Diamond cutter path error compensation method based on cutter wear evolution

Abstract

The invention discloses a diamond cutter path error compensation method based on cutter wear evolution, which comprises the following steps: S1, multi-modal sensor data acquisition and processing: in the machining process, key state information between a diamond cutter and a workpiece is acquired in real time through a multi-modal sensor; s2, tool wear modeling: key geometric parameters and evolution trend of tool wear are extracted from the collected and processed data, and tool wear modeling is carried out. And S3, artificial intelligence prediction of the abrasion loss: training a tool abrasion prediction model based on a deep learning network structure. And S4, tool path compensation is conducted, specifically, the predicted abrasion loss and the actual free-form surface model are combined, the tool path compensation amount and the compensated control point coordinates are calculated, a new machining path is generated, and dynamic compensation of tool cutting edge abrasion is achieved. And S5, system integration and closed-loop control: integrating the steps on processing equipment, and realizing real-time data interaction and synchronization through a unified data protocol.

Description

Technical Field

[0001] This invention belongs to the technical field of tool condition detection and precision and ultra-precision machining accuracy improvement, specifically involving a diamond tool path error compensation method based on tool wear evolution. Background Technology

[0002] In modern manufacturing technology, especially in the field of freeform surface optical component processing, the requirements for machining accuracy are increasing. However, due to the minute geometric errors and wear of the cutting tools themselves during production and use, especially the irregular and minute changes in the cutting edge profile, errors can directly occur on the machined surface.

[0003] Current tool wear detection methods generally rely on offline methods, which require interrupting production and lead to low efficiency. More importantly, existing toolpath planning systems use ideal tool models, ignoring the geometric errors of the tool itself during manufacturing and the wear accumulation during use, thus limiting the improvement of machining accuracy.

[0004] Existing cutting tools are prone to geometric errors and wear during production and use. In particular, minute changes in the cutting edge profile can lead to errors in the machined surface. Traditional inspection relies on offline methods, which are inefficient and cannot adapt to actual wear accumulation, thus limiting machining accuracy.

[0005] In recent years, sensor technology and artificial intelligence technology have provided new approaches for the detection and prediction of tool wear conditions. By establishing a closed-loop control scheme that includes tool wear condition detection, a big data model of tool wear evolution, and a dynamic compensation system for toolpath errors, the aforementioned problems can be effectively solved. Summary of the Invention

[0006] The purpose of this invention is to solve the above-mentioned problems and provide a diamond tool path error compensation method based on tool wear evolution that can dynamically predict path errors caused by tool wear and perform error compensation in the path planning stage to improve the accuracy and stability of ultra-precision machining.

[0007] To solve the above-mentioned technical problems, the technical solution of the present invention is: a diamond toolpath error compensation method based on tool wear evolution, comprising the following steps:

[0008] S1. Multimodal sensor data acquisition and processing: During the machining process, key state information between the diamond tool and the workpiece is acquired in real time through multimodal sensors; this information is processed, denoised, filtered and feature extracted to provide accurate input data for subsequent tool wear modeling and prediction.

[0009] S2. Tool Wear Modeling: Key geometric parameters of tool wear are extracted from the collected and processed data to perform tool wear modeling; tool wear forms are classified and wear stages are divided according to machining time; a three-dimensional geometric model of the tool cutting edge profile is constructed by combining high-order surface fitting and wear mechanics model, quantifying the deviation of the cutting edge radius and the changes in morphology at different stages, and forming the law of wear evolution; the key geometric parameters include: tool cutting edge radius, cutting edge profile deviation curve, local wear amount of the cutting edge, wear width of the flank face, size of the grinding pit on the rake face, and changes in surface roughness of the cutting edge, etc.

[0010] S3. Artificial intelligence predicts wear: Based on a deep learning network structure, a tool wear prediction model is trained using multimodal features and historical wear data as input. During the machining process, the collected machining data is input in real time to predict the cutting edge radius deviation and wear within a set time in the future, and to provide data for tool path compensation.

[0011] S4. Toolpath Compensation: Combining the predicted wear amount with the actual freeform surface model, the toolpath compensation amount and the coordinates of the control points after compensation are calculated to generate a new machining path and realize dynamic compensation for tool edge wear; at the same time, the paths and errors before and after compensation are visualized and the data is exported.

[0012] S5. System Integration and Closed-Loop Control: Integrate the above steps on the machining equipment, realize real-time data interaction and synchronization through a unified data protocol; based on the prediction results and compensation calculations, dynamically update the CNC machining program, realize online correction of the machining path, and form a closed-loop control system for tool status detection, prediction and compensation.

[0013] Furthermore, step S1 also includes the following sub-steps:

[0014] S11, Sensor Integration: All sensors are installed in key areas around the tool spindle, workpiece fixture, or machining platform;

[0015] S12. Data Acquisition and Real-time Transmission: Various sensors synchronously acquire status information during the processing through the data acquisition system and transmit it to the central monitoring system in real time via wired or wireless means for online analysis and processing.

[0016] S13. Data Processing and Storage: Process the raw data from the multimodal sensor, including noise reduction, filtering, normalization, and feature extraction; update the above data information to the tool wear modeling and model prediction algorithm dataset to improve the accuracy of wear prediction.

[0017] Furthermore, step S2 also includes the following sub-steps:

[0018] S21. Classification and Stage Division of Tool Wear: Further analysis of the processed sensor data is performed to classify tool wear, including but not limited to: flank wear, rake face wear, and cutting edge wear.

[0019] Based on the machining time, the tool wear process is divided into three wear stages: initial wear stage, intermediate wear stage, and late wear stage, which facilitates the establishment of a stage-based tool wear model in the later stage.

[0020] S22. Tool geometry and wear evolution modeling: Combining high-order surface fitting methods and mechanical wear models, the tool edge profile and key parts at different wear stages are discretized geometrically represented, and then the tool edge curve is fitted to form a curve sequence corresponding to time t. This enables a comprehensive description and prediction of different wear forms and multiple wear stages, and further derives the evolution law of tool edge wear.

[0021] Furthermore, step S3 also includes the following sub-steps:

[0022] S31. Deep Learning Predictive Modeling: Based on the establishment of a tool wear model and the acquisition of multimodal feature values, a tool wear prediction model is constructed. This model uses deep learning algorithms to predict tool wear and can analyze the changing trend of tool condition during machining. During model training, historical machining data and tool wear measurement results are used as input data for model learning and optimization to improve prediction accuracy. The model training can be adjusted using common optimization algorithms to improve prediction performance.

[0023] S32. Online Prediction: During actual machining, the multimodal feature data collected in real time is input into the trained prediction model to obtain the predicted value of future tool wear, and the prediction result is directly transmitted to the path compensation step for online correction; at the same time, the predicted wear index is used as an auxiliary basis to determine whether the tool needs to be replaced, thereby realizing the proactive management and maintenance of tool status.

[0024] Furthermore, step S4 also includes the following sub-steps:

[0025] S41. Model the machining tool and freeform surface: Based on the tool's geometric parameters and the function of the freeform surface to be machined, and combined with the numerical solution method of nonlinear equations, solve for the coordinates of the ideal contact point and ideal control point of the tool on the freeform surface.

[0026] S42. Calculate the included angle of the contact point on the ideal cutting edge curve: Based on the coordinates of the ideal contact point on the freeform surface, calculate the radian value corresponding to each contact point on the ideal cutting edge curve to provide correct angle data for compensation calculation and ensure that the direction of contact between the tool and the workpiece meets the processing requirements.

[0027] S43. Calculation of actual tool cutting edge curve error: After obtaining the tool contact angle, the actual error of the tool cutting edge curve is calculated by interpolation or fitting methods to provide error compensation values ​​for subsequent compensation calculations.

[0028] S44. Compensate the actual tool cutting edge curve model: Solve the radial unit vector at each contact point in the tool coordinate system, and further solve the three-dimensional compensation vector in the workpiece coordinate system in combination with the error value. Compensate the actual wear amount corresponding to the ideal control point and generate the coordinates of the compensated control point.

[0029] S45. Data Export and Visualization: Visualize the tool control point trajectory, control point height comparison, and compensation amount before and after compensation. Export the tool control point coordinates and compensation amount before and after compensation as text, which can be used for data analysis and subsequent use.

[0030] Furthermore, the generation of the compensated control point coordinates in S44 specifically involves: in the tool coordinate system, the direction from the contact point of the tool edge curve to the tool center is the tool radial direction, the unit vector of this direction is set as n, and the compensation error in this direction is e. sd The compensation amount in the tool coordinate system is ΔP c The rotation matrix is ​​R, the actual compensation vector is ΔP, and the ideal control point coordinates are P. I The actual control point coordinates are P. T The specific derivation formula is as follows:

[0031]

[0032] Furthermore, step S5 includes the following sub-steps:

[0033] S51. Data Interaction and Synchronization: Through industrial bus or real-time data interface, the functional modules are efficiently connected, the data protocol specifications are unified, and the stable transmission and time synchronization of data between the modules are ensured.

[0034] S52. Closed-loop control execution and dynamic path adjustment: Based on real-time wear prediction and geometric error analysis results, the tool path in the CNC program is automatically corrected to achieve online dynamic updating of the path and ensure machining accuracy and stability.

[0035] The beneficial effects of this invention are:

[0036] 1. The diamond tool path error compensation method based on tool wear evolution provided by the present invention collects tool status information in real time through multi-modal sensors, combines high-order surface fitting and wear mechanics model to model the geometric characteristics of tool wear, and further introduces deep learning prediction model to predict tool wear trend in advance and perform machining path compensation, so as to realize the integration of prediction and compensation.

[0037] 2. The diamond tool path error compensation method based on tool wear evolution proposed in this invention can effectively overcome the shortcomings of existing technologies that rely on offline measurement and cannot compensate in real time, significantly reduce machining deviations caused by tool geometric errors and wear, and improve the stability and surface accuracy of freeform surface ultra-precision machining. Attached Figure Description

[0038] Figure 1 This is a flowchart of the tool wear prediction and path compensation method of the present invention;

[0039] Figure 2 This is a schematic diagram of the cutting edge curve fitting of the tool of the present invention;

[0040] Figure 3 This is a trajectory diagram showing the changes in the compensation amount between the ideal control point and the compensated control point in this invention;

[0041] Figure 4 This is a three-dimensional distribution diagram of the ideal control points and the compensated control points of this invention. Detailed Implementation

[0042] The present invention will be further described below with reference to the accompanying drawings and specific embodiments:

[0043] like Figures 1 to 4 As shown, the present invention provides a diamond toolpath error compensation method based on tool wear evolution, comprising the following steps:

[0044] S1. Multimodal Sensor Data Acquisition and Processing: During the machining process, multimodal sensors are used to acquire key state information between the diamond tool and the workpiece in real time. This information undergoes data processing, noise reduction, filtering, and feature extraction to provide accurate input data for subsequent tool wear modeling and prediction.

[0045] Step S1 also includes the following sub-steps:

[0046] S11, Sensor Integration: All sensors are installed in critical areas around the tool spindle, workpiece fixture, or machining platform.

[0047] S12. Data Acquisition and Real-time Transmission: Various sensors synchronously acquire status information during the processing through the data acquisition system, and transmit it to the central monitoring system in real time via wired or wireless means for online analysis and processing.

[0048] S13. Data Processing and Storage: Raw data from multimodal sensors is processed, including noise reduction, filtering, normalization, and feature extraction. This data is then updated in the tool wear modeling and prediction algorithm dataset to improve the accuracy of wear prediction.

[0049] In this embodiment, tool status information during the machining process is first acquired using multimodal sensors. The key status information acquired by the multimodal sensors includes: cutting force, vibration signal, current signal, acoustic emission signal, and temperature. The sensors used include: force sensors, vibration sensors, current sensors, acoustic emission sensors, and temperature sensors.

[0050] Noise reduction and filtering are performed on different types of sensor signals to suppress environmental interference and machine tool noise. Specifically, for acoustic emission signals, bandpass filters can be used to remove low-frequency background noise and high-frequency interference. For vibration and force signals, low-pass filtering and adaptive filtering algorithms can be used to suppress mechanical noise. For non-stationary signals, wavelet packet decomposition, empirical mode decomposition, and other methods can be used to extract the main effective components.

[0051] Subsequently, a unified sampling frequency and time reference were established to achieve time alignment of the data. The processed multi-source signals were then normalized, mapping them to the [0,1] or [-1,1] interval to eliminate differences in dimensions and amplitudes between different physical signals, laying the foundation for multi-signal fusion and comparison. The calculation formula is as follows:

[0052]

[0053] Where X is the sensor's raw reading, X0 min The minimum value of the sensor data over a period of time, X max X is the maximum value of the sensor data over a period of time. n This is the result of normalizing the original data.

[0054] Finally, feature values ​​are extracted from the processed signal and standardized. In the time domain analysis, the root mean square value, peak value, peak-to-peak value, skewness, and kurtosis can be obtained. In the frequency domain analysis, the power spectral density, spectral centroid, bandwidth, and dominant frequency peak value can be obtained. In the time-frequency domain analysis, data analysis is performed by combining time and frequency variations.

[0055] The extracted multi-dimensional features are standardized, transforming them into a dataset with a mean of 0 and a standard deviation of 1. This satisfies the assumptions of machine learning algorithms regarding the distribution of input data, thereby accelerating model convergence and improving prediction accuracy. The calculation formula is as follows:

[0056]

[0057] Where x is the original eigenvalue, μ is the mean, σ is the standard deviation, and X is the mean value. z This is the standardized deviation value.

[0058] S2. Tool Wear Modeling: Key geometric parameters of tool wear are extracted from the acquired and processed data to perform tool wear modeling. Tool wear forms are classified and wear stages are divided according to machining time. Combining high-order surface fitting and wear mechanics models, a three-dimensional geometric model of the tool cutting edge profile is constructed to quantify the deviation of the cutting edge radius and the changes in morphology at different stages, thus forming the law of wear evolution. The key geometric parameters include: tool cutting edge radius, cutting edge profile deviation curve, local wear amount of the cutting edge, flank wear width, rake face grinding pit size, and changes in cutting edge surface roughness, etc.

[0059] The wear evolution law is as follows:

[0060] Initial wear stage: Wear increases rapidly over time, then tends to slow down; Intermediate wear stage: Wear is roughly linearly related to time, and the wear rate is stable; Late wear stage: Wear accelerates.

[0061] Step S2 also includes the following sub-steps:

[0062] S21. Classification and Stages of Tool Wear: Further analysis of the processed sensor data is conducted to classify tool wear, including but not limited to: flank wear, rake face wear, and cutting edge wear.

[0063] Based on the machining time, the tool wear process is divided into three stages: initial wear stage, intermediate wear stage, and late wear stage, which facilitates the establishment of a stage-based tool wear model in the later stage.

[0064] S22. Tool Geometry and Wear Evolution Modeling: Combining high-order surface fitting methods and mechanical wear models, the tool edge profile and key components at different wear stages are discretized geometrically. Then, the tool edge curve is fitted to form a curve sequence corresponding to time t, achieving a comprehensive description and prediction of different wear forms and multiple wear stages, and further deriving the evolution law of tool edge wear. For any time t, the corresponding edge profile curve is:

[0065] r(θ,t)=a0(t)+a1(t)θ+a2(t)θ 2 +...+a 10 (t)θ 10

[0066] Where the coefficients are a0(t), a1(t), ..., a 10 (t) is a coefficient that changes with time.

[0067] In this embodiment, a geometric measurement method is used to obtain discrete contour point sets of the tool cutting edge at different machining times t. Each point set contains a large number of (x, y) coordinate points. Based on the collected data, the wear of the tool at each stage of the machining process is analyzed. Specifically, tool wear is typically divided into three stages: initial wear stage, intermediate wear stage, and late wear stage. In each stage, the shape and radius of the tool cutting edge will change to varying degrees; therefore, different geometric models are established according to different wear stages.

[0068] like Figure 2 As shown, at time t0, the collected point set is preprocessed by least squares fitting, and the collected tool edge profile is modeled to quantify the radius deviation and shape change caused by wear. Figure 2 The fitted cutting edge curve expression is shown below:

[0069] r(θ,t0)=49.93-1.06θ+5.15θ 2 -20.95θ 3 -148.55θ 4 +812.70θ 5 +1372.37θ 6

[0070] -5812.76θ 7 -4790.58θ 8 +11758.40θ 9 +5557.29θ 10

[0071] S3. Artificial Intelligence-Based Wear Prediction: A deep learning-based network structure is used to train a tool wear prediction model, taking multimodal features and historical wear data as input. During machining, real-time processing data is input to predict the cutting edge radius deviation and wear amount within a set future timeframe, providing data for toolpath compensation.

[0072] Step S3 also includes the following sub-steps:

[0073] S31. Deep Learning Predictive Modeling: Based on the established tool wear model and the acquisition of multimodal feature values, a tool wear prediction model is constructed. This model uses deep learning algorithms to predict tool wear and can analyze the changing trends of tool condition during machining. During model training, historical machining data and tool wear measurement results are used as input data for model learning and optimization to improve prediction accuracy. The model training can be adjusted using common optimization algorithms to improve prediction performance.

[0074] S32. Online Prediction: During actual machining, real-time collected multimodal feature data is input into the trained prediction model to obtain the predicted value of future tool wear. The prediction result is then directly transmitted to the path compensation step for online correction. Simultaneously, the predicted wear index is used as an auxiliary basis to determine whether the tool needs to be replaced, thereby achieving proactive management and maintenance of the tool's condition.

[0075] In this embodiment, an LSTM network is selected for model training, using multimodal features and historical wear data as input to generate a tool wear prediction model. To measure the difference between the predicted value and the actual tool wear, mean squared error (MSE) can be selected as the loss function, and the Adam optimizer can be chosen as the optimization algorithm. The prediction results can be directly passed to the toolpath compensation step, thereby dynamically adjusting the toolpath during machining to ensure machining accuracy.

[0076] The state update formula for LSTM is as follows:

[0077]

[0078] Where, x t h is the input feature vector at time t; t-1 W is the hidden state vector from the previous time step. f W is the weight matrix of the forget gate. i W is the weight matrix of the input gate. o W is the weight matrix of the output gate. c b is the weight matrix for the candidate states; f b i b o b c All are biases; σ(·) is the activation function; f t For the output of the forget gate, i t For input gate output, o t For output gate output, These are candidate memory units.

[0079] S4. Toolpath Compensation: Combining the predicted wear amount with the actual freeform surface model, the toolpath compensation amount and the coordinates of the control points after compensation are calculated to generate a new machining path and realize dynamic compensation for tool edge wear; at the same time, the paths and errors before and after compensation are visualized and the data is exported.

[0080] Step S4 also includes the following sub-steps:

[0081] S41. Model the machining tool and freeform surface: Based on the tool's geometric parameters and the function of the freeform surface to be machined, and combined with the numerical solution method of nonlinear equations, solve for the coordinates of the ideal contact point and ideal control point of the tool on the freeform surface.

[0082] S42. Calculate the included angle of the contact point on the ideal cutting edge curve: Based on the coordinates of the ideal contact point on the freeform surface, calculate the radian value corresponding to each contact point on the ideal cutting edge curve to provide correct angle data for compensation calculation and ensure that the direction of contact between the tool and the workpiece meets the processing requirements.

[0083] Furthermore, in the tool coordinate system, let the included angle of the cutting edge be γ, and the parameter θ∈[-γ / 2, γ / 2]. In the xz plane, uniformly discretize the tool cutting edge curve into N contact points. Then, the coordinates of each discrete point in the local coordinate system are:

[0084]

[0085] Let R be the rotation matrix of the tool in the workpiece coordinate system, and T be the position vector of the origin of the tool coordinate system in the workpiece coordinate system. Then the i-th local point P c，i =[X c，i ,0,Z c，i ] T The expression mapped to the workpiece coordinate system is:

[0086] P w，i =RP c，i +T

[0087] Let P be the position of the expected contact point in the workpiece coordinate system. d Then each discrete point P w,i The Euclidean distance to the ideal point is: d i =||P w,i -P d The contact point corresponding to the smallest di is the actual contact point, and the corresponding angle is the actual contact angle, in radians.

[0088] S43. Calculation of actual tool cutting edge curve error: After obtaining the tool contact angle, the actual error of the tool cutting edge curve is calculated by interpolation or fitting methods to provide error compensation values ​​for subsequent compensation calculations.

[0089] Furthermore, the cubic spline interpolation calculation process used in this embodiment is as follows:

[0090] Record the tool edge wear in 1° increments and store it as a two-column text file:

[0091] {si, ei}, si∈[-60°, 60°], ei∈[-0.0003, 0.0003]

[0092] Where si represents any edge contact angle in degrees; ei represents the corresponding radial wear in mm. The error is estimated at the actual contact angle sd by treating the original discrete data as a curve and using a one-dimensional cubic spline interpolation function to calculate the error e. sd And limit sd within the measured angle range: s min ≤sd≤s max If the value exceeds the boundary value, then take the boundary value.

[0093] S44. Compensate the actual tool cutting edge curve model: Solve the radial unit vector at each contact point in the tool coordinate system, and further solve the three-dimensional compensation vector in the workpiece coordinate system based on the error value. Compensate the actual wear amount corresponding to the ideal control point and generate the coordinates of the compensated control point.

[0094] In step S44, generating the compensated control point coordinates specifically involves: In the tool coordinate system, the direction from the contact point of the tool edge curve to the tool center is the tool radial direction. The unit vector of this direction is set as n, and the compensation error in this direction is e. sd The compensation amount in the tool coordinate system is ΔP c The rotation matrix is ​​R, the actual compensation vector is ΔP, and the ideal control point coordinates are P. I The actual control point coordinates are P. T The specific derivation formula is as follows:

[0095]

[0096] S45. Data Export and Visualization: Visualize the tool control point trajectory, control point height comparison, and compensation amount before and after compensation. Export the tool control point coordinates and compensation amount before and after compensation as text, which can be used for data analysis and subsequent use.

[0097] S5. System Integration and Closed-Loop Control: Integrate the above steps on the machining equipment, realize real-time data interaction and synchronization through a unified data protocol; based on the prediction results and compensation calculations, dynamically update the CNC machining program, realize online correction of the machining path, and form a closed-loop control system for tool status detection, prediction and compensation.

[0098] Step S5 includes the following sub-steps:

[0099] S51. Data Interaction and Synchronization: Through industrial bus or real-time data interface, the various functional modules are efficiently connected, and the data protocol specifications are unified to ensure stable data transmission and time synchronization between the modules.

[0100] S52. Closed-loop control execution and dynamic path adjustment: Based on real-time wear prediction and geometric error analysis results, the tool path in the CNC program is automatically corrected to achieve online dynamic updating of the path and ensure machining accuracy and stability.

[0101] This embodiment targets a spherical optical element with a diameter D = 50 mm, whose target surface shape is described by the freeform surface equation z = f(x,y). The machining tool is a single-point diamond tool with a nose radius r = 0.3 mm, a rake angle α = -30°, and a clearance angle β = 10°. To simulate the minute wear conditions of the tool in actual use, a cutting edge profile error curve with an amplitude of approximately ±0.3 μm was pre-established. This curve was obtained by fitting offline measurement data, with an angular resolution of 1°, and stored as a text file representing the correspondence between contact angle and radius error.

[0102] The implementation steps are as follows:

[0103] 1. Freeform surface and tool geometry modeling.

[0104] Based on the equation of the freeform surface z = f(x,y) and its first-order partial derivatives Establish a workpiece surface model. Combine this with the tool geometry parameter r. t A 3D model of the cutting tool is constructed using α,β, including the cutting edge curve, the rake face, and the flank face. The cutting edge curve fitting is as follows: Figure 2 As shown.

[0105] 2. Calculation of ideal contact point and control point.

[0106] The data of the helical trajectory with equal arc length [ρ] i θ i Convert to Cartesian coordinates, projected (x) c ,y c ,z c The ideal contact point is defined as ). Using a nonlinear equation solver, with the tangency between the tool and the freeform surface as a constraint, the coordinates (x, y) of the ideal control point are calculated. d ,y d ,z d ).

[0107] 3. Determine the contact angle of the cutting edge.

[0108] For each contact point, calculate the freeform surface normal vector, using it as the z-axis of the tool's local coordinate system. Combined with the cutting feed direction, determine the x and y axes, forming a rotation matrix R. Map the contact points to the tool coordinate system to obtain the cutting edge contact position parameters s (in radians), and convert them to angles.

[0109] 4. Edge error mapping and compensation vector calculation.

[0110] According to the angle The cutting edge radius error e is calculated by interpolation. sdIn the tool coordinate system, a radial unit vector n is taken and transformed to the workpiece coordinate system using a rotation matrix R to obtain the global radial direction. Multiplying this direction by the given vector yields the three-dimensional compensation vector ΔP.

[0111] 5. Control point coordinate correction.

[0112] The three-dimensional compensation vector ΔP is applied to the coordinates of the ideal control point to obtain the coordinates of the compensated control point. The compensation amount in the z-axis is recorded for subsequent accuracy analysis.

[0113] 6. Results visualization and data export.

[0114] Export the tool control point coordinates, radial compensation amount, and Z-axis compensation components before and after compensation to a text file, and plot the tool edge curve fitting diagram (e.g., Figure 2 The curves showing the changes in compensation amounts between the ideal control point and the compensated control point (e.g.) Figure 3 ) and a comparison image of three-dimensional trajectories (such as Figure 4 Visualization results such as )

[0115] Experiments show that the trajectory before compensation exhibits a deviation in the normal direction of the freeform surface at the μm level. After compensation, the deviation is significantly reduced, with the radial compensation ranging from -0.28 μm to +0.29 μm and the maximum Z-axis compensation component being approximately 0.25 μm. Three-dimensional trajectory comparison results show that the compatibility between the compensated trajectory and the target freeform surface is significantly improved, and the surface shape error is significantly reduced. This embodiment verifies the dynamic correction capability of the method of the present invention when the tool has minor wear, making it suitable for ultra-precision machining scenarios.

[0116] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A diamond toolpath error compensation method based on tool wear evolution, characterized in that, Includes the following steps: S1. Multimodal sensor data acquisition and processing: During the machining process, key state information between the diamond tool and the workpiece is acquired in real time through multimodal sensors; this information is processed, denoised, filtered and feature extracted to provide accurate input data for subsequent tool wear modeling and prediction. S2. Tool Wear Modeling: Extract key geometric parameters of tool wear from the acquired and processed data, and perform tool wear modeling. Tool wear patterns are classified and wear stages are divided according to machining time. A three-dimensional geometric model of the tool cutting edge profile is constructed by combining high-order surface fitting and wear mechanics model to quantify the cutting edge radius deviation and morphological changes at different stages, thus forming the law of wear evolution. The key geometric parameters include: tool cutting edge radius, cutting edge profile deviation curve, local wear amount of the cutting edge, flank wear width, rake face grinding pit size, and cutting edge surface roughness change. S3. Artificial intelligence predicts wear: Based on a deep learning network structure, a tool wear prediction model is trained using multimodal features and historical wear data as input. During the machining process, the collected machining data is input in real time to predict the cutting edge radius deviation and wear within a set time in the future, and to provide data for tool path compensation. S4. Toolpath Compensation: Combining the predicted wear amount with the actual freeform surface model, the toolpath compensation amount and the coordinates of the control points after compensation are calculated to generate a new machining path and realize dynamic compensation for tool edge wear; at the same time, the paths and errors before and after compensation are visualized and the data is exported. S5. System Integration and Closed-Loop Control: Integrate the above steps on the machining equipment, realize real-time data interaction and synchronization through a unified data protocol; based on the prediction results and compensation calculations, dynamically update the CNC machining program, realize online correction of the machining path, and form a closed-loop control system for tool status detection, prediction and compensation.

2. The diamond toolpath error compensation method based on tool wear evolution according to claim 1, characterized in that, S1 further includes the following sub-steps: S11, Sensor Integration: All sensors are installed in key areas around the tool spindle, workpiece fixture, or machining platform; S12. Data Acquisition and Real-time Transmission: Various sensors synchronously acquire status information during the processing through the data acquisition system and transmit it to the central monitoring system in real time via wired or wireless means for online analysis and processing. S13. Data Processing and Storage: Process the raw data from the multimodal sensor, including noise reduction, filtering, normalization, and feature extraction; update the above data information to the tool wear modeling and model prediction algorithm dataset to improve the accuracy of wear prediction.

3. The diamond toolpath error compensation method based on tool wear evolution according to claim 1, characterized in that, S2 further includes the following sub-steps: S21. Classification and Stage Division of Tool Wear: Further analysis of the processed sensor data is performed to classify tool wear, including but not limited to: flank wear, rake face wear, and cutting edge wear. Based on the machining time, the tool wear process is divided into three wear stages: initial wear stage, intermediate wear stage, and late wear stage, which facilitates the establishment of a stage-based tool wear model in the later stage. S22. Tool geometry and wear evolution modeling: Combining high-order surface fitting methods and mechanical wear models, the tool edge profile and key parts at different wear stages are discretized geometrically represented, and then the tool edge curve is fitted to form a curve sequence corresponding to time t. This enables a comprehensive description and prediction of different wear forms and multiple wear stages, and further derives the evolution law of tool edge wear.

4. The diamond toolpath error compensation method based on tool wear evolution according to claim 1, characterized in that, S3 further includes the following sub-steps: S31. Deep Learning Predictive Modeling: Based on the establishment of a tool wear model and the acquisition of multimodal feature values, a tool wear prediction model is constructed. This model uses deep learning algorithms to predict tool wear and can analyze the changing trend of tool condition during machining. During model training, historical machining data and tool wear measurement results are used as input data for model learning and optimization to improve prediction accuracy. The model training can be adjusted using common optimization algorithms to improve prediction performance. S32. Online Prediction: During actual machining, the multimodal feature data collected in real time is input into the trained prediction model to obtain the predicted value of future tool wear, and the prediction result is directly transmitted to the path compensation step for online correction; at the same time, the predicted wear index is used as an auxiliary basis to determine whether the tool needs to be replaced, thereby realizing the proactive management and maintenance of tool status.

5. The diamond toolpath error compensation method based on tool wear evolution according to claim 1, characterized in that, S4 further includes the following sub-steps: S41. Model the machining tool and freeform surface: Based on the tool's geometric parameters and the function of the freeform surface to be machined, and combined with the numerical solution method of nonlinear equations, solve for the coordinates of the ideal contact point and ideal control point of the tool on the freeform surface. S42. Calculate the included angle of the contact point on the ideal cutting edge curve: Based on the coordinates of the ideal contact point on the freeform surface, calculate the radian value corresponding to each contact point on the ideal cutting edge curve to provide correct angle data for compensation calculation and ensure that the direction of contact between the tool and the workpiece meets the processing requirements. S43. Calculation of actual tool cutting edge curve error: After obtaining the tool contact angle, the actual error of the tool cutting edge curve is calculated by interpolation or fitting methods to provide error compensation values ​​for subsequent compensation calculations. S44. Compensate the actual tool cutting edge curve model: Solve the radial unit vector at each contact point in the tool coordinate system, and further solve the three-dimensional compensation vector in the workpiece coordinate system in combination with the error value. Compensate the actual wear amount corresponding to the ideal control point and generate the coordinates of the compensated control point. S45. Data Export and Visualization: Visualize the tool control point trajectory, control point height comparison, and compensation amount before and after compensation. Export the tool control point coordinates and compensation amount before and after compensation as text, which can be used for data analysis and subsequent use.

6. The diamond toolpath error compensation method based on tool wear evolution according to claim 1, characterized in that, Specifically, in S44, the coordinates of the compensated control points are generated as follows: In the tool coordinate system, the direction from the contact point of the tool edge curve to the tool center is the radial direction of the tool. The unit vector of this direction is set as n, and the compensation error in this direction is e. sd The compensation amount in the tool coordinate system is ΔP c The rotation matrix is ​​R, the actual compensation vector is ΔP, and the ideal control point coordinates are P. I The actual control point coordinates are P. T The specific derivation formula is as follows:

7. The diamond toolpath error compensation method based on tool wear evolution according to claim 1, characterized in that, S5 includes the following steps: S51. Data Interaction and Synchronization: Through industrial bus or real-time data interface, the functional modules are efficiently connected, the data protocol specifications are unified, and the stable transmission and time synchronization of data between the modules are ensured. S52. Closed-loop control execution and dynamic path adjustment: Based on real-time wear prediction and geometric error analysis results, the tool path in the CNC program is automatically corrected to achieve online dynamic updating of the path and ensure machining accuracy and stability.

Images