Soft material accurate cutting machining method and system

Abstract

According to the precise cutting machining method and system for the soft material, firstly, the three-dimensional model and the cutting parameters of the material are obtained, the initial cutting track is generated, in the cutting process, material deformation is monitored in real time through the multi-sensor fusion technology, the cutting track and depth are adjusted through the dynamic compensation algorithm, and the cutting precision is improved. And meanwhile, a laser interference measurement system is adopted to track the position of a cutting tool, and micron-level precision is achieved through closed-loop feedback control. According to the method, pressure control and cutting depth adjustment are combined, a specific microstructure is created to achieve differential friction coefficients, in addition, the surface chemical composition is monitored through real-time spectrum analysis, and it is ensured that the preset surface characteristic requirement is met. The cutting track, depth and surface texture of the soft material can be accurately controlled, high-precision and customizable soft material machining is achieved, and key technical support is provided for intelligent machining in the fields of flexible electronics, bionic robots and the like.

Description

Technical Field

[0001] The present invention relates to the field of intelligent processing technology, and in particular to a method and system for precise cutting of soft materials. Background Art

[0002] Profiling soft materials holds significant potential in fields such as robotics, medical devices, and human-machine interaction. This technology, capable of imparting specific shapes and functional properties to soft materials, is crucial for improving product performance and expanding its application. However, existing soft material cutting methods often struggle to achieve complex shapes while maintaining precision, and their ability to manipulate the material's surface properties is limited.

[0003] Precisely controlling the cutting path and depth while avoiding material deformation or damage during the cutting of special-shaped soft materials has been a long-standing technical challenge. Traditional cutting methods struggle to meet the demands of refined processing, particularly when manufacturing soft parts with unique surface textures and contact characteristics. Furthermore, the inherent softness and deformability of soft materials pose additional challenges to positioning and securing the parts during the cutting process.

[0004] The surface contact properties of soft materials are crucial to their functionality, especially in applications requiring precise control of friction. However, achieving controllable adjustment of the surface friction coefficient of soft materials through cutting processes, as well as creating regions with differentiated contact properties on the same component, remains a pressing technical challenge. This requires not only innovative cutting techniques but also the integrated application of knowledge from multiple disciplines, including materials science and surface engineering.

[0005] Therefore, how to develop a special-shaped cutting method that can accurately control the surface texture and contact characteristics of soft materials, achieve differentiated regulation of the friction coefficient of different areas of soft parts, and at the same time ensure cutting accuracy and efficiency has become a key issue in promoting the advancement of soft material processing technology. Summary of the Invention

[0006] In order to solve the technical problems raised in the above background technology, the first aspect of the present invention provides a method for precise cutting of soft materials, the method comprising:

[0007] S1, obtains the 3D model data and target cutting parameters of the soft material, including cutting trajectory, depth and surface texture information;

[0008] S2, based on the 3D model data and target cutting parameters, an adaptive cutting path planning algorithm is used to generate the initial cutting trajectory;

[0009] S3 uses multi-sensor fusion technology to monitor material deformation during cutting in real time and obtain material surface morphology and internal stress distribution data;

[0010] S4, based on the monitored material deformation data, uses dynamic compensation algorithm to adjust the cutting trajectory and depth in real time to ensure cutting accuracy;

[0011] S5, uses a high-precision laser interferometry measurement system to continuously track the cutting tool position and obtain the deviation value between the actual cutting trajectory and the target trajectory;

[0012] S6, based on the deviation value, uses a closed-loop feedback control algorithm to perform micron-level position correction on the cutting tool to ensure accurate execution of the cutting trajectory;

[0013] S7, through the programmable pressure control system to adjust the contact pressure between the cutting tool and the material, to achieve precise processing of surface textures in different areas;

[0014] S8, for areas requiring differentiated friction coefficients, uses a multi-scale surface structure generation algorithm to create specific microstructures by combining pressure control and cutting depth adjustment;

[0015] S9 uses real-time spectral analysis technology to monitor the changes in the chemical composition of the cutting surface to determine whether the preset surface property requirements are met. If not, it returns to adjust the cutting parameters and re-process the area.

[0016] Optionally, the step S1, obtaining the three-dimensional model data and target cutting parameters of the soft material, including cutting trajectory, depth and surface texture information, includes:

[0017] Step S11, obtaining three-dimensional model data of the soft material, and obtaining complete model data through a three-dimensional scanner;

[0018] Step S12, extracting target parameters from the complete model data, including geometric dimensions, material properties, and surface texture;

[0019] Step S13, using a finite element analysis tool to mesh the model and determine initial values of the cutting trajectory, cutting depth, and surface texture;

[0020] Step S14, generating a trajectory path for the initial value, calculating the deviation between the trajectory path and the target parameter, and if the deviation exceeds a first deviation threshold, performing iterative adjustment using a gradient descent method to obtain an optimized trajectory path;

[0021] Step S15, calculating the depth value according to the optimized trajectory path, and using a cubic spline interpolation algorithm to determine the smoothed depth value;

[0022] Step S16, extracting texture features by smoothing the depth value, calculating the difference between the texture features and the surface texture, and if the difference exceeds a difference threshold, adjusting the depth value until a matching texture feature is obtained;

[0023] Step S17, after obtaining the matching texture features, the optimized trajectory path, the smoothed depth value and the matching texture features are integrated to obtain the final cutting parameters;

[0024] Step S18: Generate cutting data of the three-dimensional model from the final cutting parameters, perform simulation verification using MATLAB, and determine the integrity of the cutting data.

[0025] Optionally, the step S3, using multi-sensor fusion technology to monitor the material deformation during the cutting process in real time and obtain the material surface morphology and internal stress distribution data, includes:

[0026] Step S31, collecting raw signals during the cutting process through a multi-sensor system to obtain deformation data and stress data;

[0027] Step S32: integrating the collected original signals using weighted average fusion to determine surface morphological features and internal stress features;

[0028] Step S33, if the surface morphological feature exceeds the deformation threshold, the deformation trend is determined by the least squares method;

[0029] Step S34, adjusting the acquisition technology according to the deformation trend to obtain more accurate distribution information;

[0030] Step S35, optimizing the distribution information through Kalman filtering to obtain smooth stress distribution data;

[0031] Step S36: if the smoothed stress distribution data is inconsistent with the internal information, the deviation area is determined by the difference method;

[0032] Step S37: Use the deviation area data to update the weighted average fusion parameters to obtain more accurate real-time acquisition results.

[0033] Optionally, the step S34, adjusting the acquisition technology according to the deformation trend to obtain more accurate distribution information, includes:

[0034] Step S341, obtaining initial distribution data through dynamic sampling, where the sampling interval is determined according to a preset spatial grid density;

[0035] Step S342: input the initial distribution data into a deformation trend calculation module based on moving average, and output the displacement change rate of each grid point;

[0036] Step S343, adjusting the sampling frequency according to the displacement change rate, doubling the sampling frequency when the displacement change rate is higher than the displacement threshold, and generating optimized distribution data;

[0037] Step S344, using OpenCV's KalmanFilter to smooth the optimized distribution data, setting the state variables as grid point coordinates and the observation matrix as the identity matrix to obtain a smooth distribution result;

[0038] Step S345 , comparing the smoothed distribution result with the surface morphology data acquired by the 3D scanner grid point by grid point, and marking as a deviation area when the coordinate difference exceeds the tolerance threshold;

[0039] Step S346, using the Sobel operator to calculate the spatial gradient of the deviation area, and determining that the deviation is valid when the gradient amplitude is greater than the gradient threshold;

[0040] Step S347, dynamically adjusting the weighted fusion coefficient according to the gradient amplitude, setting the weight of the optimized distribution data to the inverse of the gradient amplitude, and setting the weight of the surface morphology data to the normalized gradient amplitude, to generate the corrected distribution data;

[0041] Step S348: Calculate the second-order derivative of the displacement change rate of each grid point based on the corrected distribution data. If the second-order derivative exceeds the derivative threshold for three consecutive frames, the area is determined to be the final feature distribution area.

[0042] Optionally, the step S4 uses a dynamic compensation algorithm to adjust the cutting trajectory and depth in real time based on the monitored material deformation data to ensure cutting accuracy, including:

[0043] Step S41, acquiring material deformation data through a sensor and obtaining deformation characteristics using signal processing technology;

[0044] Step S42, applying a dynamic compensation algorithm to determine the trajectory adjustment amount and the depth adjustment amount according to the deformation characteristics;

[0045] Step S43: if the adjustment amount exceeds the adjustment threshold, the cutting trajectory and cutting depth are updated through the control system;

[0046] Step S44, using trajectory optimization technology to obtain a smooth path for the updated cutting trajectory;

[0047] Step S45, adjusting the cutting depth through the depth control module to obtain a stable output;

[0048] Step S46, obtaining the adjusted cutting data and determining whether it meets the requirements through the accuracy detection module;

[0049] Step S47: If the accuracy test result is lower than the standard, the dynamic compensation algorithm is repeated to adjust the trajectory and depth.

[0050] Optionally, the step S42, applying a dynamic compensation algorithm to determine the trajectory adjustment amount and the depth adjustment amount according to the deformation characteristics, includes:

[0051] Step S421, collecting deformation features through sensors and extracting feature input from the collected data;

[0052] Step S422, classifying the feature input using preset rules to obtain a classification result;

[0053] Step S423, processing the classification results through Kalman filtering to determine trajectory data;

[0054] Step S424: if the trajectory data exceeds the trajectory threshold, the trajectory data is updated by the control module;

[0055] Step S425: Using cubic spline interpolation processing based on the updated trajectory data to obtain a smooth path;

[0056] Step S426, adjusting the depth data through principal component analysis to obtain an adjustment result;

[0057] Step S427: extract the output result from the adjustment result and determine whether it meets the standard through the detection module.

[0058] Optionally, the step S5, wherein a high-precision laser interferometry system is used to continuously track the position of the cutting tool to obtain a deviation value between the actual cutting trajectory and the target trajectory, includes:

[0059] Step S51, initializing the measurement system through laser interferometry technology to obtain the initial position information of the cutting tool, and passing the initial position information to the Kalman filter for continuous tracking processing to obtain real-time data;

[0060] Step S52: passing the real-time data to the least square method for trajectory comparison, comparing the actual trajectory with the target trajectory, and determining the deviation value;

[0061] Step S53: if the deviation value exceeds the second deviation threshold, updating the position information of the cutting tool by adjusting the tool state;

[0062] Step S54: passing the real-time data to a time series analysis to analyze the changing trend of the actual trajectory and determine the stability of the trajectory comparison;

[0063] Step S55, obtaining the fluctuation range of the change trend and determining the correction direction of the deviation value;

[0064] Step S56: Update the target trajectory using the corrected deviation value to obtain an optimized cutting path.

[0065] Optionally, the step S7, adjusting the contact pressure between the cutting tool and the material by a programmable pressure control system to achieve precise processing of surface textures in different regions, includes:

[0066] Step S71, obtaining initial height data of the material surface by a contact profilometer;

[0067] Step S72, using the K-means clustering algorithm to divide the height data into regions, and the number of clusters is automatically determined according to the contour curvature;

[0068] Step S73 , generating a tool pressure parameter using a preset pressure-height mapping table according to the average height and roughness of each cluster area;

[0069] Step S74, writing the pressure parameter into the register address of the PLC controller to control the servo motor to adjust the tool spindle pressure;

[0070] Step S75, after the processing is performed, the surface height data is collected again using the profilometer, and the root mean square error of the height before and after the processing of each area is calculated;

[0071] Step S76 , if the error in any region exceeds 5 microns, trigger the PID controller to dynamically adjust the gain coefficient of the pressure mapping table;

[0072] Step S77 , repeatedly performing processing and measurement until the errors of all regions are lower than the error threshold.

[0073] Optionally, in step S8, for the area requiring differentiated friction coefficient, a multi-scale surface structure generation algorithm is used to create a specific microstructure in combination with pressure control and cutting depth adjustment, including:

[0074] Step S81, acquiring regional difference data through sensors, processing the sensor data using a multi-scale surface structure generation algorithm, and generating a friction coefficient distribution;

[0075] Step S82, extracting a pressure control parameter from the friction coefficient distribution, and determining a cutting depth adjustment range using a first pressure threshold;

[0076] Step S83, generating an initial surface structure model according to the cutting depth adjustment range, and extracting microstructure features in the model;

[0077] Step S84, adjusting the pressure control accuracy according to the microstructure characteristics and determining the surface structure optimization scheme;

[0078] Step S85, using a convolutional neural network to iterate the surface structure optimization scheme to obtain a specific structural morphology;

[0079] Step S86, determining whether a second pressure threshold is met by calculating the generation efficiency of the specific structure form, and if not, adjusting the multi-scale method parameters;

[0080] Step S87: generating final microstructure data according to the adjusted multi-scale method parameters.

[0081] A second aspect of the present invention provides a system for precisely cutting and processing soft materials, which uses the above-described method to precisely cut and process soft materials, and the system further includes:

[0082] Data acquisition module, used to obtain 3D model data and target cutting parameters of soft materials, including cutting trajectory, depth and surface texture information;

[0083] The path planning module is used to generate the initial cutting trajectory based on the 3D model data and target cutting parameters using an adaptive cutting path planning algorithm, taking into account material properties and cutting tool limitations;

[0084] The deformation monitoring module is used to monitor the material deformation during the cutting process in real time through multi-sensor fusion technology, and obtain the material surface morphology and internal stress distribution data;

[0085] Dynamic compensation module is used to adjust the cutting trajectory and depth in real time based on the monitored material deformation data using dynamic compensation algorithm to ensure cutting accuracy;

[0086] The trajectory tracking module is used to continuously track the position of the cutting tool using a high-precision laser interferometer measurement system to obtain the deviation value between the actual cutting trajectory and the target trajectory;

[0087] The closed-loop control module is used to make micron-level position corrections to the cutting tool based on the deviation value using a closed-loop feedback control algorithm to ensure accurate execution of the cutting trajectory.

[0088] The pressure regulation module is used to adjust the contact pressure between the cutting tool and the material through a programmable pressure control system to achieve precise processing of surface textures in different areas;

[0089] The texture generation module is used to create specific microstructures in areas where differentiated friction coefficients are required, combining pressure control and cutting depth adjustment with a multi-scale surface structure generation algorithm;

[0090] The quality inspection module is used to monitor the chemical composition changes of the cut surface using real-time spectral analysis technology to determine whether the preset surface property requirements are met. If not, it returns to adjust the cutting parameters and re-process the area.

[0091] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0092] The present invention provides a method and system for precise cutting of soft materials. First, the three-dimensional model of the material and cutting parameters are acquired to generate an initial cutting trajectory. During the cutting process, material deformation is monitored in real time through multi-sensor fusion technology, and the cutting trajectory and depth are adjusted using a dynamic compensation algorithm. At the same time, a laser interferometry measurement system is used to track the position of the cutting tool, and closed-loop feedback control is used to achieve micron-level precision. The present invention also combines pressure control and cutting depth adjustment to create a specific microstructure to achieve a differentiated friction coefficient. In addition, real-time spectral analysis is used to monitor the surface chemical composition to ensure that the preset surface property requirements are met. The present invention can accurately control the cutting trajectory, depth and surface texture of soft materials, achieve high-precision, customizable soft material processing, and provide key technical support for intelligent processing in the fields of flexible electronics, bionic robots, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 The figure is a flow chart of a method for accurately cutting soft materials according to the present invention.

[0094] Figure 2 This is a structural diagram of a system for precisely cutting and processing soft materials according to the present invention. DETAILED DESCRIPTION

[0095] The following will describe the technical solutions in the embodiments of the present invention in detail with reference to the accompanying drawings. The described embodiments are only a part of the embodiments of the present invention.

[0096] like Figure 1 As shown, the first aspect of the present invention provides a method for precise cutting of soft materials, which may specifically include:

[0097] S1, obtain the 3D model data and target cutting parameters of the soft material, including cutting trajectory, depth and surface texture information.

[0098] Optionally, this step also includes:

[0099] Step S11 , obtaining three-dimensional model data of the soft material, and obtaining complete model data through a three-dimensional scanner.

[0100] Step S12: extracting target parameters from the complete model data, including geometric dimensions, material properties, and surface texture.

[0101] Step S13: Using a finite element analysis tool to mesh the model, and determine the initial values of the cutting trajectory, cutting depth, and surface texture.

[0102] Step S14: Generate a trajectory path for the initial value, calculate the deviation between the trajectory path and the target parameter, and if the deviation exceeds a first deviation threshold, use the gradient descent method to perform iterative adjustment to obtain the optimized trajectory path.

[0103] Step S15 , calculating the depth value according to the optimized trajectory path, and using a cubic spline interpolation algorithm to determine a smooth depth value.

[0104] Step S16: extracting texture features by smoothing the depth value, calculating the difference between the texture features and the surface texture, and if the difference exceeds a difference threshold, adjusting the depth value until a matching texture feature is obtained.

[0105] Step S17: After obtaining the matching texture features, the optimized trajectory path, the smoothed depth value and the matching texture features are integrated to obtain the final cutting parameters.

[0106] Step S18: Generate cutting data of the three-dimensional model from the final cutting parameters, perform simulation verification using MATLAB, and determine the integrity of the cutting data.

[0107] For example, to obtain 3D model data for a soft material, a 3D scanner can be used to scan a piece of silicone material, such as a 20 cm long, 15 cm wide, and 5 cm thick soft sample. The scanner uses laser or structured light technology to capture point cloud data on the object's surface, forming a complete 3D model. This method accurately records geometric shapes and surface details, ensuring the integrity of the underlying data for subsequent analysis.

[0108] For example, the scanning resolution is set to 0.1 mm to ensure that subtle features are not lost. When extracting target parameters from the complete model, geometric dimensions can be directly measured, such as length, width, and thickness of 20, 15, and 5 cm, respectively. Material properties can be inferred from known silicone properties, such as a Young's modulus of 2 MPa. Surface texture is analyzed using point cloud data to determine roughness, such as an average roughness Ra of 0.05 mm. These parameters provide a basis for subsequent modeling.

[0109] When meshing the model using finite element analysis tools, a tetrahedral mesh was used, dividing the silicone model into approximately 100,000 elements. When determining the cutting trajectory, it was assumed that the cut would be along the centerline of the long edge. The initial cutting depth was set to 2 cm, and the initial surface texture value was adjusted based on the roughness Ra.

[0110] Preferably, after the trajectory path is generated, its deviation from the target parameters is calculated.

[0111] For example, if a trajectory deviates from the centerline by 0.2 mm, exceeding the first deviation threshold of 0.1 mm, we can iteratively adjust the path using gradient descent until the deviation is less than the first deviation threshold. This method improves cutting accuracy and ensures that the path matches the design intent.

[0112] In one possible implementation, when calculating the depth value based on the optimized trajectory path, if abrupt depth changes are found, a cubic spline interpolation algorithm may be used for smoothing.

[0113] For example, the depth gradually changes from 2 cm to 2.5 cm, generating a continuous curve and avoiding uneven cutting surfaces.

[0114] It's important to note that smoothing the depth value can reduce stress concentration and improve the stability of the material after cutting. When extracting texture features using the smoothing depth value, the cut surface roughness can be analyzed. If the target texture Ra is 0.06 mm and the actual value is 0.07 mm, exceeding the 0.01 mm difference threshold, the depth is fine-tuned to 2.3 cm until the texture matches. This iterative adjustment ensures that surface quality meets standards.

[0115] Specifically, the final cutting parameters are obtained after integrating the optimized trajectory path, smoothed depth value, and matched texture features.

[0116] For example, the trajectory has a straight line deviation of less than 0.1 mm, a smooth depth of 2.3 cm, and a texture Ra of 0.06 mm. Cutting data generated from these parameters can be output as G-code for use with CNC equipment. In MATLAB simulation verification, the cutting data is loaded to simulate stress distribution and deformation.

[0117] For example, the maximum stress after cutting was 1.5 MPa, which did not exceed the material limit, proving the integrity of the data. This type of simulation can identify potential problems in advance and improve process reliability.

[0118] For example, if a soft material is used to manufacture a flexible gasket, the above method can ensure accurate dimensions and smooth surface, thereby improving sealing performance.

[0119] Understandably, deviation adjustment and texture matching can also reduce material waste and extend service life.

[0120] In one embodiment, a slight curve optimization of the cutting trajectory can also accommodate complex shapes, expanding application scenarios. Through precise modeling and iterative optimization, the overall solution ensures high efficiency throughout the entire process, from data acquisition to processing verification. S2 uses an adaptive cutting path planning algorithm to generate the initial cutting trajectory based on the 3D model data and target cutting parameters.

[0121] Optionally, in this step, material properties and cutting tool limitations need to be considered.

[0122] Optionally, this step also includes:

[0123] Step S21 : obtaining STEP format data through the three-dimensional model, and using the OpenCASCADE geometry kernel to extract length error and curvature radius as geometric features.

[0124] Step S22: Match the Young's modulus data in the material library according to the geometric feature results, and establish a linear relationship table between the modulus and the cutting speed.

[0125] Step S23: Obtain the feed speed and spindle speed from the relationship table, and use a preset speed safety threshold to determine whether they exceed the tool limit value.

[0126] In step S24 , a gradient descent method is used to generate a collision-free initial machining path based on the adjusted rotation speed parameter and the maximum feed rate limit of the tool.

[0127] Step S25 , calculating the material removal rate through the initial path, and correcting the path deviation according to the cutting force feedback.

[0128] Step S26: Replan the tool axis direction using the Dijkstra algorithm based on the corrected path and tool orientation constraints.

[0129] Step S27: extracting the layered cutting sequence from the planning result and generating G-code instructions executable by the CNC machine tool.

[0130] For example, when obtaining STEP format data through a three-dimensional model, the OpenCASCADE geometry kernel can be used to process complex geometric information.

[0131] For example, consider a 3D model of a soft material measuring 30 cm long, 20 cm wide, and 8 cm thick, with curved surfaces. When extracting length errors using OpenCASCADE, the model's edges can be analyzed for deviations from theoretical values. For example, a 0.15 mm deviation in the length of a particular edge can be detected. When extracting curvature radius, the minimum curvature radius for the curved surface can be calculated to be 5 mm. These geometric features provide the basis for subsequent matching.

[0132] In a possible implementation, Young's modulus data in a material library is matched according to the extracted geometric features and can be searched from a preset database.

[0133] For example, a length error of 0.15 mm and a curvature radius of 5 mm correspond to a Young's modulus of 1.8 MPa for the soft material.

[0134] It should be noted that when establishing a linear relationship table between modulus and cutting speed, experimental data can be fitted. For example, a modulus of 1.8 MPa corresponds to a cutting speed range of 50 to 80 mm per second, and a table can be formed for query.

[0135] Specifically, when obtaining the feed rate and spindle speed from the relationship table, assuming that the modulus 1.8 MPa corresponds to a cutting speed of 60 mm / s, the feed rate can be set to 40 mm / s and the spindle speed to 12,000 rpm.

[0136] Preferably, a preset speed safety threshold is used for judgment, for example, the tool limit is 15,000 revolutions per minute, and the speed of 12,000 is not exceeded to ensure processing safety.

[0137] In one embodiment, a gradient descent method is used to generate a collision-free path for the adjusted rotation speed and maximum tool feed rate limit.

[0138] For example, the initial path was 2 mm close to the edge of the model, which posed a collision risk. Through iterative adjustment, the offset was adjusted to 3 mm to ensure a smooth path.

[0139] It is understandable that when calculating the material removal rate, it is assumed that 5 cubic centimeters of material are removed per minute. Based on the cutting force feedback, it is found that the path deviation is 0.2 mm. After correction, the deviation is reduced to 0.05 mm, improving the accuracy.

[0140] For example, when using the Dijkstra algorithm to replan the tool axis direction, the optimal angle can be calculated based on the corrected path and tool orientation constraints.

[0141] In one embodiment, the knife axis is adjusted from vertical to inclined at 15 degrees to avoid interference. When extracting the layered cutting sequence from the planning results, the 8 cm thickness can be divided into 4 layers, each 2 cm, and the generation order is from top to bottom.

[0142] In a possible implementation, when generating G-code instructions, the layering sequence and path data may be converted into a format recognizable by the machine tool.

[0143] For example, the starting coordinates of the first layer cutting path are X0Y0Z8 and the end coordinates are X30Y0Z8. After outputting the G code, it can be directly used for CNC equipment processing.

[0144] It should be noted that this method can ensure that the processing path is efficient and stable, and improve the overall consistency of soft material processing.

[0145] S3 uses multi-sensor fusion technology to monitor material deformation during the cutting process in real time and obtain material surface morphology and internal stress distribution data.

[0146] Optionally, this step also includes:

[0147] Step S31: collecting original signals during the cutting process through a multi-sensor system to obtain deformation data and stress data.

[0148] Step S32: weighted average fusion is used to integrate the collected original signals to determine the surface morphology characteristics and internal stress characteristics.

[0149] Step S33: If the surface morphological feature exceeds the deformation threshold, the deformation trend is determined by the least square method.

[0150] Step S34: adjusting the acquisition technology according to the deformation trend to obtain more accurate distribution information.

[0151] Step S35: Optimize the distribution information through Kalman filtering to obtain smooth stress distribution data.

[0152] Step S36: If the smoothed stress distribution data is inconsistent with the internal information, the deviation area is determined by the difference method.

[0153] Step S37: Use the deviation area data to update the weighted average fusion parameters to obtain more accurate real-time acquisition results.

[0154] For example, deformation data and stress data can be obtained by collecting raw signals during the cutting process through a multi-sensor system.

[0155] For example, when processing a piece of soft material with dimensions of 25 cm long, 15 cm wide and 6 cm thick, strain gauges and pressure sensors can be arranged on the surface and inside of the material respectively.

[0156] For example, a strain gauge detects a deformation of 0.2 mm at a surface point, while a pressure sensor records a stress of 2.5 MPa in an internal region. These data provide the basis for subsequent analysis. Using weighted average fusion to integrate the raw signals, surface morphology and internal stress characteristics can be determined.

[0157] Specifically, assuming that the surface deformation data and internal stress data are assigned weights of 0.6 and 0.4 respectively, through fusion calculation, it is concluded that the comprehensive deformation characteristics of a certain area on the surface are 0.18 mm, and the internal stress characteristics are 2.2 MPa.

[0158] In one possible implementation, this method can effectively reduce noise interference from a single sensor and improve data reliability. If the surface morphology exceeds a deformation threshold, for example, the set deformation threshold is 0.15 mm, and the actual value is 0.18 mm, the least squares method is used to determine the deformation trend.

[0159] For example, after collecting multiple sets of data points, a curve is fitted to show the gradual increase in deformation over time. This trend analysis helps to identify potential problems in advance. By adjusting the collection technology based on the deformation trend, more accurate distribution information can be obtained.

[0160] Preferably, if it is found that the deformation is concentrated in the edge area, the sensor density in this area can be increased, for example, from 1 sensor per square centimeter to 2 sensors per square centimeter.

[0161] It should be noted that this can capture local changes more carefully and improve the spatial resolution of the data. By optimizing the distribution information through Kalman filtering, smooth stress distribution data can be obtained.

[0162] In one embodiment, the raw stress data exhibits fluctuations, for example, the stress value at a certain point fluctuates between 2.3 and 2.7 MPa, but stabilizes at 2.4 MPa after filtering. This smoothing process reduces interference from sudden changes, facilitating subsequent analysis. If the smoothed stress distribution data is inconsistent with the internal information, for example, if the surface stress is 2.4 MPa while the internal measurement is 2.8 MPa, the deviation area is determined through a differential method.

[0163] As expected, calculating the difference between the two sets of data revealed that the deviation was primarily concentrated in the middle layer of the material thickness, with a deviation of 0.4 MPa. This positioning method helps focus on the problem area. Using the deviation area data to update the weighted average fusion parameters can achieve more accurate real-time acquisition results.

[0164] For example, adjusting the weight of the middle layer from 0.4 to 0.5 reduces the deviation of surface and internal stress features to 0.1 MPa after reintegration.

[0165] In one possible implementation, this dynamic adjustment can improve the system's adaptability to complex stress distributions and ensure the real-time and consistency of data during the processing process.

[0166] Optionally, the step S34 of adjusting the acquisition technology according to the deformation trend to obtain more accurate distribution information may also include:

[0167] Step S341 : acquiring initial distribution data through dynamic sampling, wherein the sampling interval is determined according to a preset spatial grid density.

[0168] Step S342: input the initial distribution data into a deformation trend calculation module based on moving average, and output the displacement change rate of each grid point.

[0169] Step S343 , adjusting the sampling frequency according to the displacement change rate. When the displacement change rate is higher than the displacement threshold, the sampling frequency is doubled to generate optimized distribution data.

[0170] Step S344: Use KalmanFilter of OpenCV to smooth the optimized distribution data, set the state variables as grid point coordinates, and set the observation matrix as the unit matrix to obtain a smooth distribution result.

[0171] Step S345 : Compare the smoothed distribution result with the surface morphology data acquired by the three-dimensional scanner grid point by grid point, and mark as a deviation area when the coordinate difference exceeds the tolerance threshold.

[0172] In step S346, the Sobel operator is used to calculate the spatial gradient of the deviation area, and when the gradient amplitude is greater than the gradient threshold, it is determined to be a valid deviation.

[0173] Step S347, dynamically adjust the weighted fusion coefficient according to the gradient amplitude, set the weight of the optimized distribution data to the inverse of the gradient amplitude, and set the weight of the surface morphology data to the normalized gradient amplitude, to generate corrected distribution data.

[0174] Step S348: Calculate the second-order derivative of the displacement change rate of each grid point based on the corrected distribution data. If the second-order derivative exceeds the derivative threshold for three consecutive frames, the area is determined to be the final feature distribution area.

[0175] For example, when dynamic sampling is performed to obtain initial distribution data, it is understood that the selection of the sampling interval directly affects the spatial resolution of the data.

[0176] For example, during the cutting process, assuming the material surface is divided into a 10mm x 10mm grid, if the preset grid density is 1mm, the sampling interval is 1mm. This method can capture subtle deformation features.

[0177] For example, if the cutting speed is fast, the grid density can be adjusted to 0.5 mm to adapt to the rapidly changing material state.

[0178] In a possible implementation, when the initial distribution data is input into a deformation trend calculation module based on moving average, the displacement change can be analyzed through a time window.

[0179] For example, if we set a 5-second time window and calculate the average displacement of each grid point during this time, we can obtain the rate of change. For example, if the displacement of a grid point changes from 0 mm to 0.2 mm, the rate of change is 0.04 mm / s. This method can effectively reflect the deformation trend of a local area.

[0180] Specifically, when adjusting the sampling frequency according to the displacement change rate, if the change rate of a grid point reaches 0.05 mm / s, which is higher than the derivative threshold of 0.03 mm / s, the sampling frequency is doubled from 1 time per second to 2 times per second.

[0181] Preferably, this dynamic adjustment can more accurately capture areas of rapid deformation, such as material portions near cutting tools, where displacement changes are typically more dramatic. When using OpenCV's KalmanFilter for smoothing.

[0182] It should be noted that the state variables are set to grid point coordinates in order to track position changes.

[0183] For example, if the initial coordinates of a grid point are (5,5), and the observed value is biased to (5.1,5.2) due to noise, it can be corrected to (5.05,5.1) after filtering. This smoothing process can reduce noise interference and improve data reliability.

[0184] In one embodiment, when comparing the smoothed distribution result with the 3D scanner data, assuming that the coordinate difference of a certain grid point is 0.3 mm, which exceeds the tolerance threshold of 0.2 mm, it is marked as a deviation area.

[0185] For example, areas near the cutting edge often exhibit large deviations due to stress concentration. This comparison allows for quick location of abnormal areas.

[0186] For example, when using the Sobel operator to calculate spatial gradients, if the gradient amplitude of a deviation region is 10, which is higher than the gradient threshold of 5, it is considered a valid deviation. This method can highlight areas of severe deformation, such as localized bulges on the material surface caused by high temperatures. Gradient analysis helps identify key features.

[0187] Preferably, when dynamically adjusting the weighted fusion coefficient according to the gradient amplitude, assuming that the gradient amplitude is 10, the weight of the optimized distribution data is set to 0.1 and the weight of the surface morphology data is set to 0.9. This adjustment can more accurately reflect the actual state.

[0188] For example, when cutting thin plates, the surface morphology data is more reliable due to direct measurement, and increased weighting can help correct the results.

[0189] It is understandable that when calculating the second-order derivative based on the corrected distribution data, if the second-order derivative of a grid point for three consecutive frames is 0.02, 0.03, and 0.04 mm / s 2 , both exceeded the derivative threshold of 0.01 mm / s 2 , it is determined as the final characteristic distribution area. This determination can pinpoint the area with the most significant deformation, such as the part that continues to deform due to stress release during the cutting process, which helps to optimize subsequent processes.

[0190] Optionally, the step S37, in which the weighted average fusion parameters are updated using the deviation area data to obtain a more accurate real-time acquisition result, further includes:

[0191] Step S371: extract the data update content from the deviation area to obtain the updated data set.

[0192] Step S372: Input the updated data set into the weighted average algorithm, adjust the weighted average parameters, and obtain a new weighted average result.

[0193] Optionally, use the following formula to calculate the weighted average result:

[0194]

[0195] Among them, W(x) is the weighted average result, n represents the number of samples in the data set, and X i represents the i-th sample value, α i represents the corresponding weight coefficient, β represents the adjustment factor, λi represents the updated weight, Δx i Indicates the amount of change in sample values.

[0196] Step S373: Calculate the change value of the fusion parameter based on the new weighted average result.

[0197] Step S374: Apply the changed value of the fusion parameter to real-time data processing to obtain intermediate acquisition results.

[0198] Step S375: If the intermediate acquisition result does not match the deviation area, the K-means clustering algorithm is used to analyze the deviation area to determine the correction direction.

[0199] Step S376: Adjust the fusion result according to the correction direction to obtain optimized data.

[0200] Step S377: Input the optimized data into the judgment conditions to determine the final collection results.

[0201] Exemplarily, when extracting data update content from the deviation area and obtaining an updated data set, it can be understood that this process aims to obtain the latest information from the previously marked abnormal area.

[0202] For example, when cutting the surface of a material, areas close to the tool may experience large deviations due to stress concentration. Extracting displacement data or morphological change data from these areas can provide a basis for subsequent analysis.

[0203] In one possible implementation, assume that the displacement value of a certain deviation area changes from 0.2 mm to 0.3 mm. By recording these changes, the updated data set can reflect the real-time state changes of the material.

[0204] When the updated data set is input into the weighted average algorithm, the weighted average parameters are adjusted, and a new weighted average result is obtained, it should be noted that the core of weighted average is to balance the contribution of different data.

[0205] For example, if the updated data set contains five sample points, which respectively record displacement values of 0.1 mm, 0.2 mm, 0.3 mm, 0.4 mm, and 0.5 mm, weights can be assigned according to the temporal sequence or spatial position of the sample points.

[0206] For example, the weight of sample points near the cutting edge can be set to 0.3 because the changes are more significant, while those far from the edge can be set to 0.1. The adjustment factor can be adjusted according to the material type, such as 1.2 when cutting thin plates to amplify the impact of critical areas.

[0207] When calculating the change value of the fusion parameter based on the new weighted average result, specifically, this change value reflects the dynamic adjustment of the data fusion.

[0208] For example, if the weighted average result changes from 0.25 mm to 0.28 mm, the change is 0.03 mm, which can be used for calibration in subsequent real-time processing.

[0209] In one embodiment, assuming that the cutting speed is increased, the variation value may increase to 0.05 mm, indicating that the fusion parameters need to be updated more frequently.

[0210] Preferably, the changing values of the fusion parameters are applied to real-time data processing to obtain intermediate acquisition results, and this process can quickly respond to changes in the material state.

[0211] For example, during the cutting process, if the displacement data collected in real time is combined with the change value of 0.03 mm, an intermediate result can be generated to preliminarily determine whether the deviation is still within the tolerance range.

[0212] Understandably, this real-time nature helps to adjust process parameters in a timely manner.

[0213] If the intermediate acquisition result does not match the deviation area, the K-means clustering algorithm is used to analyze the deviation area to determine the correction direction. In one possible implementation, the K-means clustering algorithm can divide the deviation area into multiple clusters.

[0214] For example, if the deviation region contains 10 grid points with displacement values ranging from 0.1 mm to 0.5 mm, clustering can result in two groups: one close to the tool with large variations, and the other farther away with more moderate variations. Correction can be prioritized for clusters with large variations.

[0215] When the fusion result is adjusted in the correction direction to obtain optimized data, specifically, this adjustment can more accurately reflect the status of the key area.

[0216] For example, if the cluster analysis shows that the sampling frequency needs to be increased in the area close to the tool, the fusion result can increase the weight of the data in this area from 0.2 to 0.4, and the final optimized data is closer to the actual deformation characteristics.

[0217] The optimized data is input into the preset judgment conditions to determine the final acquisition results. For example, the displacement threshold is set to 0.3 mm. If the displacement of a grid point in the optimized data is 0.35 mm, it is marked as an area of concern.

[0218] In one embodiment, if the optimized data for the cut edge exceeds the preset judgment criteria, the system can automatically record it as the final result to guide subsequent process improvements. This approach effectively targets key feature areas and improves acquisition accuracy. In step S4, based on the monitored material deformation data, a dynamic compensation algorithm is used to adjust the cutting trajectory and depth in real time to ensure cutting accuracy.

[0219] Optionally, this step also includes:

[0220] Step S41: Obtain material deformation data through sensors and obtain deformation characteristics using signal processing technology.

[0221] Step S42 : applying a dynamic compensation algorithm to determine the trajectory adjustment amount and the depth adjustment amount according to the deformation characteristics.

[0222] Step S43: If the adjustment amount exceeds the adjustment threshold, the cutting trajectory and cutting depth are updated through the control system.

[0223] Step S44: For the updated cutting trajectory, a trajectory optimization technique is used to obtain a smooth path.

[0224] Step S45: adjusting the cutting depth through the depth control module to obtain a stable output.

[0225] Step S46: Obtain the adjusted cutting data and determine whether it meets the requirements through the accuracy detection module.

[0226] Step S47: If the accuracy test result is lower than the standard, the dynamic compensation algorithm is repeated to adjust the trajectory and depth.

[0227] For example, after obtaining material deformation data through sensors, extracting deformation features using signal processing technology is a key first step.

[0228] For example, when machining a piece of soft material, a sensor can capture changes in displacement in a certain area of the surface.

[0229] Specifically, the deformation at a certain point is detected to be 0.3 mm. This raw signal may contain noise and needs to be processed through filtering technology.

[0230] In one possible implementation, a low-pass filter is used to remove high-frequency interference, preserving the true deformation trend and obtaining clear feature data. This approach provides a reliable foundation for subsequent steps. The core of the next step is to apply a dynamic compensation algorithm to determine the amount of trajectory and depth adjustment based on the deformation characteristics.

[0231] For example, if a deformation feature of 0.25 mm is detected somewhere on the material surface, the algorithm will analyze the impact of this change on the cutting path.

[0232] It's understandable that if the trajectory is not adjusted, cutting deviations may occur. Therefore, based on the deformation distribution, the algorithm deduces that the trajectory adjustment should be offset to the left by 0.1 mm, while also increasing the depth adjustment by 0.05 mm. This dynamic adjustment effectively adapts to material changes. If the adjustment exceeds the adjustment threshold—for example, if the trajectory offset threshold is set to 0.08 mm and the actual value is 0.1 mm—the control system updates the cutting trajectory and depth.

[0233] Specifically, the control system sends the new path parameters to the actuators.

[0234] In one embodiment, if the trajectory deviates due to deformation at the material edge, the updated trajectory will realign with the target area. This real-time update ensures process stability. It is particularly important to use trajectory optimization technology to generate a smooth path for the updated cutting trajectory.

[0235] For example, the new trajectory may have sharp corners, which can be smoothed into a continuous curve through spline optimization.

[0236] Preferably, this smooth path can reduce equipment vibration and improve cutting consistency.

[0237] It should be noted that the smoothed path can also reduce mechanical wear. Adjusting the cutting depth through the depth control module is a key step to ensure stable output.

[0238] In one possible implementation, assuming the initial depth is 5 mm and needs to be adjusted to 5.05 mm due to deformation, the module will accurately control the tool's downward pressure distance.

[0239] For example, this fine adjustment can avoid incomplete cutting due to insufficient depth and ensure processing quality. The adjusted cutting data needs to be verified by the accuracy detection module.

[0240] For example, after collecting a set of data, it is found that the deviation of a certain section of the cutting path is 0.02 mm, while the standard requires the deviation to be less than 0.03 mm, then it is considered to meet the requirements.

[0241] In one embodiment, the inspection module uses a laser measuring instrument to scan the surface and obtain the actual cutting results. This inspection method can quickly determine whether the processing meets the standards. If the accuracy test result is below the standard, for example, the deviation reaches 0.04 mm, the dynamic compensation algorithm is repeated to adjust the trajectory and depth.

[0242] Specifically, the system will reanalyze the deformation characteristics and derive new adjustment amounts, such as a trajectory offset of 0.05 mm and a depth increase of 0.03 mm.

[0243] It is understandable that this iterative adjustment can gradually approach the target accuracy.

[0244] Preferably, multiple adjustments can also optimize the system's adaptability to complex materials and ensure that the final output meets the requirements.

[0245] Optionally, the step S42 of applying a dynamic compensation algorithm to determine the trajectory adjustment amount and the depth adjustment amount according to the deformation characteristics further includes:

[0246] Step S421: collect deformation features through sensors and extract feature input from the collected data.

[0247] Step S422: classify the feature input using preset rules to obtain a classification result.

[0248] Step S423: Process the classification result through Kalman filtering to determine trajectory data.

[0249] Step S424: If the trajectory data exceeds the trajectory threshold, the trajectory data is updated by the control module.

[0250] Step S425 : Using cubic spline interpolation processing based on the updated trajectory data to obtain a smooth path.

[0251] Step S426: adjust the depth data through principal component analysis to obtain an adjustment result.

[0252] Step S427: extract the output result from the adjustment result and determine whether it meets the standard through the detection module.

[0253] For example, when using sensors to collect deformation characteristics, a high-precision displacement sensor can be used to monitor small changes in the material surface in real time. The sensor sampling frequency can be set to 100 times per second to capture key data points during the dynamic deformation process.

[0254] In one possible implementation, the collected raw data may contain noise, so it needs to be initially processed through a low-pass filter to retain the main deformation trends. When extracting feature input, two key parameters, amplitude and frequency, can be separated from the data.

[0255] For example, an amplitude of 0.05 mm and a frequency of 10 Hz indicate that the material has undergone significant deformation.

[0256] When feature input is classified using pre-set rules, three levels can be set based on the degree of deformation: mild, moderate, and severe. An amplitude less than 0.02 mm is classified as mild, 0.02 to 0.05 mm as moderate, and over 0.05 mm as severe. This classification method facilitates quick assessment of the material's real-time status during subsequent processing.

[0257] It should be noted that the classification results directly affect the subsequent trajectory adjustment strategy, so the setting of rules needs to be based on a large amount of experimental data to ensure accuracy.

[0258] When processing the classification results through Kalman filtering, for example, the prediction step size of the filter can be set to 0.1 seconds, combining the previous trajectory data and the current deformation characteristics to generate a more stable output.

[0259] Preferably, if the prediction result shows that the trajectory deviation reaches 0.03 mm and exceeds the trajectory threshold of 0.02 mm, it indicates that an immediate update is required. This method can effectively reduce the error caused by external interference and improve the reliability of trajectory data.

[0260] In one embodiment, when the trajectory data is updated by the control module, an adjustment instruction can be sent to the servo motor to change the position of the cutting head in real time.

[0261] For example, when the offset is 0.03 mm, the motor adjusts the angle by 1 degree, and the response time is controlled within 50 milliseconds. This ensures that the update process is fast and without noticeable delay.

[0262] According to the updated trajectory data, when cubic spline interpolation is used, it can be understood that this method generates a smooth curve by calculating multiple control points.

[0263] For example, on the path between three consecutive sampling points, the curvature change rate after interpolation is controlled within 5%, ensuring that the path is smooth and easy to execute.

[0264] When adjusting depth data through principal component analysis, in one possible implementation, main influencing factors can be extracted from multidimensional data.

[0265] For example, analysis results show that temperature changes account for 60% of the impact on depth, while pressure affects only 30%. Based on this, changes in temperature are prioritized when adjusting depth. For example, a 10°C increase in temperature results in a 0.01mm decrease in depth. This approach more accurately reflects actual operating conditions.

[0266] When extracting output results from the adjustment results, for example, the adjusted depth is stabilized at 2 mm, and the fluctuation range is controlled within 0.005 mm.

[0267] When the inspection module determines whether the standard is met, specifically, the accuracy standard can be set to ±0.01 mm. If the test data is 1.99 mm, which meets the requirements, the next process will be continued; if it is 1.95 mm, further adjustment will be required. This inspection method can detect deviations in a timely manner and ensure that the final output meets expectations. For example, the advantage of this method is that it improves the stability of the process, especially in high-precision processing scenarios, which can effectively reduce the scrap rate and improve consistency.

[0268] S5, uses a high-precision laser interferometry measurement system to continuously track the cutting tool position and obtain the deviation value between the actual cutting trajectory and the target trajectory.

[0269] Optionally, this step also includes:

[0270] Step S51, initialize the measurement system through laser interferometry technology, obtain the initial position information of the cutting tool, pass the initial position information to the Kalman filter for continuous tracking processing, and obtain real-time data.

[0271] In step S52 , the real-time data is passed to the least square method for trajectory comparison, and the actual trajectory is compared with the target trajectory to determine the deviation value.

[0272] Step S53: If the deviation value exceeds the second deviation threshold, the position information of the cutting tool is updated by adjusting the tool state.

[0273] In step S54, the real-time data is passed to a time series analysis to analyze the changing trend of the actual trajectory and determine the stability of the trajectory comparison.

[0274] Step S55: Obtain the fluctuation range of the change trend and determine the correction direction of the deviation value.

[0275] Step S56: Update the target trajectory using the corrected deviation value to obtain an optimized cutting path.

[0276] For example, when initializing the measurement system through laser interferometry technology, the core is to use the high-precision characteristics of the laser to obtain the initial position information of the cutting tool.

[0277] For example, when machining a sheet of metal, a laser interferometer can emit a beam and use the reflected signal to determine the relative distance between the tool and the material surface. Assuming the initial position is recorded as 10 mm in the X axis and 15 mm in the Y axis, this method can provide a precise starting point for subsequent tracking.

[0278] In one possible implementation, the initial data is passed directly to a Kalman filter, which continuously tracks the tool's position by fusing the sensor data with a predictive model.

[0279] For example, if a tool experiences slight vibration while moving, the position captured by the sensor may fluctuate by 0.2 mm. The Kalman filter combines historical data with current input to smooth the output of the real-time position, such as 10.1 mm on the X axis and 15.05 mm on the Y axis. This real-time data lays the foundation for subsequent analysis.

[0280] When passing real-time data to the least squares method for trajectory comparison, the goal is to quantify the difference between the actual trajectory and the target trajectory.

[0281] Specifically, suppose the target trajectory is a straight line starting at 10 mm on the X axis and 15 mm on the Y axis, and ending at 20 mm on the X axis and 25 mm on the Y axis. However, due to tool offset, the actual trajectory is recorded as starting at 10.2 mm on the X axis and 15.3 mm on the Y axis, and ending at 20.5 mm on the X axis and 25.4 mm on the Y axis. The least squares method fits these two sets of data and calculates the deviation, for example, an average deviation of 0.3 mm.

[0282] Understandably, this comparison can clearly reflect the tool's operating status.

[0283] Preferably, if the deviation value exceeds a second deviation threshold, such as 0.25 mm, the tool state needs to be adjusted.

[0284] For example, if a deviation of 0.3 mm is detected, the control system fine-tunes the tool position via the motor to bring it closer to the target trajectory. The updated position might be 10.1 mm on the X axis and 15.1 mm on the Y axis. This adjustment ensures that the machining path closely matches the intended path. When transmitting real-time data to time series analysis, the key is to identify the trend of trajectory changes.

[0285] In one embodiment, assuming that the actual trajectory points recorded continuously for 10 seconds show that the Y-axis position gradually shifts from 15 mm to 15.4 mm, the time series analysis will identify this slowly increasing trend.

[0286] It’s important to note that this type of analysis not only focuses on current deviations but also predicts future trends, providing a basis for subsequent corrections. By determining the fluctuation range of the trend, we can further determine the direction of deviation correction.

[0287] For example, if the fluctuation range is 0.2 mm to 0.4 mm and the trend is shifting upward, the correction direction should be to adjust the tool position downward.

[0288] Specifically, the system may determine that the correction amount is 0.1 mm each time to gradually reduce the deviation. When the target trajectory is updated with the corrected deviation value, the purpose is to generate an optimized cutting path.

[0289] In one possible implementation, assuming the original target trajectory is a straight line, it is then corrected to a slightly curved path based on the deviation distribution, such as a 0.1mm downward offset at 15mm on the X axis. This optimized path can better adapt to actual processing conditions.

[0290] Preferably, after the new path is generated, sharp change points can be reduced through smoothing processing.

[0291] For example, if there are sudden corners in the adjusted path, they can be optimized into smooth transitions through curve fitting.

[0292] It is understandable that this approach can improve the stability of tool operation and extend the service life of the equipment.

[0293] S6, based on the deviation value, uses a closed-loop feedback control algorithm to perform micron-level position correction on the cutting tool to ensure accurate execution of the cutting trajectory.

[0294] Optionally, this step also includes:

[0295] Step S61: Obtain the millimeter-level deviation value through the sensor, calculate the adjustment value using the PID algorithm, and output a preliminary micron-level correction instruction.

[0296] In step S62 , the servo motor is driven to adjust the position of the cutting tool according to the X / Y axis coordinate values of the preliminary correction instruction.

[0297] Step S63: collecting the position feedback signal through the grating ruler and calculating the offset between the actual position and the command.

[0298] In step S64 , if the offset exceeds the offset threshold of 5 microns, an incremental PID algorithm is used to output a pulse signal to control the piezoelectric ceramic actuator to perform micron-level compensation.

[0299] In step S65 , the compensated position coordinate sequence is input into the least square method to calculate the slope change of the cutting trajectory.

[0300] Step S66: When the slope deviation exceeds 1 degree, the path point coordinates are regenerated using a cubic spline interpolation algorithm.

[0301] Step S67: Modify the feed speed and acceleration parameters of the servo motor according to the new path point coordinates to generate a final execution path.

[0302] Step S68: The actual path coordinates are collected by a laser interferometer, and the adjustment is completed when it is determined that the sampling errors are all less than 2 microns for three consecutive times.

[0303] For example, after obtaining the millimeter-level deviation value through the sensor, using the PID algorithm to calculate the adjustment amount is a key step in ensuring the position accuracy of the cutting tool.

[0304] For example, when machining sheet metal, a sensor might detect a tool deviation of 0.8 mm in the X-axis. The PID algorithm would then calculate an adjustment based on the proportional, integral, and differential parameters, outputting a preliminary correction of, for example, 50 microns. The key to this approach is rapid response to deviations and the appropriate direction for correction.

[0305] It can be understood that the PID algorithm avoids deviation accumulation through real-time adjustment, laying the foundation for subsequent steps.

[0306] In a possible implementation, the servo motor is driven to adjust the tool position according to the X / Y axis coordinate values of the preliminary correction instruction.

[0307] For example, if the instruction requires the tool to move from 10 mm to 10.05 mm on the X-axis, the servo motor will achieve this fine adjustment through precise rotation after receiving the signal.

[0308] Specifically, the servo motor, with its high torque and fast response, can adjust the tool position close to the target value, ensuring efficient execution of instructions.

[0309] It should be noted that after the grating ruler collects the position feedback signal, the adjustment effect can be further verified.

[0310] In one embodiment, the scale detects an actual position of 10.048 mm on the X axis, which is 2 microns offset from the commanded position of 10.05 mm. By calculating this offset, it is possible to determine whether the adjustment has achieved the desired result. The scale's high resolution provides reliable data for subsequent fine-tuning. If the offset exceeds the 5-micron offset threshold, the incremental PID algorithm intervenes and outputs a pulse signal.

[0311] For example, if the actual position is 10.045 mm on the X axis and the offset is 5 microns, the algorithm will generate a series of pulse signals to drive the piezoelectric ceramic actuator to compensate.

[0312] Preferred piezoelectric ceramics, due to their high precision and fast response, can adjust the position to 10.049 mm on the X axis. This micron-level compensation significantly improves trajectory accuracy. The compensated position coordinate sequence is input into the least squares method to analyze the overall trajectory trend.

[0313] For example, assuming the X-axis coordinates of five consecutive sampling points are 10.049 mm, 10.050 mm, 10.052 mm, 10.053 mm, and 10.055 mm, the least squares method will fit a straight line and calculate the slope change. This analysis can reveal the stability of tool movement. When the slope deviates by more than 1 degree, the cubic spline interpolation algorithm will regenerate the path point coordinates.

[0314] For example, the original path points may form a steep curve due to deviations. The interpolation algorithm will generate new coordinates through smoothing, such as adjusting the X-axis position of 10.050 mm to 10.051 mm. This method can optimize the continuity of the path.

[0315] In one embodiment, the feed speed and acceleration parameters of the servo motor are modified according to the new path point coordinates.

[0316] For example, a new path requires adjusting the speed from 5 mm / s to 4.8 mm / s and reducing the acceleration from 2 mm / s² to 1.8 mm / s². This adjustment makes the tool run more smoothly. The effectiveness of the adjustment can be determined by collecting the actual path coordinates using a laser interferometer.

[0317] For example, three consecutive samplings showed errors of 1.8 microns, 1.5 microns and 1.9 microns, all less than 2 microns, indicating that the adjustment is complete.

[0318] Understandably, this high-precision verification ensures the reliability of the machining path.

[0319] S7, through the programmable pressure control system to adjust the contact pressure between the cutting tool and the material, to achieve precise processing of surface textures in different areas.

[0320] Optionally, this step also includes:

[0321] Step S71 , obtaining initial height data of the material surface by a contact profilometer, with a data point spacing of 50 microns.

[0322] Step S72: Use K-means clustering algorithm to divide the height data into regions, and the number of clusters is automatically determined according to the contour curvature.

[0323] Step S73 : generating a tool pressure parameter according to the average height and roughness of each cluster region through a preset pressure-height mapping table.

[0324] Step S74, writing the pressure parameter into the register address of the PLC controller to control the servo motor to adjust the tool spindle pressure.

[0325] Step S75: After the processing is performed, the surface height data is collected again using the profilometer to calculate the root mean square error of the height of each area before and after the processing.

[0326] Step S76: If the error in any region exceeds 5 microns, the PID controller is triggered to dynamically adjust the gain coefficient of the pressure mapping table.

[0327] Step S77 , repeatedly performing processing and measurement until the errors of all regions are lower than the error threshold.

[0328] For example, when obtaining initial height data of a material surface using a contact profilometer, the data point spacing is set to 50 microns, which means that the device scans the surface at fixed intervals to generate a series of height values.

[0329] For example, when machining a sheet of aluminum, a profilometer might record the height every 50 microns along a straight path, giving it a rough representation of the surface.

[0330] It is understandable that this high-density sampling can capture tiny deformations and provide a reliable basis for subsequent analysis.

[0331] In one embodiment, assuming a 100 mm long area is scanned, 2,000 data points are generated, and the height values may range from -20 microns to +30 microns, depending on the material properties. When the K-means clustering algorithm is used to partition the height data into regions, the number of clusters is automatically determined by the contour curvature.

[0332] Specifically, the algorithm divides the surface into several areas by analyzing the distribution and change trend of height values.

[0333] For example, flat areas might be classified into one category, while areas with significant bumps and depressions might be classified into another.

[0334] In one possible implementation, if the surface of an aluminum plate has smooth areas and worn areas, the algorithm may identify three clusters: the average height of the smooth area is 5 microns, the transition area is 15 microns, and the worn area is 25 microns.

[0335] It should be noted that curvature can dynamically adapt to different surface morphologies to ensure the rationality of the division. When generating tool pressure parameters based on the average height and roughness of each cluster area, the preset pressure-height mapping table plays a key role.

[0336] For example, if the smooth area has low roughness, the mapping table may specify a pressure of 10 Newtons, while the worn area has high roughness and the pressure is increased to 15 Newtons. These parameters are then written to the register address of the PLC controller, driving the servo motor to adjust the tool spindle pressure.

[0337] For example, after the PLC receives a 15 Newton command, the servo motor adjusts the spindle downward displacement by 0.2 mm to apply the corresponding force.

[0338] Preferably, this mapping method can be adaptively adjusted according to the surface characteristics to avoid insufficient or excessive pressure. After processing, the surface height data is collected again with a profilometer and the height root mean square error is calculated to verify the processing consistency.

[0339] In one embodiment, the average height of the smooth area before machining was 5 microns, and after machining it became 4.8 microns, with an error of 0.2 microns, well below the error threshold of 5 microns. However, if the error in the worn area reaches 6 microns, it indicates insufficient pressure regulation.

[0340] It is understandable that an error exceeding the limit will trigger the PID controller to dynamically adjust the gain coefficient of the pressure mapping table.

[0341] For example, the original gain factor was 1.0, but after adjustment it was increased to 1.2, causing the pressure in the next round of machining to increase from 15 Newtons to 18 Newtons. Repeating the machining and measurement process until the error falls below the error threshold demonstrates the advantages of closed-loop control.

[0342] Specifically, if the wear zone error is reduced to 4.5 microns after the first round of adjustments and further reduced to 1.8 microns in the second round, the process is complete.

[0343] In one possible implementation, multiple iterations can also optimize the stability of the pressure parameters.

[0344] For example, the initial mapping table may be conservative, but it becomes more accurate after iteration. This method ensures high quality and consistency of the machined surface by gradually approaching the target value.

[0345] It should be noted that the data feedback provided by each measurement lays the foundation for the next round of adjustments and significantly improves the adaptability of the process.

[0346] S8, for areas where differentiated friction coefficients are required, combines pressure control and cutting depth adjustment to create specific microstructures using a multi-scale surface structure generation algorithm.

[0347] Optionally, this step also includes:

[0348] Step S81, obtain regional difference data through sensors, use multi-scale surface structure generation algorithm to process sensor data, and generate friction coefficient distribution. Step S82, extract pressure control parameters from the friction coefficient distribution, and use the first pressure threshold to determine the cutting depth adjustment range. Step S83, generate an initial model of the surface structure based on the cutting depth adjustment range, and extract the microstructure features in the model. Step S84, adjust the pressure control accuracy based on the microstructure features, and determine the surface structure optimization scheme. Step S85, use a convolutional neural network to iterate the surface structure optimization scheme to obtain a specific structural morphology. Step S86, determine whether the second pressure threshold is met by calculating the generation efficiency of the specific structural morphology. If not, adjust the multi-scale method parameters. Step S87, generate the final microstructure data based on the adjusted multi-scale method parameters.

[0349] For example, acquiring regional difference data through sensors is a key step in surface structure processing.

[0350] For example, a laser triangulation sensor may be used to scan the surface of the material to obtain height distribution data.

[0351] In one possible implementation, the sensor samples at 0.1mm intervals, covering a 100mm x 100mm area, generating a 1000 x 1000 dot matrix. This high-density sampling effectively captures microscopic topographical features. The multiscale surface structure generation algorithm can utilize wavelet transforms when processing sensor data.

[0352] Specifically, features of different scales are extracted through the decomposition and reconstruction process to achieve a comprehensive analysis from macro to micro.

[0353] For example, by analyzing the surface of titanium alloy, structural information can be extracted at three scales: 1mm, 100μm, and 10μm, and a friction coefficient distribution map can be generated.

[0354] It should be noted that when extracting the pressure control parameters from the friction coefficient distribution, a mapping relationship between the friction coefficient and the cutting force can be established.

[0355] In one embodiment, when the friction coefficient is in the range of 0.1-0.3, the corresponding cutting force is set to 10-30 N. This mapping method can dynamically adjust the processing parameters according to local characteristics.

[0356] Preferably, using the first pressure threshold to determine the cutting depth adjustment range can improve machining accuracy.

[0357] For example, the cutting depth can be set to vary no more than 5% of the material thickness to avoid overcutting. Based on this range, an initial model of the surface structure, such as spiral grooves and meshes, can be generated. Adaptive control strategies can be used to adjust pressure control accuracy based on microstructural features.

[0358] It is understandable that this method can dynamically adjust the pressure based on real-time feedback to ensure processing quality.

[0359] For example, when a change in material hardness is detected, the system automatically increases pressure to maintain consistent cutting results. Using convolutional neural networks to iterate surface structure optimization schemes enables intelligent design.

[0360] In one embodiment, the surface structure is optimized by simulating the fluid dynamics of different structural forms. This method can improve processing efficiency while ensuring functionality.

[0361] like Figure 2 As shown, the second aspect of the present invention provides a system for accurately cutting and processing soft materials, which uses the above method to accurately cut and process soft materials. The system mainly includes:

[0362] Data acquisition module, used to obtain 3D model data and target cutting parameters of soft materials, including cutting trajectory, depth and surface texture information;

[0363] The path planning module is used to generate the initial cutting trajectory based on the 3D model data and target cutting parameters using an adaptive cutting path planning algorithm, taking into account material properties and cutting tool limitations;

[0364] The deformation monitoring module is used to monitor the material deformation during the cutting process in real time through multi-sensor fusion technology, and obtain the material surface morphology and internal stress distribution data;

[0365] Dynamic compensation module is used to adjust the cutting trajectory and depth in real time based on the monitored material deformation data using dynamic compensation algorithm to ensure cutting accuracy;

[0366] The trajectory tracking module is used to continuously track the position of the cutting tool using a high-precision laser interferometer measurement system to obtain the deviation value between the actual cutting trajectory and the target trajectory;

[0367] The closed-loop control module is used to make micron-level position corrections to the cutting tool based on the deviation value using a closed-loop feedback control algorithm to ensure accurate execution of the cutting trajectory.

[0368] The pressure regulation module is used to adjust the contact pressure between the cutting tool and the material through a programmable pressure control system to achieve precise processing of surface textures in different areas;

[0369] The texture generation module is used to create specific microstructures in areas where differentiated friction coefficients are required, combining pressure control and cutting depth adjustment with a multi-scale surface structure generation algorithm;

[0370] The quality inspection module is used to monitor the chemical composition changes of the cut surface using real-time spectral analysis technology to determine whether the preset surface property requirements are met. If not, it returns to adjust the cutting parameters and re-process the area.

[0371] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for precise cutting of soft materials, characterized in that: The method comprises: S1, obtains the 3D model data and target cutting parameters of the soft material, including cutting trajectory, depth and surface texture information; S2, based on the 3D model data and target cutting parameters, an adaptive cutting path planning algorithm is used to generate the initial cutting trajectory; S3 uses multi-sensor fusion technology to monitor material deformation during cutting in real time and obtain material surface morphology and internal stress distribution data; S4, based on the monitored material deformation data, uses dynamic compensation algorithm to adjust the cutting trajectory and depth in real time to ensure cutting accuracy; S5, uses a high-precision laser interferometry measurement system to continuously track the cutting tool position and obtain the deviation value between the actual cutting trajectory and the target trajectory; S6, based on the deviation value, uses a closed-loop feedback control algorithm to perform micron-level position correction on the cutting tool to ensure accurate execution of the cutting trajectory; S7, through the programmable pressure control system to adjust the contact pressure between the cutting tool and the material, to achieve precise processing of surface textures in different areas; S8, for areas requiring differentiated friction coefficients, uses a multi-scale surface structure generation algorithm to create specific microstructures by combining pressure control and cutting depth adjustment; S9 uses real-time spectral analysis technology to monitor the changes in the chemical composition of the cutting surface to determine whether the preset surface property requirements are met. If not, it returns to adjust the cutting parameters and re-process the area.

2. The method according to claim 1, characterized in that The step S1, obtaining the three-dimensional model data and target cutting parameters of the soft material, including cutting trajectory, depth and surface texture information, includes: Step S11, obtaining three-dimensional model data of the soft material, and obtaining complete model data through a three-dimensional scanner; Step S12, extracting target parameters from the complete model data, including geometric dimensions, material properties, and surface texture; Step S13, using a finite element analysis tool to mesh the model and determine initial values of the cutting trajectory, cutting depth, and surface texture; Step S14, generating a trajectory path for the initial value, calculating the deviation between the trajectory path and the target parameter, and if the deviation exceeds a first deviation threshold, performing iterative adjustment using a gradient descent method to obtain an optimized trajectory path; Step S15, calculating the depth value according to the optimized trajectory path, and using a cubic spline interpolation algorithm to determine the smoothed depth value; Step S16, extracting texture features by smoothing the depth value, calculating the difference between the texture features and the surface texture, and if the difference exceeds a difference threshold, adjusting the depth value until a matching texture feature is obtained; Step S17, after obtaining the matching texture features, the optimized trajectory path, the smoothed depth value and the matching texture features are integrated to obtain the final cutting parameters; Step S18: Generate cutting data of the three-dimensional model from the final cutting parameters, perform simulation verification using MATLAB, and determine the integrity of the cutting data.

3. The method according to claim 1, characterized in that The step S3, which uses multi-sensor fusion technology to monitor the material deformation during the cutting process in real time and obtain the material surface morphology and internal stress distribution data, includes: Step S31, collecting raw signals during the cutting process through a multi-sensor system to obtain deformation data and stress data; Step S32: integrating the collected original signals using weighted average fusion to determine surface morphological features and internal stress features; Step S33, if the surface morphological feature exceeds the deformation threshold, the deformation trend is determined by the least squares method; Step S34, adjusting the acquisition technology according to the deformation trend to obtain more accurate distribution information; Step S35, optimizing the distribution information through Kalman filtering to obtain smooth stress distribution data; Step S36: if the smoothed stress distribution data is inconsistent with the internal information, the deviation area is determined by the difference method; Step S37: Use the deviation area data to update the weighted average fusion parameters to obtain more accurate real-time acquisition results.

4. The method according to claim 3, characterized in that The method further includes: step S34, adjusting the acquisition technology according to the deformation trend to obtain more accurate distribution information, specifically including: Step S341, obtaining initial distribution data through dynamic sampling, where the sampling interval is determined according to a preset spatial grid density; Step S342: input the initial distribution data into a deformation trend calculation module based on moving average, and output the displacement change rate of each grid point; Step S343, adjusting the sampling frequency according to the displacement change rate, doubling the sampling frequency when the displacement change rate is higher than the displacement threshold, and generating optimized distribution data; Step S344, using OpenCV's KalmanFilter to smooth the optimized distribution data, setting the state variables as grid point coordinates and the observation matrix as the identity matrix to obtain a smoothed distribution result; Step S345 , comparing the smoothed distribution result with the surface morphology data acquired by the 3D scanner grid point by grid point, and marking as a deviation area when the coordinate difference exceeds the tolerance threshold; Step S346, using the Sobel operator to calculate the spatial gradient of the deviation area, and determining that the deviation is valid when the gradient amplitude is greater than the gradient threshold; Step S347, dynamically adjusting the weighted fusion coefficient according to the gradient amplitude, setting the weight of the optimized distribution data to the inverse of the gradient amplitude, and setting the weight of the surface morphology data to the normalized gradient amplitude, to generate the corrected distribution data; Step S348: Calculate the second-order derivative of the displacement change rate of each grid point based on the corrected distribution data. If the second-order derivative exceeds the derivative threshold for three consecutive frames, the area is determined to be the final feature distribution area.

5. The method according to claim 1, characterized in that The step S4 uses a dynamic compensation algorithm to adjust the cutting trajectory and depth in real time based on the monitored material deformation data to ensure cutting accuracy, including: Step S41, acquiring material deformation data through a sensor and obtaining deformation characteristics using signal processing technology; Step S42, applying a dynamic compensation algorithm to determine the trajectory adjustment amount and the depth adjustment amount according to the deformation characteristics; Step S43: if the adjustment amount exceeds the adjustment threshold, the cutting trajectory and cutting depth are updated through the control system; Step S44, using trajectory optimization technology to obtain a smooth path for the updated cutting trajectory; Step S45, adjusting the cutting depth through the depth control module to obtain a stable output; Step S46, obtaining the adjusted cutting data and determining whether it meets the requirements through the accuracy detection module; Step S47: If the accuracy test result is lower than the standard, the dynamic compensation algorithm is repeated to adjust the trajectory and depth.

6. The method according to claim 5, characterized in that The step S42, applying a dynamic compensation algorithm to determine the trajectory adjustment amount and the depth adjustment amount according to the deformation characteristics, includes: Step S421, collecting deformation features through sensors and extracting feature input from the collected data; Step S422, classifying the feature input using preset rules to obtain a classification result; Step S423, processing the classification results through Kalman filtering to determine trajectory data; Step S424: if the trajectory data exceeds the trajectory threshold, the trajectory data is updated by the control module; Step S425: Using cubic spline interpolation processing based on the updated trajectory data to obtain a smooth path; Step S426, adjusting the depth data through principal component analysis to obtain an adjustment result; Step S427: extract the output result from the adjustment result and determine whether it meets the standard through the detection module.

7. The method according to claim 1, characterized in that The step S5, using a high-precision laser interferometry system to continuously track the position of the cutting tool to obtain the deviation value between the actual cutting trajectory and the target trajectory, includes: Step S51, initializing the measurement system through laser interferometry technology to obtain the initial position information of the cutting tool, and passing the initial position information to the Kalman filter for continuous tracking processing to obtain real-time data; Step S52: passing the real-time data to the least square method for trajectory comparison, comparing the actual trajectory with the target trajectory, and determining the deviation value; Step S53: if the deviation value exceeds the second deviation threshold, updating the position information of the cutting tool by adjusting the tool state; Step S54: passing the real-time data to a time series analysis to analyze the changing trend of the actual trajectory and determine the stability of the trajectory comparison; Step S55, obtaining the fluctuation range of the change trend and determining the correction direction of the deviation value; Step S56: Update the target trajectory using the corrected deviation value to obtain an optimized cutting path.

8. The method according to claim 1, characterized in that The step S7, adjusting the contact pressure between the cutting tool and the material by a programmable pressure control system to achieve precise processing of surface textures in different areas, includes: Step S71, obtaining initial height data of the material surface by a contact profilometer; Step S72, using the K-means clustering algorithm to divide the height data into regions, and the number of clusters is automatically determined according to the contour curvature; Step S73 , generating a tool pressure parameter using a preset pressure-height mapping table according to the average height and roughness of each cluster area; Step S74, writing the pressure parameter into the register address of the PLC controller to control the servo motor to adjust the tool spindle pressure; Step S75, after the processing is performed, the surface height data is collected again using the profilometer, and the root mean square error of the height before and after the processing of each area is calculated; Step S76 , if the error in any region exceeds 5 microns, trigger the PID controller to dynamically adjust the gain coefficient of the pressure mapping table; Step S77 , repeatedly performing processing and measurement until the errors of all regions are lower than the error threshold.

9. The method according to claim 1, characterized in that The step S8, for the area requiring differentiated friction coefficient, uses a multi-scale surface structure generation algorithm to create a specific microstructure in combination with pressure control and cutting depth adjustment, including: Step S81, acquiring regional difference data through sensors, processing the sensor data using a multi-scale surface structure generation algorithm, and generating a friction coefficient distribution; Step S82, extracting a pressure control parameter from the friction coefficient distribution, and determining a cutting depth adjustment range using a first pressure threshold; Step S83, generating an initial surface structure model according to the cutting depth adjustment range, and extracting microstructure features in the model; Step S84, adjusting the pressure control accuracy according to the microstructure characteristics and determining the surface structure optimization scheme; Step S85, using a convolutional neural network to iterate the surface structure optimization scheme to obtain a specific structural morphology; Step S86, determining whether a second pressure threshold is met by calculating the generation efficiency of the specific structure form, and if not, adjusting the multi-scale method parameters; Step S87: generating final microstructure data according to the adjusted multi-scale method parameters.

10. A system for precise cutting of soft materials, characterized in that: The method according to any one of claims 1 to 9 is used to perform precise cutting processing on soft materials, wherein the system further comprises: Data acquisition module, used to obtain 3D model data and target cutting parameters of soft materials, including cutting trajectory, depth and surface texture information; The path planning module is used to generate the initial cutting trajectory based on the 3D model data and target cutting parameters using an adaptive cutting path planning algorithm, taking into account material properties and cutting tool limitations; The deformation monitoring module is used to monitor the material deformation during the cutting process in real time through multi-sensor fusion technology, and obtain the material surface morphology and internal stress distribution data; Dynamic compensation module is used to adjust the cutting trajectory and depth in real time based on the monitored material deformation data using dynamic compensation algorithm to ensure cutting accuracy; The trajectory tracking module is used to continuously track the position of the cutting tool using a high-precision laser interferometer measurement system to obtain the deviation value between the actual cutting trajectory and the target trajectory; The closed-loop control module is used to make micron-level position corrections to the cutting tool based on the deviation value using a closed-loop feedback control algorithm to ensure accurate execution of the cutting trajectory. The pressure regulation module is used to adjust the contact pressure between the cutting tool and the material through a programmable pressure control system to achieve precise processing of surface textures in different areas; The texture generation module is used to create specific microstructures in areas where differentiated friction coefficients are required, combining pressure control and cutting depth adjustment with a multi-scale surface structure generation algorithm; The quality inspection module is used to monitor the chemical composition changes of the cut surface using real-time spectral analysis technology to determine whether the preset surface property requirements are met. If not, it returns to adjust the cutting parameters and re-process the area.

Images