A method for forming chamfering of the ball track of a four-row linear ball guide

Abstract

The present invention discloses a method for chamfering a four-row ball linear guide auxiliary ballway, and relates to the field of chamfering forming technology. The method for chamfering a four-row ball linear guide auxiliary ballway is based on the double chamfering simulation process training of the four-row ball linear guide auxiliary ballway to obtain double chamfering training data. Through the double chamfering training data, a first machine learning model of a round of chamfering trajectory, a second machine learning model of a round of chamfering bevel angle, a third machine learning model of two rounds of chamfering trajectory, and a fourth machine learning model of two rounds of chamfering bevel angle are trained respectively; the chamfering forming process is controlled and corrected based on the real-time forming data and the machine learning model, and then the process processing status after chamfering forming is detected and maintained. It solves the problem that the traditional mechanical processing method is difficult to meet the requirements of complex shapes and fine chamfering, which easily leads to uneven chamfering, more burrs on the finished product, and deformation of parts due to vibration.

Description

Technical Field

[0001] The invention relates to the technical field of chamfering, and in particular to a method for chamfering a ball track of a four-row linear ball guide. Background Art

[0002] Four-row ball linear guideway is a common component in mechanical equipment, used to support and guide moving parts, such as machine tools, automation equipment, etc. These guides need to be chamfered during the manufacturing process to improve their surface quality and reduce edge stress concentration, thereby increasing service life and performance stability. In engineering manufacturing, chamfering is a common processing technology that helps improve the functionality, safety and aesthetics of the workpiece by processing the edge or corner of the workpiece into an inclined surface.

[0003] Traditional chamfering processing often relies on the skills and experience of the operator. There may be differences in the chamfering process, resulting in inconsistent chamfer shapes or unstable quality. It is difficult to meet the requirements of the chamfer quality and shape of each lane. At the same time, traditional machining methods are difficult to meet the requirements of complex shapes and fine chamfers, which can easily lead to uneven chamfering, more burrs on the finished product, deformation of parts due to vibration, and thermal deformation of parts due to the heat generated during the chamfering process.

[0004] Therefore, in view of the above problems, a method for chamfering the ball track of a four-row ball linear guide is urgently needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a method for chamfering the ball track of a four-row ball linear guide, which solves the problem that traditional mechanical processing methods are difficult to meet the requirements of complex shapes and fine chamfers, and easily lead to uneven chamfers, more burrs on the finished product, and deformation of parts due to vibration.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for chamfering the auxiliary ball track of a four-row ball linear guide, comprising the following steps: S1, based on the double chamfering simulation process training of the auxiliary ball track of the four-row ball linear guide, obtaining double chamfering training data; S2, based on the double chamfering training data, respectively training a first machine learning model of a round of chamfering trajectory, a second machine learning model of a round of chamfering bevel angle, a third machine learning model of a second round of chamfering trajectory, and a fourth machine learning model of a second round of chamfering bevel angle; S3, in the actual chamfering forming process, collecting the real-time forming data of the double chamfer of the auxiliary ball track of the four-row ball linear guide, and controlling and correcting the one-round chamfering forming process based on the real-time forming data, the first machine learning model and the second machine learning model; S4, monitoring the process processing status after the one-round chamfering forming on the basis of controlling and correcting the one-round chamfering forming process, and maintaining the abnormal process processing status after the one-round chamfering forming; S5, repeating S3 and S4 to control the two-round chamfering forming until the two-round chamfering is completed.

[0007] Furthermore, the double chamfering training data specifically includes one round of chamfering training data and two rounds of chamfering training data; the one round of chamfering training data specifically includes one round of chamfering trajectory training data and one round of chamfering bevel angle training data, the one round of chamfering trajectory training data is specifically the X-axis coordinate and Y-axis coordinate of the one round of chamfering training process in the spatial coordinate system, and the one round of chamfering bevel angle training data is specifically the chamfering bevel angle during the one round of chamfering training process; the two rounds of chamfering training data specifically includes two rounds of chamfering trajectory training data and two rounds of chamfering bevel angle training data, the two rounds of chamfering trajectory training data is specifically the X-axis coordinate and Y-axis coordinate of the two rounds of chamfering training process in the spatial coordinate system, and the two rounds of chamfering bevel angle training data is specifically the chamfering bevel angle during the two rounds of chamfering training process.

[0008] Furthermore, the specific training method of the first machine learning model of a round of chamfering trajectory is: cleaning, missing value processing and standardization of the training data of a round of chamfering trajectory; based on the speed consistency requirement of chamfering forming, dividing the training data of a round of chamfering trajectory into four columns of X-axis coordinates and Y-axis coordinates corresponding to each time point of the ball linear guide auxiliary ballway chamfering cutting tool; based on the linear regression model as the basic model of the first machine learning model of a round of chamfering trajectory, the X-axis coordinates and Y-axis coordinates corresponding to each time point of the four columns of ball linear guide auxiliary ballway chamfering cutting tools are used as the input of the first machine learning model of a round of chamfering trajectory. The first machine learning model of a round of chamfering trajectory takes the X-axis coordinates, Y-axis coordinates and cutting speed predicted cutting position labels corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool as prediction targets, and minimizes the sum of prediction errors for all cutting position labels as training targets; the first machine learning model of a round of chamfering trajectory is trained until the sum of prediction errors converges, and the model training is stopped, and the first machine learning model of the cutting position labels predicted by the X-axis coordinates, Y-axis coordinates and cutting speed output corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool is trained.

[0009] Furthermore, the specific training method of the third machine learning model of the two-round chamfering trajectory is: cleaning, missing value processing and standardization of the two-round chamfering trajectory training data; based on the speed consistency requirement of chamfering forming, dividing the two-round chamfering trajectory training data into four columns of X-axis coordinates and Y-axis coordinates corresponding to each time point of the ball linear guide auxiliary ballway chamfering cutting tool; based on the linear regression model as the basic model of the third machine learning model of the two-round chamfering trajectory, the X-axis coordinates, Y-axis coordinates and cutting speeds corresponding to each time point of the four-column ball linear guide auxiliary ballway chamfering cutting tool are used as the third machine learning model of the two-round chamfering trajectory. The third machine learning model of the two-round chamfering trajectory takes the X-axis coordinates, Y-axis coordinates and cutting speed predicted cutting position labels corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool as prediction targets, and minimizes the sum of prediction errors for all cutting position labels as training targets; the third machine learning model of the two-round chamfering trajectory is trained until the sum of prediction errors converges, and the model training is stopped, and a third machine learning model is trained that outputs predicted cutting position labels based on the X-axis coordinates and Y-axis coordinates corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool.

[0010] Furthermore, the specific training method of the second machine learning model of the round of chamfer bevel angle is: cleaning, missing value processing and standardization of the training data of the round of chamfer bevel angle; based on the consistent speed requirement of chamfering forming, dividing the training data of the round of chamfer bevel angle into four columns of chamfer bevel angles corresponding to each time point of the chamfering cutting tool of the ball linear guide auxiliary ballway; based on the linear regression model as the second machine learning model of the round of chamfer bevel angle, the chamfer bevel angles corresponding to each time point of the four columns of ball linear guide auxiliary ballways and the cutting positions are used as the first machine learning model of the round of chamfer bevel angle. Two machine learning models are input, and the second machine learning model of the round of chamfer bevel angles uses the chamfer bevel angles corresponding to each time point of the four-row ball linear guide secondary ball track and the chamfer bevel angle labels predicted at the cutting position as prediction targets, and minimizes the sum of all chamfer bevel angles as training targets; the second machine learning model of the round of chamfer bevel angles is trained until the sum of the prediction errors converges, and the model training is stopped to train a second machine learning model based on the chamfer bevel angles corresponding to each time point of the four-row ball linear guide secondary ball track and the chamfer bevel angle labels predicted at the cutting position.

[0011] Furthermore, the specific training method of the fourth machine learning model of the two-round chamfer bevel angle is: cleaning, missing value processing and standardization of the two-round chamfer bevel angle training data; based on the consistent speed requirement of chamfering forming, dividing the two-round chamfer bevel angle training data into four columns of chamfer bevel angles corresponding to each time point of the chamfering cutting tool of the ball linear guide auxiliary ballway; based on the linear regression model as the fourth machine learning model of the two-round chamfer bevel angle, the chamfer bevel angles and cutting positions corresponding to each time point of the four columns of ball linear guide auxiliary ballways are used as the fourth machine learning model of the two-round chamfer bevel angle. Four machine learning models are input, and the fourth machine learning model of the two-round chamfer bevel angle uses the chamfer bevel angle corresponding to each time point of the four-row ball linear guide secondary ball track and the chamfer bevel angle label predicted at the cutting position as the prediction target, and minimizes the sum of all chamfer bevel angles as the training target; the fourth machine learning model of the two-round chamfer bevel angle is trained until the sum of the prediction errors converges, and the model training is stopped to train a fourth machine learning model based on the chamfer bevel angle corresponding to each time point of the four-row ball linear guide secondary ball track and the chamfer bevel angle label predicted at the cutting position.

[0012] Furthermore, the specific analysis of controlling and correcting a round of chamfering forming process based on real-time forming data, the first machine learning model and the second machine learning model is as follows: real-time monitoring of the actual chamfering forming process to obtain real-time forming data, the real-time forming data including the cutting speed of a round of chamfering, the X-axis coordinate, the Y-axis coordinate and the chamfering bevel angle of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway in a round of chamfering; using the trained first machine learning model, according to the cutting speed of a round of chamfering at the current moment, the X-axis coordinate and the Y-axis coordinate of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway in a round of chamfering, predict the position of the four-row ball linear guide auxiliary ballway chamfering cutting tool at the next time point; and then using the trained second machine learning model, according to the predicted position of the four-row ball linear guide auxiliary ballway chamfering cutting tool at the next time point and the chamfering bevel angle at the current moment, predict the chamfering bevel angle at the next time point; the four-row ball linear guide auxiliary ballway chamfering cutting tool predicted by the first machine learning model The position of the chamfering cutting tool of the secondary ballway of the ball linear guide and the chamfer bevel angle predicted by the second machine learning model are compared with the real-time forming data respectively to obtain the position error value and the chamfer bevel angle error value of the secondary ballway of the four-row ball linear guide; the position error threshold and the chamfer bevel angle error threshold of the secondary ballway of the four-row ball linear guide based on the chamfering forming process of the secondary ballway of the four-row ball linear guide are obtained; when the position error value and the chamfer bevel angle error value of the secondary ballway of the four-row ball linear guide are within the position error threshold and the chamfer bevel angle error threshold of the secondary ballway of the four-row ball linear guide, respectively, the sensor parameters are automatically controlled for real-time adjustment and correction; when the position error value and the chamfer bevel angle error value of the secondary ballway of the four-row ball linear guide are greater than the position error threshold and the chamfer bevel angle error threshold of the secondary ballway of the four-row ball linear guide, respectively, an alarm is triggered to warn.

[0013] Furthermore, the specific analysis of the maintenance of abnormal process conditions after a round of chamfering is as follows: based on monitoring the process conditions after a round of chamfering, process data is obtained; the process data specifically includes: surface roughness after chamfering and fixture clamping force; fuzzy reasoning is performed on the surface roughness after chamfering and the fixture clamping force through fuzzy logic to obtain a chamfering process inspection evaluation value; a chamfering process inspection threshold is obtained based on the chamfering process requirements of the four-row ball linear guide secondary ballway; the chamfering process inspection evaluation value is compared with the chamfering process inspection threshold, and when the chamfering process inspection evaluation value is greater than or equal to the chamfering process inspection threshold, the process of the current round of chamfering is normal; when the chamfering process inspection evaluation value is less than the chamfering process inspection threshold, the process of the current round of chamfering is abnormal, and an alarm is triggered for warning maintenance.

[0014] The present invention has the following beneficial effects:

[0015] (1) This method for chamfering the ball track of a four-row ball linear guide can predict the tool position and bevel angle in real time during the actual chamfering process by training multiple machine learning models, and can achieve precise control of the first and second rounds of chamfering processes to ensure the chamfering quality and accuracy of each workpiece; using real-time forming data and machine learning models, deviations or anomalies in the chamfering process can be discovered and corrected in a timely manner to avoid the production of defective products; automated machine learning models can reduce the risks caused by human operation errors and subjective judgments, and machine learning models make decisions and adjustments based on data, which is relatively more accurate and consistent, thereby improving the stability and reliability of chamfering; through precise control and real-time correction, downtime caused by problems in the processing process can be reduced, and the optimized chamfering process can improve production efficiency, speed up production cycles, reduce maintenance costs, and ensure production continuity while ensuring quality.

[0016] (2) This method for chamfering the ball track of a four-row ball linear guide adopts a double chamfering process. On the basis of the traditional single chamfering process, a secondary chamfering process is added, which can make the surface of the part after the first chamfering smoother, thereby reducing the number of burrs generated during the subsequent second chamfering. At the same time, the second chamfering can make the surface of the part after the first chamfering more uniform, reducing the impact of dimensional deviation and vibration on the part.

[0017] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The present invention is a flow chart of a method for chamfering a ball track of a four-row ball linear guide. DETAILED DESCRIPTION

[0019] The embodiment of the present application adopts a four-row ball linear guide rail auxiliary ballway chamfering forming method to solve the problem that traditional mechanical processing methods are difficult to meet the requirements of complex shapes and fine chamfers, which easily leads to uneven chamfering, more burrs on the finished product, and deformation of parts due to vibration.

[0020] The overall idea of ​​the problem in the embodiment of this application is as follows:

[0021] Through simulated process training, the training data of double chamfering is obtained, and the data includes the changes in tool position and bevel angle during the chamfering process; using the double chamfering training data, four machine learning models are trained respectively: the first machine learning model is used to predict the tool trajectory of one round of chamfering, the second machine learning model is used to predict the bevel angle of one round of chamfering, the third machine learning model is used to predict the tool trajectory of two rounds of chamfering, and the fourth machine learning model is used to predict the bevel angle of two rounds of chamfering; in the actual chamfering forming process, the forming data of double chamfering of four-row ball linear guide secondary ballway is collected in real time. Based on the real-time forming data, the previously trained machine learning model is applied to control and correct the chamfering forming process in real time, and the process processing status after the chamfering is completed is detected and repaired.

[0022] See also Figure 1 The embodiment of the present invention provides a technical solution: a method for chamfering the auxiliary ball track of a four-row ball linear guide, comprising the following steps: S1, based on the double chamfering simulation process training of the auxiliary ball track of the four-row ball linear guide, obtaining double chamfering training data; S2, based on the double chamfering training data, respectively training a first machine learning model of a round of chamfering trajectory, a second machine learning model of a round of chamfering bevel angle, a third machine learning model of a second round of chamfering trajectory, and a fourth machine learning model of a second round of chamfering bevel angle; S3, in the actual chamfering forming process, collecting the real-time forming data of the double chamfer of the auxiliary ball track of the four-row ball linear guide, and controlling and correcting the one-round chamfering forming process based on the real-time forming data, the first machine learning model and the second machine learning model; S4, monitoring the process processing status after the one-round chamfering forming on the basis of controlling and correcting the one-round chamfering forming process, and maintaining the abnormal process processing status after the one-round chamfering forming; S5, repeating S3 and S4 to control the two-round chamfering forming until the two-round chamfering is completed.

[0023] Specifically, the double chamfering training data specifically includes one round of chamfering training data and two rounds of chamfering training data; the one round of chamfering training data specifically includes one round of chamfering trajectory training data and one round of chamfering bevel angle training data, the one round of chamfering trajectory training data is specifically the X-axis coordinate and Y-axis coordinate of the one round of chamfering training process in the spatial coordinate system, and the one round of chamfering bevel angle training data is specifically the chamfering bevel angle during the one round of chamfering training process; the two rounds of chamfering training data specifically includes two rounds of chamfering trajectory training data and two rounds of chamfering bevel angle training data, the two rounds of chamfering trajectory training data is specifically the X-axis coordinate and Y-axis coordinate of the two rounds of chamfering training process in the spatial coordinate system, and the two rounds of chamfering bevel angle training data is specifically the chamfering bevel angle during the two rounds of chamfering training process.

[0024] In this implementation scheme, the first round of chamfering trajectory training data includes the coordinate changes of the tool on the X-axis and Y-axis in the spatial coordinate system. Through these data, the motion path and trajectory law of the tool in the first round of chamfering process can be learned, which is helpful to improve the accuracy and consistency of chamfering; the first round of chamfering bevel angle training data records the angle change of the chamfering bevel in the first round of chamfering process, which can help determine the optimal angle of the chamfering bevel and the chamfering end condition to ensure the functionality and aesthetics of the workpiece; the second round of chamfering trajectory training data includes the coordinate changes of the tool on the X-axis and Y-axis in the spatial coordinate system. For the two-round chamfering process, it not only continues the advantages of the first round of chamfering trajectory data, but also can provide more accurate motion trajectory and control information in complex chamfering processes; the second round of chamfering bevel angle training data records the angle change of the chamfering bevel in the second round of chamfering process. Similar to the first round of chamfering bevel angle training data, these data can help optimize the chamfering process and ensure that the bevel angle of each workpiece meets the design requirements.

[0025] Specifically, the specific training method of the first machine learning model of a round of chamfering trajectory is: cleaning, missing value processing and standardization of the training data of a round of chamfering trajectory; based on the speed consistency requirement of chamfering forming, dividing the training data of a round of chamfering trajectory into four columns of X-axis coordinates and Y-axis coordinates corresponding to each time point of the ball linear guide auxiliary ballway chamfering cutting tool; based on the linear regression model as the basic model of the first machine learning model of a round of chamfering trajectory, the X-axis coordinates and Y-axis coordinates corresponding to each time point of the four-column ball linear guide auxiliary ballway chamfering cutting tool are used as the input of the first machine learning model of a round of chamfering trajectory. A machine learning model uses the X-axis coordinates, Y-axis coordinates and cutting speed corresponding to each time point of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway as the prediction target, and minimizes the sum of the prediction errors of all cutting position labels as the training target; the first machine learning model of a round of chamfering trajectory is trained until the sum of the prediction errors reaches convergence, and the model training is stopped, and the first machine learning model of the cutting position label predicted by the X-axis coordinates, Y-axis coordinates and cutting speed corresponding to each time point of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway is trained. The specific expression of the first machine learning model is: Where t represents the time point number, The X-axis coordinate of the position of the chamfering tool of the four-row ball linear guideway pair at the next time point predicted by the model, The Y-axis coordinates of the position of the chamfering tool of the four-row ball linear guideway at the next time point predicted by the model, and the X-axis coordinates of the chamfering tool of the four-row ball linear guideway at the next time point predicted by the model. t , Y t They represent the X-axis coordinates and Y-axis coordinates of the positions of the chamfering cutting tools of the four-row ball linear guideway at the current time point t, Z represents the cutting speed of one round of chamfering, β0x , β 1x , β 2x , β 0y , β 1y , β 2y Represents the model coefficients of the first machine learning model, which are used to predict the X-axis coordinate and Y-axis coordinate of the next time point. Represents the model error term of the first machine learning model.

[0026] In this implementation scheme, the sum of prediction errors reaching convergence refers to a stopping condition in the training process of the machine learning model, which means that the prediction error of the model on the training data has been reduced to an acceptable level and no longer changes significantly or changes very little; specifically, the prediction error refers to the difference between the predicted value of the model on the training data and the true label value, which is usually measured by a loss function such as mean square error and cross entropy; the sum of errors is the sum of the prediction errors of all samples; convergence refers to the process in which the prediction error gradually decreases and tends to be stable during model training. When the training error is reduced to a sufficiently small value and the error does not change much in subsequent training steps, the error of the model converges; in each training iteration, the prediction error of the current model on the training data is calculated, and the sum of these errors is monitored. Once this sum no longer changes significantly or changes small enough, further model training is stopped, and it is considered that the model has sufficiently fit the training data, avoiding the risk of overfitting.

[0027] The training objective is set to minimize the sum of the prediction errors for all cutting position labels, which means that the purpose of model training is to make the predicted value as close to the actual cutting position as possible. During the training process, the model iteratively learns and adjusts the parameters according to the cutting position labels predicted by the input X-axis coordinates, Y-axis coordinates and cutting speed until the sum of the prediction errors converges. The model training is stopped to ensure that the model performs well on the training set and can be generalized to new data.

[0028] The X-axis coordinate and Y-axis coordinate corresponding to each time point of the cutting tool represent the position of the cutting tool in the spatial coordinate system. During the chamfering process, the cutting tool performs XY interpolation based on the linear equation and slowly performs the first chamfering at a certain rate. After the chamfering is completed, the grinding process is continued to process the straight surface and the inclined surface formed by the chamfering. The machine tool parameters are adjusted to make the two transition smoothly with basically no speed change. After a period of grinding process is completed, the second chamfering is continued. The cutting speed represents the moving speed of the cutting tool at each time point. During the chamfering process, the consistency of the cutting speed is crucial to ensuring the chamfering quality and process stability. The cutting speed can be calculated by the change in the X-axis and Y-axis coordinates per unit time or directly obtained by sensor measurement.

[0029] The specific method of obtaining the X-axis coordinates and Y-axis coordinates corresponding to each time point of the cutting tool is as follows: a sensor installed near the cutting tool or the workbench can capture the position and moving speed information of the cutting tool in real time. The sensor can be a laser rangefinder, an encoder, or a photoelectric switch, which can provide high-precision position and speed data; in the chamfering process, a data recording and acquisition system can be used to regularly record and store the position and speed data of the cutting tool; or the software monitoring and control system equipped with modern chamfering equipment can be used to monitor and record the movement of the cutting tool in real time, including the position and speed.

[0030] By training the first machine learning model of a round of chamfering trajectory, the X-axis and Y-axis coordinates of the cutting tool at different time points can be accurately predicted, which helps to ensure the accuracy and consistency of the cutting position in each chamfering process; the design clearly requires that according to the speed consistency requirement of chamfering forming, the goal of model training is to minimize the prediction error to ensure that the prediction of the cutting position at different time points can meet the consistency of the cutting speed during the chamfering process, which can avoid process quality problems caused by excessive changes in the cutting position; through data preprocessing steps such as cleaning, missing value processing and standardization processing, the quality and reliability of the input data are ensured, which helps to improve the effect of model training, so that it can better learn and generalize from the data, and then optimize the chamfering forming process; using machine learning models to predict and control chamfering trajectories realizes the automation and intelligence of the chamfering forming process, which not only reduces the need for manual intervention, but also improves production efficiency and consistency; once the trained model can be used in actual production, it can continuously collect real-time data and provide feedback to achieve the ability to continuously improve and adapt to different process conditions, which helps to maintain the efficiency and quality of chamfering forming.

[0031] Specifically, the specific training method of the third machine learning model of the two-round chamfering trajectory is as follows: cleaning, missing value processing and standardization of the two-round chamfering trajectory training data; based on the speed consistency requirement of chamfering forming, dividing the two-round chamfering trajectory training data into four columns of X-axis coordinates and Y-axis coordinates corresponding to each time point of the ball linear guide auxiliary ballway chamfering cutting tool; based on the linear regression model as the basic model of the third machine learning model of the two-round chamfering trajectory, the X-axis coordinates, Y-axis coordinates and cutting speed corresponding to each time point of the four-column ball linear guide auxiliary ballway chamfering cutting tool are used as the input of the third machine learning model of the two-round chamfering trajectory. The third machine learning model of the chamfering trajectory uses the X-axis coordinates, Y-axis coordinates and cutting position labels predicted by the cutting speed corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool as the prediction target, and minimizes the sum of the prediction errors of all cutting position labels as the training target; the third machine learning model of the two-round chamfering trajectory is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The third machine learning model is trained to output the predicted cutting position labels according to the X-axis coordinates and Y-axis coordinates corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool. The specific expression of the third machine learning model is: Where t represents the time point number, The X-axis coordinate of the position of the chamfering tool of the four-row ball linear guideway pair at the next time point predicted by the model, The Y-axis coordinates of the position of the chamfering tool of the four-row ball linear guideway at the next time point predicted by the model, and the X-axis coordinates of the chamfering tool of the four-row ball linear guideway at the next time point predicted by the model. t ′、Y t ′ respectively represent the X-axis coordinate and Y-axis coordinate of the position of the chamfering cutting tool of the four-row ball linear guideway at the current time point t, Z' represents the cutting speed of the two-wheel chamfering, β 0x ', β 1x ', β 2x ', β 0y ', β 1y ', β 2y ' represents the model coefficients of the third machine learning model, which are used to predict the X-axis coordinates and Y-axis coordinates of the next time point, Represents the model error term of the third machine learning model.

[0032] In this implementation scheme, the two-round chamfering trajectory training data is cleaned, missing values ​​are processed and standardized to ensure data quality and consistency; the processed data is divided into four columns of X-axis coordinates and Y-axis coordinates corresponding to the ball track chamfering cutting tool of the ball linear guide at each time point as input features; the linear regression model is selected as the basic model to establish the third machine learning model of the two-round chamfering trajectory; the training goal is set to minimize the sum of the prediction errors of all cutting position labels, which means that the purpose of model training is to make the predicted value as close as possible to the actual cutting position; during the training process, the model adjusts parameters according to the cutting position labels predicted by the input X-axis coordinates, Y-axis coordinates and cutting speed through iterative learning, and stops model training when the sum of the prediction errors converges, ensuring that the model performs well on the training set and can be generalized to new data.

[0033] The trained third machine learning model can accurately predict the cutting position of the two-round chamfering trajectory, which helps to ensure the chamfering quality and product consistency; the emphasis on the speed consistency requirement of chamfering forming means that the influence of cutting speed is taken into account in the model training process, which can effectively maintain the consistency of cutting speed in the two-round chamfering process and avoid process problems caused by speed fluctuations; through the analysis and training of actual chamfering trajectory data, the model can learn from a large amount of historical data, continuously optimize the prediction performance, and improve production efficiency and the stability of the chamfering process; the use of machine learning models for the prediction and control of the two-round chamfering trajectory makes the chamfering forming process more intelligent and automated, which not only reduces the need for manual intervention, but also improves production efficiency and product quality.

[0034] Specifically, the specific training method of the second machine learning model of a round of chamfer bevel angle is as follows: cleaning, missing value processing and standardization of the training data of a round of chamfer bevel angle; based on the consistent speed requirement of chamfering forming, dividing the training data of a round of chamfer bevel angle into four columns of chamfer bevel angles corresponding to each time point of the chamfering cutting tool of the ball linear guide auxiliary ballway; based on the linear regression model as the second machine learning model of a round of chamfer bevel angle, the chamfer bevel angles corresponding to each time point of the four-column ball linear guide auxiliary ballway and the cutting position are used as the input of the second machine learning model of a round of chamfer bevel angle. The second machine learning model of a round of chamfer bevel angles takes the chamfer bevel angles corresponding to each time point of the four-row ball linear guide auxiliary ballway and the chamfer bevel angle labels predicted at the cutting position as prediction targets, and minimizes the sum of all chamfer bevel angles as training targets; the second machine learning model of a round of chamfer bevel angles is trained until the sum of the prediction errors converges, and the model training is stopped. The second machine learning model is trained based on the chamfer bevel angles corresponding to each time point of the four-row ball linear guide auxiliary ballway and the chamfer bevel angle labels predicted at the cutting position. The specific expression of the second machine learning model is: In the formula represents the chamfer angle at the next time point predicted by the model, β 0d , β 1d , β 2d , β 3d represents the model coefficient of the second machine learning model, which is used to predict the chamfer angle at the next time point, X t , Y t The X-axis coordinates and Y-axis coordinates of the positions of the chamfering cutting tools of the four-column ball linear guideway at the current time point t, respectively, D t Indicates the chamfer angle at the current time point t, Represents the model error term of the second machine learning model.

[0035] In this implementation scheme, the chamfer bevel angle refers to the angle of the bevel when chamfering is performed in engineering processing. In the field of mechanical processing, chamfering is to reduce the sharpness of the edge of the workpiece, and a bevel is often formed on the edge of the workpiece, and the angle of the bevel is the chamfer bevel angle; the specific method of obtaining the chamfer bevel angle is: use special measuring tools, such as chamfering gauges, microscopes, etc. to directly measure the bevel angle formed on the workpiece, or use optical measuring equipment, such as projectors, laser measuring instruments, etc. to perform non-contact bevel angle measurement on the workpiece, and you can also use a camera to shoot the workpiece surface, and then use image processing software to analyze the bevel angle.

[0036] The training goal is to minimize the error between the predicted value and the actual slope angle label. The square loss function or other appropriate loss functions can be used for optimization. The training process updates the model parameters through optimization algorithms such as gradient descent until the training stop condition is reached.

[0037] By training a second machine learning model of the chamfering bevel angle, the accurate prediction of the bevel angle at each time point in the chamfering process can be achieved, which is crucial to ensuring the quality of chamfering and process stability; emphasizing the need for consistent speed according to chamfering forming, it means that the influence of cutting position and bevel angle is considered in the model training process. By optimizing the model training, the consistency of the bevel angle in the chamfering process can be effectively maintained to avoid process problems caused by fluctuations in the bevel angle; through data preprocessing steps such as cleaning, missing value processing and standardization processing, the quality and reliability of the input data are ensured, which helps to improve the effect of model training, enable it to better learn and generalize from the data, and thus optimize the prediction accuracy of the bevel angle in the chamfering forming process; using machine learning models to predict and control the bevel angle makes the chamfering forming process more intelligent and automated, which not only reduces the need for manual intervention, but also improves production efficiency and the stability of the chamfering process.

[0038] Specifically, the specific training method of the fourth machine learning model of the two-round chamfer bevel angle is as follows: the two-round chamfer bevel angle training data is cleaned, missing values ​​are processed and standardized; based on the consistent speed requirement of chamfering forming, the two-round chamfer bevel angle training data is divided into four columns of chamfer bevel angles corresponding to each time point of the chamfering cutting tool of the ball linear guide auxiliary ballway; based on the linear regression model as the fourth machine learning model of the two-round chamfer bevel angle, the chamfer bevel angles corresponding to each time point of the four-column ball linear guide auxiliary ballway and the cutting position are used as the fourth machine learning model input of the two-round chamfer bevel angle. The fourth machine learning model of the two-round chamfer bevel angle takes the chamfer bevel angle corresponding to each time point of the four-row ball linear guide auxiliary ballway and the chamfer bevel angle label predicted by the cutting position as the prediction target, and minimizes the sum of all chamfer bevel angles as the training target; the fourth machine learning model of the two-round chamfer bevel angle is trained until the sum of the prediction errors reaches convergence, and the model training is stopped, and the fourth machine learning model is trained according to the chamfer bevel angle corresponding to each time point of the four-row ball linear guide auxiliary ballway and the chamfer bevel angle label predicted by the cutting position. The specific expression of the fourth machine learning model is: In the formula represents the chamfer angle at the next time point predicted by the model, β 0d ′、β 1d ′、β 2d ′、β 3d ′ represents the model coefficient of the fourth machine learning model, which is used to predict the chamfer angle at the next time point, X t ′、Y t ′ respectively represent the X-axis coordinate and Y-axis coordinate of the position of the chamfering cutting tool of the four-row ball linear guideway at the current time point t, D t ′ represents the chamfer angle at the current time point t, represents the model error term of the fourth machine learning model.

[0039] In this implementation, the training goal is to minimize the error between the predicted value and the actual slope angle label. The square loss function or other appropriate loss functions can be used for optimization. The training process will update the model parameters through optimization algorithms such as gradient descent until the training stop condition is reached.

[0040] By training the fourth machine learning model of the two-round chamfering bevel angle, the bevel angle at each time point in the two-round chamfering process can be accurately predicted. In complex processes, process problems caused by fluctuations in the bevel angle can be avoided; emphasizing the cleaning, missing value processing and standardization of the two-round chamfering bevel angle training data will help improve the effect of model training, enabling it to learn from the processed high-quality data and better generalize to the actual process; considering the speed consistency requirements of chamfering forming, the effects of cutting position and bevel angle are combined in model training. By optimizing model training, the consistency of the bevel angle during the chamfering process can be effectively maintained, ensuring that the cutting bevel angle at different time points remains stable; using machine learning models to predict and control the bevel angle makes the chamfering forming process more intelligent and automated, which not only reduces the need for manual intervention, but also improves production efficiency and the stability of the chamfering process.

[0041] The training goal is to minimize the error between the predicted value and the actual slope angle label. The square loss function or other appropriate loss functions can be used for optimization. The training process updates the model parameters through optimization algorithms such as gradient descent until the training stop condition is reached.

[0042] Specifically, the specific analysis of controlling and correcting a round of chamfering forming process based on real-time forming data, the first machine learning model and the second machine learning model is as follows: real-time monitoring of the actual chamfering forming process to obtain real-time forming data, the real-time forming data including the cutting speed of a round of chamfering, the X-axis coordinate, Y-axis coordinate and chamfering bevel angle of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway in a round of chamfering; using the trained first machine learning model, according to the cutting speed of a round of chamfering at the current moment, the X-axis coordinate and Y-axis coordinate of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway in a round of chamfering, predict the four-row ball linear guide at the next time point. The position of the chamfer cutting tool of the auxiliary ballway of the guide rail; then, using the trained second machine learning model, according to the predicted position of the chamfer cutting tool of the auxiliary ballway of the four-row ball linear guide rail at the next time point and the chamfer bevel angle at the current moment, the chamfer bevel angle at the next time point is predicted; the position of the chamfer cutting tool of the auxiliary ballway of the four-row ball linear guide rail predicted by the first machine learning model and the chamfer bevel angle predicted by the second machine learning model are respectively compared with the real-time forming data to obtain the position error value and the chamfer bevel angle error value of the auxiliary ballway of the four-row ball linear guide rail. The position error value of the chamfer cutting tool of the auxiliary ballway of the four-row ball linear guide rail is expressed as: Where η1 represents the position error value of the chamfering cutting tool of the four-row ball linear guideway pair, The X-axis coordinate of the position of the chamfering cutting tool of the four-row ball linear guideway pair at the next time point predicted by the model, X t+1 Indicates the real-time X-axis coordinate, The Y-axis coordinate of the position of the chamfering tool of the four-row ball linear guideway at the next time point predicted by the model, Y t+1 Indicates the real-time Y-axis coordinate; the chamfer angle error value is expressed as: Where η2 represents the angle error of the chamfered surface. Denotes the chamfer angle at the next time point predicted by the model, D t+1 Indicates the real-time chamfer bevel angle; obtains the four-row ball linear guide auxiliary ballway chamfer cutting tool position error threshold and chamfer bevel angle error threshold based on the four-row ball linear guide auxiliary ballway chamfer forming process setting; when the four-row ball linear guide auxiliary ballway chamfer cutting tool position error value and chamfer bevel angle error value are respectively within the four-row ball linear guide auxiliary ballway chamfer cutting tool position error threshold and chamfer bevel angle error threshold, automatically controls the sensor parameters for real-time adjustment and correction; when the four-row ball linear guide auxiliary ballway chamfer cutting tool position error value and chamfer bevel angle error value are respectively greater than the four-row ball linear guide auxiliary ballway chamfer cutting tool position error threshold and chamfer bevel angle error threshold, triggers an alarm for warning.

[0043] In this implementation scheme, the position error threshold of the chamfering cutting tool and the chamfer bevel angle error threshold of the four-row ball linear guide auxiliary ballway can be set by analyzing past production data and experimental results to identify the typical error range that causes problems in actual production. The chamfering forming test can also be carried out in a controlled laboratory or simulated environment to directly measure and evaluate the cutting tool position and bevel angle errors under different conditions, and set the threshold value based on the experimental results. Through real-time monitoring and prediction, the position and chamfer bevel angle of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway can be controlled and adjusted in real time during the chamfering forming process to ensure that the cutting position and bevel angle of each step meet expectations, thereby improving the accuracy and stability of the process. The traditional chamfering forming process requires operation The operator constantly adjusts the cutting parameters to adapt to the different conditions of the workpiece. The automatic prediction and adjustment of the machine learning model reduces the need for human intervention and the risk of human error. Through automated real-time adjustment and correction, it can quickly adapt to changes in the workpiece and fluctuations in process conditions, thereby improving production efficiency. Optimized cutting parameters and process control can reduce unnecessary downtime and adjustment time, thereby increasing the stability and throughput of the production line. Precise control of the cutting position and bevel angle can ensure the consistency and accuracy of the chamfering process, thereby improving the quality of the final product. By reducing the error value and setting a reasonable error threshold, it can promptly detect and handle abnormal situations, avoid product quality problems caused by poor chamfering, avoid the production of defective products, and improve resource utilization.

[0044] Specifically, the specific analysis of maintenance for abnormal process conditions after a round of chamfering is as follows: based on monitoring the process conditions after a round of chamfering, process data is obtained; the process data specifically includes: surface roughness after chamfering and fixture clamping force; fuzzy reasoning is performed on the surface roughness after chamfering and the fixture clamping force through fuzzy logic to obtain a chamfering process detection evaluation value, which is used to quantify the effect of the chamfering process; a chamfering process detection threshold set based on the chamfering process requirements of the four-row ball linear guide secondary ballway is obtained; the chamfering process detection evaluation value is compared with the chamfering process detection threshold, and when the chamfering process detection evaluation value is greater than or equal to the chamfering process detection threshold, the process of the current round of chamfering is normal; when the chamfering process detection evaluation value is less than the chamfering process detection threshold, the process of the current round of chamfering is abnormal, triggering an alarm for warning maintenance.

[0045] In this implementation scheme, fuzzy reasoning is performed on the surface roughness after chamfering and the clamping force of the fixture through fuzzy logic to obtain the specific analysis of the chamfering process processing detection evaluation value: the surface roughness after chamfering and the clamping force of the fixture are defined as input variables, and divided into different fuzzy sets, for example: "Low", "Medium", "High" for the surface roughness after chamfering, "Low", "Medium", "High" for the clamping force, the chamfering process processing detection evaluation value is defined as the output variable, and divided into different fuzzy sets, for example: "Low", "Medium", "High" for the chamfering process processing detection evaluation value, according to the surface roughness after chamfering is inversely proportional to the chamfering process processing detection evaluation value, the clamping force is proportional to the chamfering process processing detection evaluation value to establish a logical rule, the fuzzy rule example is:

[0046] The surface roughness after chamfering is marked as C, the clamping force is marked as J, and the chamfering process inspection evaluation value is marked as G, then it can be defined as:

[0047] Rule 1:IF(C is High)AND(J is Low)THEN(G is Low)

[0048] Rule 2:IF(C is Low)AND(J is High)THEN(G is High) ...

[0049] According to the established fuzzy rules and the fuzzy sets corresponding to the surface roughness after chamfering and the clamping force of the fixture, fuzzy reasoning is performed to obtain the fuzzy set of the chamfering process processing detection evaluation value, and then the chamfering process processing detection evaluation value is obtained by defuzzifying the fuzzy set of the chamfering process processing detection evaluation value.

[0050] The division of fuzzy sets can be adjusted according to actual conditions. For example, although this embodiment takes three fuzzy sets as examples, in fact, the surface roughness after chamfering, the clamping force of the fixture, and the chamfering process inspection evaluation value can be divided into more than three sets to facilitate a better evaluation of sleep quality.

[0051] For the judgment of the high, medium and low surface roughness after chamfering and the high, medium and low clamping force of the fixture, the threshold value can be set according to the actual situation. For example, when the surface roughness after chamfering exceeds 2 microns, it will be marked as "High". I will not go into details here.

[0052] Surface roughness refers to the roughness of the workpiece surface after chamfering. It is used to evaluate the accuracy and forming effect of the chamfering process. It is measured by surface quality inspection equipment or instruments. A surface roughness instrument, such as a surface roughness meter, is used to scan and measure the chamfered workpiece to obtain the surface roughness value. The clamping force of the fixture refers to the fixing force or supporting force of the fixture on the workpiece during the chamfering forming process. The fixture is used to maintain the stability and position of the workpiece to ensure that the chamfering tool can cut accurately without causing errors due to movement or swinging of the workpiece. The clamping force of the fixture is specifically obtained through the fixture's own sensors or load sensors. These sensors can monitor the force or pressure applied by the fixture in real time, thereby determining the size of the fixture's clamping force.

[0053] By monitoring the process data such as the surface roughness and fixture clamping force after chamfering, it is possible to capture abnormal situations that may occur in the chamfering process in real time. With the help of fuzzy logic reasoning, multiple input variables can be comprehensively considered to generate specific chamfering process detection evaluation values, so as to detect abnormalities in time; set the detection threshold based on the chamfering process requirements. When the detection evaluation value is lower than the threshold, the system automatically determines that there is an abnormality in the current round of chamfering process. The automated warning mechanism can immediately trigger an alarm to remind the operator or the automation system to perform further maintenance and adjustments to avoid the production of defective products or potential equipment damage; timely identification and handling of process abnormalities helps to reduce the quality risk of chamfered workpieces. Through real-time monitoring and detection during the forming process, it can prevent the increase in defective product rate caused by process abnormalities, ensure the stability and consistency of product quality, avoid downtime and rework caused by process abnormalities, and help improve the operating efficiency of the production line and equipment utilization.

[0054] In summary, this application has at least the following effects:

[0055] Through the control and correction of the real-time forming process by the machine learning model, the accuracy and consistency of the chamfering process can be improved. The traditional chamfering method is affected by the operator's technical level and experience, while this method can be adjusted according to real-time data to ensure that the quality and geometry of each chamfer meet the requirements; using machine learning models for automatic control and adjustment can reduce the need for manual intervention, thereby improving production efficiency. Through real-time monitoring and adjustment, defective products caused by operating errors or equipment deviations can be avoided, thereby saving time and resources; monitoring the process processing status after chamfering forming, timely detecting abnormal conditions and performing maintenance, will help improve the stability and reliability of the production line, not only to correct problems in real-time production, but also to accumulate data for subsequent process optimization; by collecting and analyzing real-time forming data, the machine learning model can be continuously optimized so that it can more accurately predict and adjust the chamfering process, helping manufacturers to continuously improve product quality and production efficiency in the long run.

[0056] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0057] The present invention is described with reference to a flowchart of a method according to an embodiment of the present invention. It should be understood that the combination of each process in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the flowchart. Figure 1 A device that specifies functions in a process or multiple processes.

[0058] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A function specified in a process or multiple processes.

[0059] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1The steps of a specified function in a process or multiple processes.

[0060] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0061] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for chamfering a ball track of a four-row linear ball guide, characterized in that: The following steps are involved: S1, based on the double chamfering simulation process training of the four-row ball linear guideway, obtain the double chamfering training data; S2, based on the double chamfering training data, respectively train a first machine learning model for one round of chamfering trajectory, a second machine learning model for one round of chamfering bevel angle, a third machine learning model for two rounds of chamfering trajectory, and a fourth machine learning model for two rounds of chamfering bevel angle; S3, in the actual chamfering forming process, collect the real-time forming data of the double chamfer of the ball track of the four-row ball linear guide, and control and correct one round of chamfering forming process based on the real-time forming data, the first machine learning model and the second machine learning model; S4, monitoring the process status after a round of chamfering and forming on the basis of controlling and correcting the process of a round of chamfering and forming, and maintaining the abnormal process status after a round of chamfering and forming; S5, repeating S3 and S4 to control the second round of chamfering until the second round of chamfering is completed; Monitor the actual chamfering process in real time and obtain real-time forming data, including the cutting speed of a round of chamfering, the X-axis coordinates, Y-axis coordinates and chamfering angle of the chamfering cutting tool of the four-row ball linear guideway in a round of chamfering; Using the trained first machine learning model, according to the cutting speed of a round of chamfering at the current moment, and the X-axis coordinate and Y-axis coordinate of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway in a round of chamfering, the position of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway at the next time point is predicted; Then, using the trained second machine learning model, the chamfering bevel angle at the next time point is predicted based on the predicted position of the chamfering cutting tool of the four-row ball linear guideway at the next time point and the chamfering bevel angle at the current moment; The position of the chamfering cutting tool of the four-row ball linear guide auxiliary ballway predicted by the first machine learning model and the chamfering bevel angle predicted by the second machine learning model are respectively compared with the real-time forming data to obtain the position error value and the chamfering bevel angle error value of the four-row ball linear guide auxiliary ballway; Obtaining a four-row ball linear guide auxiliary ball track chamfering cutting tool position error threshold and a chamfering slope angle error threshold based on a four-row ball linear guide auxiliary ball track chamfering forming process setting; When the position error value and the chamfering bevel angle error value of the four-row ball linear guide auxiliary ballway chamfering cutting tool are respectively within the position error threshold and the chamfering bevel angle error threshold of the four-row ball linear guide auxiliary ballway chamfering cutting tool, the automatic control sensor parameters are adjusted and corrected in real time.

2. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The double chamfering training data specifically includes one round of chamfering training data and two rounds of chamfering training data; The one-round chamfering training data specifically includes one-round chamfering trajectory training data and one-round chamfering bevel angle training data, wherein the one-round chamfering trajectory training data specifically includes the X-axis coordinate and the Y-axis coordinate in the spatial coordinate system during the one-round chamfering training process, and the one-round chamfering bevel angle training data specifically includes the chamfering bevel angle during the one-round chamfering training process; The two-round chamfering training data specifically includes two-round chamfering trajectory training data and two-round chamfering bevel angle training data. The two-round chamfering trajectory training data specifically refers to the X-axis coordinate and the Y-axis coordinate in the spatial coordinate system during the two-round chamfering training process. The two-round chamfering bevel angle training data specifically refers to the chamfering bevel angle during the two-round chamfering training process.

3. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The specific training method of the first machine learning model of the round of chamfering trajectory is: Clean, process missing values ​​and standardize a round of chamfer trajectory training data; Based on the speed consistency requirement of chamfering forming, the training data of a round of chamfering trajectory is divided into four columns of X-axis coordinates and Y-axis coordinates corresponding to each time point of the ball linear guideway chamfering cutting tool; Based on the linear regression model as the basic model of the first machine learning model of a round of chamfering trajectory, the X-axis coordinates and Y-axis coordinates corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool are used as the input of the first machine learning model of a round of chamfering trajectory. The first machine learning model of a round of chamfering trajectory uses the X-axis coordinates, Y-axis coordinates corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool and the cutting position labels predicted by the cutting speed as prediction targets, and minimizes the sum of prediction errors of all cutting position labels as the training target; The first machine learning model of a round of chamfering trajectory is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The first machine learning model of the cutting position label predicted by the X-axis coordinates, Y-axis coordinates and cutting speed output corresponding to each time point of the four-row ball linear guide pair ballway chamfering cutting tool is trained.

4. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The specific training method of the third machine learning model of the two-round chamfering trajectory is: Clean, process missing values ​​and standardize the two-wheel chamfer trajectory training data; Based on the speed consistency requirement of chamfering forming, the two-wheel chamfering trajectory training data is divided into four columns of X-axis coordinates and Y-axis coordinates corresponding to each time point of the ball linear guideway chamfering cutting tool; Based on the linear regression model as the basic model of the third machine learning model of the two-wheel chamfering trajectory, the X-axis coordinate, Y-axis coordinate and cutting speed corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool are used as the input of the third machine learning model of the two-wheel chamfering trajectory. The third machine learning model of the two-wheel chamfering trajectory uses the X-axis coordinate, Y-axis coordinate and cutting speed corresponding to each time point of the four-row ball linear guide auxiliary ballway chamfering cutting tool as the prediction target, and minimizes the sum of the prediction errors of all cutting position labels as the training target; The third machine learning model of the two-round chamfering trajectory is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The third machine learning model is trained to output the predicted cutting position label according to the X-axis coordinates and Y-axis coordinates corresponding to each time point of the four-row ball linear guide pair ballway chamfering cutting tool.

5. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The specific training method of the second machine learning model for the round of chamfering angle is: Clean, process missing values ​​and standardize a round of chamfer angle training data; Based on the speed consistency requirement of chamfering forming, a round of chamfering bevel angle training data is divided into four columns of chamfering bevel angles corresponding to each time point of the ball linear guide secondary ballway chamfering cutting tool; This model, which is based on a linear regression model as the second machine learning model of a round of chamfer bevel angles, uses the chamfer bevel angles and cutting positions corresponding to each time point of the four-row ball linear guide auxiliary ballway as inputs to the second machine learning model of a round of chamfer bevel angles. The second machine learning model of a round of chamfer bevel angles uses the chamfer bevel angles corresponding to each time point of the four-row ball linear guide auxiliary ballway and the chamfer bevel angle labels predicted by the cutting positions as prediction targets, and minimizes the sum of all chamfer bevel angles as a training target. The second machine learning model of a round of chamfer bevel angle is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The second machine learning model of the chamfer bevel angle label predicted by the chamfer bevel angle corresponding to each time point of the four-row ball linear guide auxiliary ball track and the cutting position is trained.

6. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The specific training method of the fourth machine learning model of the second-wheel chamfer angle is: Clean, process missing values ​​and standardize the training data of the second round of chamfer angles; Based on the requirement of consistent chamfering forming speed, the training data of the two-wheel chamfering bevel angle is divided into four columns of chamfering bevel angles corresponding to each time point of the ball linear guide secondary ballway chamfering cutting tool; This model, which is based on the linear regression model as the fourth machine learning model of the two-wheel chamfer bevel angle, uses the chamfer bevel angle and cutting position corresponding to each time point of the four-row ball linear guide auxiliary ballway as the input of the fourth machine learning model of the two-wheel chamfer bevel angle. The fourth machine learning model of the two-wheel chamfer bevel angle uses the chamfer bevel angle corresponding to each time point of the four-row ball linear guide auxiliary ballway and the chamfer bevel angle label predicted by the cutting position as the prediction target, and minimizes the sum of all chamfer bevel angles as the training target; The fourth machine learning model of the two-wheel chamfer bevel angle is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The fourth machine learning model of the chamfer bevel angle label predicted according to the chamfer bevel angle corresponding to each time point of the four-row ball linear guide secondary ball track and the cutting position is trained.

7. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The warning for a round of chamfering forming process based on real-time forming data, the first machine learning model and the second machine learning model is: When the position error value and the chamfer bevel angle error value of the chamfer cutting tool of the four-row ball linear guide auxiliary ball track are respectively greater than the position error threshold and the chamfer bevel angle error threshold of the chamfer cutting tool of the four-row ball linear guide auxiliary ball track, an alarm is triggered to warn.

8. A method for chamfering a ball track of a four-row linear ball guide according to claim 1, characterized in that: The specific analysis of maintaining the abnormal processing conditions after a round of chamfering is as follows: The process data is obtained based on monitoring the process status after a round of chamfering; The process data specifically include: surface roughness after chamfering and clamping force of the fixture; Fuzzy logic is used to perform fuzzy reasoning on the surface roughness after chamfering and the clamping force of the fixture to obtain the chamfering process processing inspection evaluation value; Obtain the chamfering process detection threshold value set based on the chamfering forming process requirements of the four-row ball linear guideway; The chamfering process processing detection evaluation value is compared with the chamfering process processing detection threshold value. When the chamfering process processing detection evaluation value is greater than or equal to the chamfering process processing detection threshold value, the process processing of the current round of chamfering is normal; When the chamfering process processing detection evaluation value is less than the chamfering process processing detection threshold, the process processing of the current round of chamfering is abnormal, and an alarm is triggered for warning maintenance.

Images