaussian process model. In this process, the Gaussian process model will learn the trends and variances in the current data points. Step 3.4: Conduct actual machining experiments under the current data point to obtain actual experimental data corresponding to the simulation process; Step 3.5: Calculate the value of the objective function MSE based on the simulation data in Step 3.3 and the actual experimental data in Step 3.4, and update the prior distribution of the Gaussian process model based on the current data point; Step 3.6: Based on the updated Gaussian process model in Step 3.5, optimize to obtain the next parameter combination to be tested, perform simulation operations on the selected next parameter combination to be tested, and obtain new simulation data And calculate the corresponding objective function MSE. Iterate this process continuously until the reduction amplitude of the objective function MSE is less than the set threshold or the maximum computing power is reached. At this time, the degree of fit between the simulation data and the actual experimental data is the highest, and the error between the simulation data and the actual experimental data is the smallest, indicating that the current simulation data corresponds to the optimal digital twin model of the machined workpiece.

5. A barrel finishing method combined with digital twin simulation technology according to claim 1, characterized in that It further includes Step 4: Perform 3D modeling of the barrel finishing equipment through the CAD method to construct a digital twin model of the barrel finishing equipment; establish an InfluxDB database to achieve real-time maintenance of the barrel finishing equipment, and its sub-steps are as follows: Step 4.1: Collect the operating status, fault maintenance, and maintenance costs of the barrel finishing equipment during the actual machining process as historical operation records. The operating status of the barrel finishing equipment includes the rotational speed r of the barrel finishing equipment. At the same time, use sensors to collect the real-time operation records of the barrel finishing equipment, and establish an InfluxDB database based on the historical operation records and real-time operation records; use InfluxQL for data query to analyze the operation trend of the barrel finishing equipment and identify potential faults; Step 4.2: Based on the InfluxDB database, retrieve the rotational speed r of the barrel finishing equipment during real-time operation and the rotational speed r from historical operation records. ＇ , when r = r ＇ , it indicates that the barrel finishing equipment is operating normally; when r ≠ r ＇ , it indicates that the barrel finishing equipment is operating abnormally, and the abnormal information is fed back to the control end of the digital twin model of the barrel finishing equipment to achieve real-time monitoring of the operating status of the barrel finishing equipment, and the barrel finishing equipment can be maintained in real time according to its operating status.

Images