Intelligent dynamic adjustment method and system based on screen metal foil processing

Abstract

The application belongs to the technical field of machine control, and particularly relates to a method and system for intelligent dynamic adjustment of an edge based on a screen metal foil processing. The method comprises: acquiring a smooth pose deviation sequence and a normal vector direction; calculating a perpendicular angle change amount between a tool head axis and the normal vector according to the smooth pose deviation sequence and the normal vector direction; acquiring a cutting depth difference value between an actual cutting depth value and a theoretical cutting depth value according to the perpendicular angle change amount and a parameter value of the metal foil; in response to the cutting depth difference value being greater than a preset difference threshold, generating an error accumulation feature vector; performing classification processing on the error accumulation feature vector to acquire a predicted metal foil edge defect category and a corresponding spatial distribution position; and outputting a control signal of a tool head rotation axis according to the perpendicular angle change amount, the edge defect category and the corresponding spatial distribution position. The application realizes real-time compensation of a pose deviation through machine intelligence to inhibit edge defects of a screen metal foil.

Description

Technical Field

[0001] This application belongs to the field of machine control technology, specifically relating to a method and system for intelligent dynamic edge adjustment based on the processing of metal foil with intensifying screen. Background Technology

[0002] As a core component of important equipment such as medical imaging diagnostics, the edge processing quality of the intensifying screen's metal foil substrate directly affects image clarity and diagnostic accuracy. Metal foil edge processing typically employs a six-degree-of-freedom industrial robot carrying a tool head to cut and shape the foil along a predetermined trajectory.

[0003] During the edge processing of metal foil, to ensure the continuity of the edge contour and surface quality, the tool head axis must always remain perpendicular to the metal foil processing surface. However, metal foil materials generally have objective defects such as uneven thickness and surface undulations, making it difficult to maintain the perpendicularity between the tool head and the metal foil surface.

[0004] Verticality deviation directly changes the actual cutting depth of the tool head, which can easily lead to quality problems such as excessive or insufficient cutting at the edge of the metal foil. In severe cases, it can cause surface defects such as burrs, collapsed edges, or tears, affecting the assembly sealing and service life of the intensifying screen.

[0005] Therefore, how to ensure the consistency of processing quality during the edge processing of metal foil is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the above issues, this application provides a method and system for intelligent dynamic edge adjustment in metal foil processing based on intensified sensing screens. By compensating for robot pose deviations in real time, edge defects in metal foil edge processing are suppressed.

[0007] To achieve the above objectives, this application provides a method for intelligent dynamic edge adjustment based on the processing of metal foil for intensifying screens, comprising: Obtain a smooth pose deviation sequence and the direction of the normal vector between the tool head and the metal foil surface. The smooth pose deviation sequence is used to represent the relative pose relationship between the tool head and the metal foil during the metal foil processing. Based on the spatial rotation matrix components in the smoothed pose deviation sequence and the direction of the normal vector between the tool head and the metal foil surface, calculate the change in the vertical angle between the tool head axis and the normal vector. Based on the change in vertical angle and the parameter value of the metal foil, the difference in cutting depth between the actual cutting depth of the tool head into the metal foil and the theoretical cutting depth is obtained. In response to the cutting depth difference value being greater than a preset difference threshold, the spatial translation coordinate components in the smoothed pose deviation sequence are accumulated to generate an error accumulation feature vector, which is used to represent the edge deformation trend of the metal foil. The error accumulation feature vector is classified to obtain the predicted metal foil edge defect category and its corresponding spatial distribution location; Based on the vertical angle change, the edge defect type, and the corresponding spatial distribution location, a control signal for the tool head rotation axis is output.

[0008] On the one hand, this application provides an intelligent dynamic edge adjustment system based on the processing of metal foil for intensifying screens, including: The pose acquisition module is used to acquire a smooth pose deviation sequence, which represents the relative pose relationship between the tool head and the metal foil during the metal foil processing. The normal vector acquisition module is used to obtain the normal vector direction between the tool head and the metal foil surface; The vertical angle calculation module is used to calculate the change in vertical angle between the tool head axis and the normal vector based on the spatial rotation matrix components in the smooth pose deviation sequence and the normal vector direction. The cutting depth difference acquisition module is used to acquire the cutting depth difference between the actual cutting depth value and the theoretical cutting depth value of the tool head cutting into the metal foil based on the vertical angle change and the parameter value of the metal foil. The error accumulation feature generation module is used to accumulate the spatial translation coordinate components in the smooth pose deviation sequence in response to the cutting depth difference value being greater than a preset difference threshold, and generate an error accumulation feature vector. The error accumulation feature vector is used to represent the edge deformation trend of the metal foil. The defect prediction module is used to classify the error accumulation feature vector and predict the edge defect category and corresponding spatial distribution location of the metal foil. The control signal output module is used to output a control signal for the tool head rotation axis based on the vertical angle change, the edge defect type, and the corresponding spatial distribution position.

[0009] The embodiments of this application have the following beneficial effects: In the embodiments of this application, by acquiring the smooth pose deviation sequence and the normal vector direction between the tool head and the metal foil surface, the relative pose relationship between the tool head and the metal foil is perceived in real time during the processing. Compared with the open-loop control mode in the prior art that only relies on theoretical kinematic models for trajectory planning and lacks online detection methods, this provides a real-time and reliable data foundation for subsequent dynamic compensation.

[0010] Simultaneously, by extracting the spatial rotation matrix deviation component from the smoothed pose deviation sequence and calculating the vertical angle change in conjunction with the surface normal vector direction of the metal foil, the tool head attitude deviation is quantified as the angular offset between the tool head axis and the normal vector. Compared to existing machining methods that use fixed attitude parameters for trajectory planning and cannot respond to changes in the surface normal vector direction, this establishes a clear physical correspondence for the assessment of cutting depth deviation. Furthermore, by mapping the vertical angle change to the actual cutting depth and subtracting it from the theoretical cutting depth to obtain the cutting depth difference value, a quantitative correlation between attitude deviation and machining quality is established. Compared to the passive quality control method in existing technologies that only detects defects through offline inspection after machining, this makes the cutting depth deviation a feedback signal that can be monitored online.

[0011] Building upon this foundation, by accumulating spatial translational coordinate deviation components to generate an error accumulation feature vector when the difference in cutting depth exceeds a preset threshold, the minute positional deviations of each trajectory point are transformed into macroscopic contour deformation features of the entire edge. Compared to the open-loop machining mode in existing technologies that cannot acquire edge contour deformation trends online, this approach allows the changing trends of edge quality to be perceived in real time during machining. Subsequently, a support vector machine algorithm is used to classify the error accumulation feature vector, mapping the contour deformation trend to specific edge defect categories such as burrs, edge collapse, and tears, along with their spatial distribution. This enables the defect prediction results to be directly used to guide compensation decisions, demonstrating the application of machine intelligence in defect identification and classification.

[0012] Finally, the embodiments of this application output control signals for each rotation axis by comprehensively considering the vertical angle change, edge defect type, and spatial distribution location. This enables the compensation action to be adjusted differently according to different defect types and specific segments where the defects are located. This achieves a complete technical closed loop from pose deviation perception, cutting quality assessment, defect type prediction to segment-specific compensation, effectively suppressing edge defects in the edge processing of intensifying screen metal foil and improving the consistency of processing quality. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating the intelligent dynamic edge adjustment method for metal foil processing based on intensifying screen provided in the first embodiment of this application; Figure 2 This is a flowchart illustrating the intelligent dynamic edge adjustment method for metal foil processing based on intensifying screen provided in the second embodiment of this application; Figure 3 This is a flowchart illustrating the intelligent dynamic edge adjustment method for metal foil processing based on intensifying screen provided in the third embodiment of this application; Figure 4 This is a flowchart illustrating the intelligent dynamic edge adjustment method based on metal foil processing for intensifying screens provided in the fourth embodiment of this application. Figure 5 This is a flowchart illustrating the intelligent dynamic edge adjustment method for metal foil processing based on intensifying screen provided in the fifth embodiment of this application; Figure 6 This is a flowchart illustrating the intelligent dynamic edge adjustment method for metal foil processing based on intensifying screen provided in the sixth embodiment of this application; Figure 7 This is a flowchart illustrating the intelligent dynamic edge adjustment method based on metal foil processing for intensifying screens provided in the seventh embodiment of this application. Figure 8 This is a flowchart illustrating the intelligent dynamic edge adjustment method based on metal foil processing for intensifying screens provided in the eighth embodiment of this application; Figure 9 This is a flowchart illustrating the intelligent dynamic edge adjustment method for metal foil processing based on intensifying screen provided in the ninth embodiment of this application; Figure 10 This is a schematic diagram of a module for an intelligent dynamic edge adjustment system based on intensifying screen metal foil processing provided in this application embodiment. Detailed Implementation

[0014] To enable those skilled in the art to better understand the technical solution, the present application will be described in detail below with reference to the embodiments. The description in this section is only exemplary and explanatory, and should not be used to limit the scope of protection of the present application in any way.

[0015] Edge processing of metal foil (such as aluminum foil) for intensifying screens typically employs a six-degree-of-freedom industrial robot carrying a tool head to cut and shape the foil along a predetermined trajectory. During processing, the tool head axis must remain perpendicular to the metal foil surface at all times to ensure the quality of the edge contour.

[0016] In related technologies, the machining trajectory uses preset fixed posture parameters, which cannot respond to the uneven thickness and surface undulations of the metal foil material itself. When the tool head moves along the edge of the three-dimensional curved surface, the direction of the surface normal vector changes continuously. The fixed posture trajectory cannot adjust the tool head pointing in real time to follow the change of the normal vector. In areas where the curvature of the surface changes greatly, the problem of the tool head axis deviating from the surface normal vector direction is particularly prominent.

[0017] Meanwhile, existing trajectory planning methods are based on theoretical kinematic models. However, factors such as transmission errors, link elastic deformation, and load disturbances can occur at various robot joints, causing deviations between the actual spatial pose reached by the end effector and the theoretical set value. These deviations accumulate continuously during the processing and lack online detection and real-time compensation methods.

[0018] Furthermore, in related technologies, quality control methods are mostly offline inspections after processing, which cannot intervene in positional deviations during processing. When edge defects are detected, the workpiece is already scrapped, and can only be passively remedied by adjusting the process parameters of subsequent batches. This makes it difficult to guarantee the consistency of each product, and the scrap rate increases significantly when dealing with batches with poor material consistency.

[0019] Therefore, surface quality problems such as burrs, collapsed edges, or tears are prone to occur at the edges of metal foil in curved transition areas and locations where material thickness changes abruptly, affecting the assembly sealing and service life of the intensifying screen.

[0020] To address the aforementioned technical problems, this application provides one or more embodiments and combinations thereof to offer an intelligent dynamic edge adjustment method for metal foil processing based on an intensifying screen. By acquiring the deviation between the actual pose data of the robot end effector and the theoretical set value, the deviation is denoised and then the change in the vertical angle between the tool head and the metal foil surface is calculated. Based on this change, the difference between the actual cutting depth and the theoretical value is evaluated, and when the difference exceeds the limit, a corresponding compensation control signal is generated and output to each rotation axis of the robot. This enables real-time dynamic adjustment of the pose deviation during processing, thereby suppressing edge defects in metal foil edge processing.

[0021] In one embodiment, see Figure 1 The intelligent dynamic edge adjustment method based on the metal foil processing of intensifying screens includes at least the following steps: Step S100: Obtain the smooth pose deviation sequence and the normal vector direction of the tool head and the metal foil surface. The smooth pose deviation sequence is used to represent the relative pose relationship between the tool head and the metal foil during the metal foil processing.

[0022] Step S120: Calculate the change in the vertical angle between the tool head axis and the normal vector based on the spatial rotation matrix components in the smoothed pose deviation sequence and the direction of the normal vector between the tool head and the metal foil surface.

[0023] Step S140: Based on the change in vertical angle and the parameter value of the metal foil, obtain the difference in cutting depth between the actual cutting depth of the tool head cutting into the metal foil and the theoretical cutting depth.

[0024] Step S160: In response to the cutting depth difference value being greater than a preset difference threshold, the spatial translation coordinate components in the smoothed pose deviation sequence are accumulated to generate an error accumulation feature vector, which is used to represent the edge deformation trend of the metal foil.

[0025] Step S180: Classify the cumulative error feature vector to obtain the predicted metal foil edge defect category and its corresponding spatial distribution location.

[0026] Step S200: Output the control signal for the tool head rotation axis based on the vertical angle change, edge defect type, and corresponding spatial distribution location.

[0027] In step S100, the tool head refers to the motion component mounted on the robot's end effector that directly acts on the edge of the intensifying screen's metal foil to perform cutting processing. The metal foil refers to the thin metal sheet (e.g., aluminum foil) that constitutes the substrate of the intensifying screen and is the object being processed in the edge processing process. The tool head is carried by the robot and moves along the edge trajectory of the metal foil. The spatial pose of the tool head determines the position of the cutting point relative to the surface of the metal foil and the cutting angle. The relative pose relationship between the two directly determines the edge processing quality.

[0028] A smooth pose deviation sequence refers to the set of deviation values ​​between the actual spatial pose and the theoretically set pose of the robot end effector head during the processing of metal foil edges. The smooth pose deviation sequence can include spatial translation coordinate deviation components and spatial rotation matrix deviation components. The spatial translation coordinate deviation components are the offsets of the actual position of the tool head relative to its theoretical position along the X, Y, and Z axes in the base coordinate system. The spatial rotation matrix deviation components can be the rotational deviations of the tool head's actual posture relative to its theoretical posture, recorded in the form of a 3×3 rotation matrix.

[0029] The direction of the normal vector between the tool head and the metal foil surface refers to the theoretical vertical orientation of the metal foil at the current processing point. It is extracted by querying the neighborhood point set from the three-dimensional point cloud data of the metal foil surface and fitting the local plane equation through the current spatial position coordinates of the tool head. It is used to compare with the actual posture of the tool head to calculate the degree of posture deviation.

[0030] In one specific implementation, a binocular vision sensor captures the optical target on the robot's end effector in real time at a sampling frequency of 100 Hz to obtain the actual spatial pose data at the current moment. An example set of values ​​is used for illustration: the actual spatial translation coordinates are (150mm, 200mm, 50mm), and the theoretical spatial translation coordinates are (150mm, 202mm, 49mm); the actual attitude Euler angles are (10°, 20°, 30°), and the theoretical attitude Euler angles are (10°, 19°, 31°). Substituting the Euler angles of the actual and theoretical attitudes into trigonometric function matrix operations, actual and theoretical spatial rotation matrices are constructed. By multiplying the inverse of the actual and theoretical rotation matrices and using the Rodrigues formula to convert the resulting matrix into an equivalent rotation vector, the spatial rotation angle deviation is calculated to be 17 milliradians. Subtracting the theoretical translation coordinates from the actual translation coordinates yields the spatial translation coordinate difference vector as (0mm, -2mm, 1mm). The above values ​​are for illustrative purposes only and do not constitute any limitation on the scope of protection of this application.

[0031] In step S120, the spatial rotation matrix deviation component is extracted from the smoothed pose deviation sequence. This spatial rotation matrix deviation component records the rotation deviation of the actual tool head posture relative to the theoretical posture, and is represented in the form of a 3×3 rotation matrix. The tool head axis refers to the direction in which the center of the tool head points in the base coordinate system. In the initial state, the tool head axis is aligned with the Z-axis direction of the base coordinate system. This tool head axis is represented by a unit vector. Represented. The spatial rotation matrix deviation component R is compared with the initial axis unit vector of the tool head. Multiplying them together yields the actual attitude vector of the tool head in the base coordinate system. .

[0032] The actual attitude vector of the tool head in the base coordinate system With respect to the direction of the normal vector of the metal foil surface Perform a vector dot product operation to obtain the cosine of the angle between the tool head axis and the normal vector. Then through the inverse cosine function Calculate the angular offset between the tool head axis and the normal vector. This angular offset is the change in vertical angle, which reflects the degree to which the tool head's current orientation deviates from the vertical orientation of the metal foil surface. As an example, let's illustrate this with a set of values: assume the spatial rotation matrix deviation component R and the initial axis unit vector... The actual attitude vector is obtained after multiplication. The direction of the normal vector of the metal foil surface Then the cosine value of the included angle angular offset The radian value, converted to an angle of approximately 7.2°, indicates that the current orientation of the tool head deviates from the vertical direction of the metal foil surface by approximately 7.2 degrees. The above values ​​are for illustrative purposes only and do not constitute any limitation on the scope of protection of this application.

[0033] In step S140, the actual cutting depth of the tool head into the metal foil is calculated based on the mapping relationship between the change in vertical angle and the parameter values ​​of the metal foil. The magnitude of the change in vertical angle directly affects the contact state between the tool head and the surface of the metal foil. When the axis of the tool head deviates from the normal vector direction, the contact point between the tool head cutting edge and the surface of the metal foil changes, and the original cutting action perpendicular to the surface becomes an inclined cutting action, resulting in a deviation between the actual amount of material removed and the theoretical design value.

[0034] The larger the deviation angle, the greater the horizontal offset of the actual contact point, and the more significant the difference between the actual path swept by the cutting edge and the theoretical path. Consequently, the actual cutting depth deviates from the theoretical value. If this deviation occurs in areas with thin metal foil, it may cause the tool head to penetrate the foil; if it occurs in areas with thicker foil, it may result in insufficient cutting and incomplete edge shaping. The theoretical cutting depth value is obtained, and the difference between the actual cutting depth value and the theoretical cutting depth value is used to quantify the degree of cutting deviation at the current machining point caused by the tool head posture deviation.

[0035] In one specific implementation, a vision sensor collects real-time posture data of the cutting tool during operation, extracting a vertical angle change of 3 degrees. A built-in quadratic polynomial mapping algorithm is invoked, which specifies that the thickness of the metal foil equals the square of the vertical angle change multiplied by 2 plus the vertical angle change multiplied by 5. Based on this, the actual thickness of the metal foil at the current machining position is calculated to be 33 micrometers. A depth calculation program is initiated based on the tool geometry and material deformation coefficient. A preset 5-micrometer elastic compression compensation is added to the 33-micrometer metal foil thickness to calculate the actual cutting depth of the tool head into the metal foil as 38 micrometers. The theoretical cutting depth value corresponding to the current batch of material is retrieved from the cloud-based process database; this value is 40 micrometers. The difference between the theoretical cutting depth value of 40 micrometers and the actual cutting depth value of 38 micrometers is calculated to obtain a cutting depth difference value of 2 micrometers. The above values ​​are for illustrative purposes only and do not constitute any limitation on the scope of protection of this application.

[0036] In step S160, when the difference in cutting depth is greater than the preset difference threshold, it indicates that the cutting deviation of the current processing point has exceeded the allowable range. If the deviation continues, it will destroy the continuity of the edge contour and the surface quality, resulting in local over-cutting or under-cutting of the metal foil edge, which in turn causes edge defects.

[0037] At this point, the spatial translation coordinate deviation components in the smoothed pose deviation sequence are accumulated along the edge processing path to generate an error accumulation feature vector. The spatial translation coordinate deviation components in the smoothed pose deviation sequence record the offset of the tool head's actual position relative to its theoretical position at each trajectory point. Accumulating these components along the path reflects the overall distribution and cumulative effect of the deviation across the entire edge. The error accumulation feature vector transforms the minute deviations at each trajectory point into macroscopic contour features of the entire edge, characterizing the overall deformation trend of the entire edge of the metal foil in three-dimensional space.

[0038] In one specific implementation, the control unit calculates the difference between the actual cutting depth and the theoretical cutting depth of the tool tip in real time. If the difference is found to be 15 micrometers, it is compared with a pre-existing surface roughness threshold of 10 micrometers in a database. This comparison determines that the current machining state has triggered a risk of surface quality degradation, thus initiating a deviation extraction mechanism. A matrix slicing algorithm is used to remove rotational attitude components from the smoothed pose deviation sequence, extracting spatial translation coordinate components in three dimensions: horizontal, vertical, and lateral. For example, the translation deviation at a sampling point might be 2 micrometers horizontally, 3 micrometers vertically, and 1 micrometer vertically. A discrete integration algorithm is then used to perform point-by-point vector addition on the extracted spatial translation coordinate components along an edge machining path containing 500 interpolation nodes, generating an error accumulation feature vector. The above values ​​are for illustrative purposes only and do not constitute any limitation on the scope of protection of this application.

[0039] In step S180, the error accumulation feature vector is classified. Different edge defects correspond to different deviation accumulation patterns. Burrs usually correspond to a peak pattern where the deviation suddenly increases and then recovers quickly in a local area, that is, the error accumulation feature vector shows a significant bulge in a certain section; collapsed edges correspond to a monotonic trend pattern where the deviation continuously shifts in the same direction along a long path, that is, the error accumulation feature vector continuously increases or decreases along the path; tearing corresponds to a sudden change pattern where the deviation changes discontinuously, that is, the error accumulation feature vector shows a step change at a certain point.

[0040] Based on the above correspondence, by classifying the error accumulation feature vector, the contour deformation trend represented by the error accumulation feature vector can be mapped to specific edge defect categories. Simultaneously, the spatial distribution of each type of defect along the processing path is output, thereby locating the specific region of each defect on the edge. Through classification processing, multiple defect types that may coexist on the entire edge and their respective segment locations are output together, providing a basis for subsequent segmented differential compensation.

[0041] In one specific implementation, a 256-dimensional error cumulative feature vector, acquired by a front-end vision sensor and after dimensionality reduction, is input into a pre-trained support vector machine model. This model uses a radial basis function kernel to map low-dimensional linearly inseparable features to a high-dimensional space. A penalty parameter of 100 and a kernel function parameter of 5 are set, and the optimal classification surface is solved by calculating the geometric distance between the input vector and the hyperplane. During the classification decision stage, when the gradient abruptness frequency of the feature vector in a local region exceeds 50 times per millimeter, the model classifies it as a burr morphology; when the edge thickness decay gradient is below a preset threshold for 30 consecutive pixels, it is identified as a collapsed edge morphology; when the contour exhibits irregular jagged edges and the fracture depth exceeds 20 pixels, it is classified as a tear morphology. These three microscopic morphological features are mapped to a macroscopic edge defect type label library, and the corresponding edge defect prediction category code is output. The above values ​​are for illustrative purposes only and do not constitute any limitation on the scope of protection of this application.

[0042] In step S200, the change in vertical angle reflects the degree to which the current tool head posture deviates from the vertical orientation of the metal foil surface, the edge defect category reflects the type of defect that needs to be corrected, and the spatial distribution location reflects the specific segment on the machining path where the defect appears. The change in vertical angle, the edge defect category, and the corresponding spatial distribution location of the edge defect are combined to generate control signals and output to each rotation axis of the robot. Different edge defect types require different tool head posture adjustments. Burr defects typically require reducing the depth of cut and adjusting the tool head pitch angle; edge collapse defects typically require increasing the depth of cut and adjusting the tool head yaw angle; and tearing defects typically require slowing down the feed rate and adjusting the overall tool head posture.

[0043] The spatial distribution determines the starting point and duration of the adjustment action on the machining path, ensuring that the compensation action is performed only in the defective section, avoiding interference with the normal machining of defect-free sections. Driving each rotary axis to perform corresponding attitude adjustment actions compensates for tool head posture deviations, thus suppressing edge defects in metal foil edge processing.

[0044] In one specific implementation, the system receives an edge defect prediction from a visual inspection model, classifying it as a depth overcut defect. Simultaneously, it reads a vertical angle change of +3 degrees from a tilt sensor. These two parameters are used as a joint primary key to retrieve a pre-established inter-axis coupling relationship table from a memory database. The Pearson correlation coefficient algorithm is used to analyze the contribution of each axis to the current overcut defect, calculating the coupling weight of the first rotation axis as 75%, the second rotation axis as 20%, and the third rotation axis as 5%. Based on the descending order of these weights, compensation priorities are assigned to each rotation axis, with the first rotation axis designated as the first-priority compensation object. The Jacobian matrix iteration algorithm is invoked, using the target pose matrix with a +3 degree deviation as input. Inverse kinematics calculation is performed with the first-priority rotation axis as the primary variable. After five matrix inversion iterations, it is calculated that the first rotation axis requires a reverse compensation of 2 degrees, and the second rotation axis requires a forward compensation of 0.5 degrees. The joint correction compensation is then converted into pulse signals and sent to the servo driver. The above values ​​are for illustrative purposes only and do not constitute any limitation on the scope of protection of this application.

[0045] In this embodiment, by acquiring the smooth pose deviation sequence and the normal vector direction between the tool head and the metal foil surface, the relative pose relationship between the tool head and the metal foil is perceived in real time during the processing. Compared with the open-loop control mode in the prior art that only relies on theoretical kinematic models for trajectory planning and lacks online detection methods, this provides a real-time and reliable data foundation for subsequent dynamic compensation.

[0046] Simultaneously, by extracting the spatial rotation matrix deviation component from the smoothed pose deviation sequence and calculating the vertical angle change in conjunction with the surface normal vector direction of the metal foil, the tool head attitude deviation is quantified as the angular offset between the tool head axis and the normal vector. Compared to existing machining methods that use fixed attitude parameters for trajectory planning and cannot respond to changes in the surface normal vector direction, this establishes a clear physical correspondence for the assessment of cutting depth deviation. Furthermore, by mapping the vertical angle change to the actual cutting depth and subtracting it from the theoretical cutting depth to obtain the cutting depth difference value, a quantitative correlation between attitude deviation and machining quality is established. Compared to the passive quality control method in existing technologies that only detects defects through offline inspection after machining, this makes the cutting depth deviation a feedback signal that can be monitored online.

[0047] Building upon this foundation, by accumulating spatial translational coordinate deviation components to generate an error accumulation feature vector when the difference in cutting depth exceeds a preset threshold, the minute positional deviations of each trajectory point are transformed into macroscopic contour deformation features of the entire edge. Compared to the open-loop machining mode in existing technologies that cannot acquire edge contour deformation trends online, this approach allows the changing trends of edge quality to be perceived in real time during machining. Subsequently, a support vector machine algorithm is used to classify the error accumulation feature vector, mapping the contour deformation trend to specific edge defect categories such as burrs, edge collapse, and tears, along with their spatial distribution. This enables the defect prediction results to be directly used to guide compensation decisions, demonstrating the application of machine intelligence in defect identification and classification.

[0048] Finally, this embodiment outputs control signals for each rotation axis by comprehensively considering the vertical angle change, edge defect type, and spatial distribution location. This enables the compensation action to be adjusted differently according to different defect types and specific segments where the defects are located. This achieves a complete technical closed loop from pose deviation perception, cutting quality assessment, defect type prediction to segment-specific compensation, effectively suppressing edge defects in the edge processing of intensifying screen metal foil and improving the consistency of processing quality.

[0049] In one embodiment, see Figure 2 Step S100 includes: Step S101: Obtain the actual pose data of the tool head during the processing of the metal foil edge and the preset theoretical setting data.

[0050] Step S102: Compare the differences between the actual pose data and the theoretical set data in the spatial rotation matrix and spatial translation coordinates to obtain the initial pose deviation sequence.

[0051] Step S103: Perform state estimation and noise filtering on the attitude deviation frequency components in the initial pose deviation sequence to obtain a smooth pose deviation sequence.

[0052] In step S101, the actual spatial pose data of the tool head during the processing of the metal foil edge is acquired in real time by a vision sensor. Simultaneously, theoretical pose data at the same moment is read from a preset 3D design model, thus achieving real-time perception of the tool head's actual pose. Compared to the open-loop control method in existing technologies that relies solely on theoretical kinematic models for trajectory planning and lacks actual pose feedback, this embodiment provides a reliable data source for subsequent pose deviation detection and compensation.

[0053] In step S102, the differences between the actual pose data and the theoretically set data are compared in two dimensions: spatial translation coordinates and spatial rotation matrix. The actual spatial translation coordinates are subtracted from the theoretical spatial translation coordinates dimension by dimension to obtain the spatial translation coordinate deviation component; the actual spatial rotation matrix is ​​compared with the theoretical spatial rotation matrix to obtain the spatial rotation matrix deviation component. The translation deviation component and rotation matrix deviation component at each sampling time are concatenated in chronological order to form the initial pose deviation sequence. Compared to existing technologies that do not consider pose deviations caused by factors such as transmission errors, link deformation, and load disturbances during actual robot operation, this embodiment quantifies the differences in two dimensions, transforming the pose deviation from a qualitative description into a processable quantitative data sequence.

[0054] In step S103, a Kalman filter algorithm is used to perform state estimation and noise filtering on the attitude deviation frequency components in the initial pose deviation sequence, filtering out high-frequency mechanical vibration noise and random interference components, and outputting a smooth pose deviation sequence. This smooth pose deviation sequence includes spatial translation coordinate deviation components and spatial rotation matrix deviation components. The former represents the offset of the actual tool head position relative to its theoretical position along the X, Y, and Z axes in the base coordinate system, while the latter represents the rotational deviation of the actual tool head posture relative to its theoretical posture, recorded in the form of a 3×3 rotation matrix. Compared to the lack of effective noise reduction methods for sensor data in existing technologies, this embodiment uses Kalman filtering to separate the effective signal from noise, providing an accurate data foundation for subsequent vertical angle calculations.

[0055] For example, the state equation for a Kalman filter is: The observation equation is The optimal estimate of the attitude deviation is gradually approximated through two iterative steps: prediction and update. Specifically, the Kalman filter is used to filter out high-frequency mechanical vibration noise and random interference components at a frequency of 50Hz from the initial attitude deviation sequence. The gain calculation formula for the Kalman filter is as follows: ,in Let R be the prior error covariance matrix, and R be the observation noise covariance matrix. The smoothed pose deviation sequence output after filtering contains the effective pose deviation signal, providing an accurate data basis for subsequent vertical angle calculation.

[0056] In this embodiment, the actual pose of the tool head is acquired in real time by a vision sensor and compared with the theoretical value, realizing the transformation of pose deviation from qualitative to quantitative. This overcomes the technical limitations of existing technologies that rely solely on theoretical models for trajectory planning and cannot perceive actual deviations. Through Kalman filtering noise reduction, the state estimation algorithm in machine intelligence is used to optimally estimate the frequency components of the pose deviation, effectively eliminating high-frequency vibrations and random interference in the original pose data. This ensures that the deviation data meets the reliability requirements for subsequent accuracy compensation, avoiding miscompensation or overcompensation caused by noise interference, and laying a data foundation for improving the edge processing quality of the intensified screen metal foil.

[0057] In one embodiment, see Figure 3 Step S100 includes: Step S104: Obtain the three-dimensional point cloud coordinates of the metal foil surface.

[0058] Step S105: Fit the local plane equation of the three-dimensional point cloud coordinates using the least squares method.

[0059] Step S106: Extract the direction of the normal vector between the tool head and the metal foil surface from the local plane equation.

[0060] In step S104, for example, the three-dimensional point cloud coordinates of the metal foil surface can be acquired using a structured light scanner. The three-dimensional point cloud coordinates are a collection of discrete position information of the metal foil surface in space. Each point records the actual spatial position of the metal foil surface in the base coordinate system in the form of three-dimensional coordinates, which are used for subsequent surface shape fitting and normal vector extraction.

[0061] In step S105, based on the current spatial coordinates of the tool head, the set of three-dimensional points in the neighborhood of that location can be queried from the three-dimensional point cloud data, and the local plane equation can be fitted using the least squares method. The least squares method determines the optimal coefficients a, b, c, and d of the plane equation by minimizing the sum of the squares of the perpendicular distances from all points to the fitted plane. This minimizes the overall deviation between the fitted plane and the actual point cloud, thus ensuring that the fitted plane equation can accurately reflect the actual spatial orientation of the metal foil surface at the current processing point.

[0062] In step S106, the normal vector can be extracted from the fitted local plane equation. This serves as the direction of the normal vector between the tool head and the metal foil surface. This normal vector direction characterizes the actual surface orientation of the metal foil at the current processing point and is used to compare with the actual orientation of the tool head to calculate the degree to which the tool head orientation deviates from the perpendicular orientation of the metal foil surface. Compared to directly reading the theoretical normal vector from the CAD model, this embodiment obtains the normal vector through actual measurement, which can truly reflect the actual changes in the normal vector caused by factors such as uneven thickness and surface undulations of the metal foil material itself.

[0063] In this embodiment, a structured light scanner is used to collect three-dimensional point clouds of the metal foil surface online and fit the actual plane equation. Compared with the existing technology that directly reads the theoretical normal vector from the CAD model without considering the actual surface undulations and thickness unevenness of the metal foil material, the actual surface normal vector of the current processing point is obtained instead of the theoretical value. This eliminates the deviation between the theoretical and actual values ​​of the normal vector caused by the difference in material consistency, provides an accurate reference direction for calculating the change in vertical angle, avoids the vertical angle calculation error caused by the deviation of the normal vector reference, and thus improves the accuracy of attitude deviation detection.

[0064] In one embodiment, the parameter values ​​for the metal foil include a thickness value. As an example, the thickness value characterizes the material thickness of the metal foil at various locations along the processing path and is read from a process database.

[0065] Please see Figure 4 Step S140 includes: Step S141: Calculate the actual cutting depth of the tool head into the metal foil based on the mapping relationship between the vertical angle change and the metal foil thickness.

[0066] Step S142: Obtain the preset theoretical cutting depth value.

[0067] Step S143: Calculate the cutting depth difference value based on the difference between the actual cutting depth value and the theoretical cutting depth value.

[0068] In step S141, the magnitude of the change in vertical angle directly affects the contact angle between the tool head and the metal foil surface. The greater the change in vertical angle, the more serious the deviation of the tool head axis from the normal vector direction of the metal foil surface, and the more the actual depth of the tool head cutting into the metal foil changes accordingly.

[0069] Based on the change in vertical angle With metal foil thickness value Mapping relationship between Calculate the actual metal foil thickness at the current processing point. Then, calculate the actual metal foil thickness. Substitute into the cutting depth calculation formula ,in The actual cutting depth of the tool head into the metal foil is obtained by setting a preset elastic compression compensation amount. .

[0070] The mapping relationship between the change in vertical angle and the thickness value describes the quantitative impact of attitude deviation on the actual thickness of the metal foil. The elastic downward pressure compensation is used to compensate for the impact of the elastic deformation of the tool head caused by the cutting force during the cutting process on the actual cutting depth.

[0071] In step S142, the theoretical cutting depth value is read from the process database. It is the target cutting depth value pre-set at the current machining point according to the metal foil design thickness and process requirements, representing the ideal depth of the tool head cutting into the metal foil under the condition of no positional deviation.

[0072] In step S143, the actual cutting depth value can be used as a reference. Compared with theoretical cutting depth value The difference between them is used to calculate the depth of cut difference value. The depth of cut difference value is used to quantify the degree of deviation in cutting amount caused by the tool head posture deviation at the current machining point. The positive or negative value of the depth of cut difference value reflects whether the actual cutting depth is too deep or insufficient relative to the theoretical cutting depth, providing a quantitative criterion for whether to trigger compensation operations in the future.

[0073] In this embodiment, the actual cutting depth is calculated by mapping the change in vertical angle to the thickness of the metal foil, and the difference between this and the theoretical cutting depth is used to obtain the cutting depth difference value, thus establishing a quantitative correlation between tool head posture deviation and machining quality. Compared to the passive quality control method in the prior art that only detects defects through offline detection after machining, this embodiment quantifies the impact of posture deviation on cutting depth in real time during machining, making the cutting depth deviation a feedback signal that can be monitored online. This provides a direct basis for subsequent compensation decisions and avoids the problem of defects accumulating to an unacceptable level due to the inability to know the cutting depth deviation online.

[0074] Meanwhile, this embodiment compensates for the elastic deformation caused by the cutting force by introducing an elastic compression compensation amount, making the calculated result of the actual cutting depth closer to the true value, improving the calculation accuracy of the cutting depth difference value, and thus improving the accuracy of the compensation decision.

[0075] In one embodiment, see Figure 5 Step S160 includes: Step S161: Extract the spatial translation coordinate deviation components of each trajectory point in the edge processing path from the smooth pose deviation sequence.

[0076] Step S162: Accumulate the spatial translation coordinate deviation components of each trajectory point to obtain the cumulative error feature vector.

[0077] In step S161, the spatial translation coordinate deviation component of each trajectory point in the edge machining path is extracted from the smoothed pose deviation sequence. The spatial translation coordinate deviation component in the smoothed pose deviation sequence records the offset of the actual position of the tool head relative to the theoretical position at each trajectory point. The spatial translation coordinate deviation component of each trajectory point includes translation deviation values ​​along the X-axis, Y-axis, and Z-axis of the base coordinate system. .

[0078] In step S162, the spatial translation coordinate deviation components of all trajectory points are accumulated point by point to obtain the error accumulation feature vector. , where N is the total number of trajectory points on the processing path. Although the spatial translation deviation of each trajectory point is small individually, the overall distribution pattern of the deviation after being accumulated along the entire processing path can reflect the overall direction of the edge contour. For example, the difference in the distribution trend of the deviation in the first half and the second half of the path can reflect macroscopic deformation characteristics such as the wavy contour or unilateral tilt of the edge. Therefore, the error accumulation feature vector is used to characterize the overall deformation trend of the entire edge of the intensifying screen metal foil in three-dimensional space.

[0079] In this embodiment, by accumulating the minute positional deviations of each trajectory point along the entire processing path, the discrete trajectory point deviations are aggregated into the macroscopic contour deformation features of the entire edge. Compared to the limitations of existing technologies that only focus on single-point deviations and cannot know the deformation trend of the entire edge contour, this embodiment enables the change trend of edge quality to be perceived and evaluated in real time during processing, providing a data foundation for subsequent defect type classification and identification.

[0080] In one embodiment, see Figure 6 Step S180 includes: Step S181: Construct a classifier using the support vector machine algorithm.

[0081] Step S182: Input the cumulative error feature vector into the classifier and calculate the geometric distance between the cumulative error feature vector and the classification hyperplane.

[0082] Step S183: Based on the numerical distribution of geometric distance, the contour shape of each segment of the metal foil edge is respectively mapped to one or more of burrs, collapsed edges, or tears.

[0083] Step S184: Use the corresponding results as the predicted metal foil edge defect category and the corresponding spatial distribution location.

[0084] In step S181, a classifier is constructed using the Support Vector Machine (SVM) algorithm. SVM is a supervised learning model that separates samples of different categories by finding the optimal classification hyperplane. The classifier constructed in this step is used to identify the defect category from the input cumulative error feature vector.

[0085] In step S182, the cumulative error feature vector is input into the classifier, and the geometric distance between the cumulative error feature vector and the classification hyperplane is calculated. This geometric distance reflects the similarity between the edge deformation trend represented by the currently input cumulative error feature vector and the typical features of various defects. The smaller the geometric distance, the higher the degree of matching between the edge deformation trend and the type of defect.

[0086] In step S183, based on the numerical distribution of geometric distance, the contour morphology of each segment of the metal foil edge is assigned to one or more of the following: burrs, collapsed edges, or tears. Different segments of the error accumulation feature vector correspond to different defect morphologies. Burrs correspond to a sharp peak pattern where the error accumulation feature vector shows a significant bulge in a certain segment; collapsed edges correspond to a monotonic trend pattern where the error accumulation feature vector continuously increases or decreases along the path; and tears correspond to a sudden change pattern where the error accumulation feature vector shows a step change at a certain point. By matching the geometric distance value of each segment with the above patterns, the defect type of each segment is determined.

[0087] In step S184, the corresponding results are output as the predicted metal foil edge defect category and the corresponding spatial distribution location, thereby locating the specific segment where each defect is located on the entire edge.

[0088] In this embodiment, a support vector machine (SVM) algorithm is used to classify the cumulative error feature vector. This maps the contour deformation trend represented by the cumulative error feature vector to specific edge defect categories such as burrs, collapsed edges, and tears, along with their spatial distribution. This allows the defect prediction results to be directly used to guide compensation decisions. Compared to existing defect identification methods that rely on manual visual inspection or simple threshold judgment, this embodiment uses a support vector machine to achieve accurate classification in the feature space by finding the optimal classification hyperplane, significantly improving the accuracy and consistency of defect type identification.

[0089] In one embodiment, see Figure 7 Step S200 includes: Step S201: Based on the edge defect type and the change in vertical angle, query the preset inter-axis coupling relationship table, extract the inter-axis coupling degree of each rotation axis to the current defect, and configure the inter-axis coupling degree to represent the degree of influence of each rotation axis on the current defect.

[0090] Step S202: Assign a compensation priority to each rotating axis based on the inter-axis coupling degree of each rotating axis.

[0091] Step S203: Use inverse kinematics to perform pose inverse solving on the rotation axis with the highest compensation priority, and calculate the joint correction compensation amount.

[0092] Step S204: Output control signals for each rotation axis of the tool head according to the joint correction compensation amount.

[0093] In step S201, the inter-axis coupling degree can be used to quantify the influence of each rotation axis on the current edge defect category. Angular deviations of the robot's rotation axes will produce a coupling effect at the end effector. Deviations of different rotation axes may collectively affect the tool head's offset in the same direction, and deviations of the same rotation axis may simultaneously affect the end effector pose in multiple directions.

[0094] Furthermore, the inter-axis coupling relationship table is pre-stored in the database. This table records the contribution of each rotation axis deviation to the end pose error under different defect types and different vertical angle changes.

[0095] In step S202, the greater the inter-axis coupling, the more significant the influence of the rotating axis on the current defect, and it should be adjusted first during compensation. Therefore, this rotating axis is set as a higher compensation priority. The rotating axis with the next higher inter-axis coupling is set as the second highest compensation priority, and so on, until all rotating axes are assigned the corresponding compensation priority.

[0096] In step S203, inverse kinematics is the calculation process of inferring the angles of each joint based on the desired pose of the robot's end effector. The rotation axis with the highest compensation priority is taken as the primary variable to be solved. A matrix inversion operation is performed using the Jacobian matrix iteration algorithm to calculate the angle value that needs to be adjusted for that rotation axis, i.e., the joint correction compensation amount. During the solution process, rotation axes with lower priorities are treated as known constraints, gradually reducing the degrees of freedom and improving the solution efficiency.

[0097] As an example, in inverse kinematics solving, the mapping relationship between the robot's end-effector pose and the angles of each joint is as follows: Where X is the end-effector pose vector. These are the angle vectors for each joint. Joint correction compensation amounts. Solve iteratively using the pseudo-inverse of the Jacobian matrix: ,in For Jacobian matrices, This is the pseudo-inverse of the Jacobian matrix. This represents the desired pose correction amount at the end effector. Through multiple iterative calculations, the joint angle adjustment amount that satisfies the pose correction requirements at the end effector is gradually approximated. During the solution process, the joint variables corresponding to the rotation axis with the highest compensation priority are iteratively updated first.

[0098] In step S204, the control signal includes the angle direction and magnitude that each rotary axis needs to be adjusted. After being output to the servo driver of each rotary axis, it drives each rotary axis to perform the corresponding angle adjustment action, thereby achieving compensation for the tool head pose.

[0099] In this embodiment, the contribution of each rotation axis to the current defect is determined through an inter-axis coupling relationship table, and compensation priorities are assigned accordingly. This concentrates compensation actions on the most significantly affected rotation axes, avoiding motion conflicts and energy waste that may occur with simultaneous multi-axis compensation, thus improving compensation efficiency and solution accuracy. This embodiment uses inverse kinematics to perform pose inverse solving on the highest-priority rotation axis. Machine intelligence kinematic calculation methods are used to accurately map the desired end-effector pose to joint angle adjustment amounts, achieving a quantitative conversion from defect identification to joint compensation, providing a reliable computational basis for real-time toolhead pose adjustment. Compared to existing technologies that rely on manual experience to allocate compensation priorities or use uniform joint angle compensation amounts, this embodiment makes compensation decisions more scientific and accurate through quantitative weight allocation and precise inverse kinematics solution.

[0100] In one embodiment, see Figure 8 Step S202 includes: Step S2021: Calculate the coupling weight of each rotation axis to the current edge defect category.

[0101] Step S2022: Set the compensation sequence for each rotation axis in descending order of coupling weight.

[0102] In step S2021, the coupling weight of each rotation axis to the current edge defect category can be calculated based on the inter-axis coupling degree of each rotation axis. The inter-axis coupling degree reflects the contribution of each rotation axis deviation to the end pose error. The inter-axis coupling degree value of each rotation axis is normalized to obtain the coupling weight corresponding to each rotation axis. The coupling weight is expressed as a percentage to represent the relative influence of each rotation axis on the current defect. The larger the coupling weight, the more significant the influence of the rotation axis on the current defect.

[0103] In step S2022, the rotation axis with the largest coupling weight is set as the first compensation order, and the pose inverse solution of this rotation axis is performed first in subsequent inverse kinematics calculations; the rotation axis with the second largest coupling weight is set as the second compensation order, and so on, until all rotation axes are set with the corresponding compensation order.

[0104] In this embodiment, the contribution of each rotation axis to the current defect is quantified by coupling weight, and the compensation order is set according to the coupling weight from large to small, so that the compensation resources are preferentially allocated to the rotation axis that has the most significant impact on the defect. This avoids motion coupling and energy waste caused by simultaneous compensation of multiple axes, improves compensation efficiency, and provides clear variable priority for inverse kinematics solution, reducing the complexity of high-dimensional space solution.

[0105] In one embodiment, see Figure 9 Step S200 includes: Step S205: Based on the edge defect category and the corresponding spatial distribution location, the control signal is decomposed into segmented control signals corresponding to each processing path segment, and the parameters of each segmented control signal are matched with the defect type of the segment.

[0106] Step S206: According to the spatial sequence of the processing path segments, output the control signal of each segment to the corresponding rotary axis in sequence.

[0107] In step S205, different machining path segments have different defect types, and each segment requires different compensation parameters. Burr segments correspond to compensation parameters that reduce the depth of cut, collapsed edge segments correspond to compensation parameters that increase the depth of cut, and tearing segments correspond to compensation parameters that reduce the feed rate. The control signal is decomposed into multiple segmented control signals according to their spatial distribution, and the parameters of each segmented control signal are matched with the defect type of its respective segment.

[0108] In step S206, the tool head passes through each section sequentially along the machining path, and the control signals are output sequentially according to the section order to ensure that the compensation action is synchronized with the machining position in terms of timing, and to avoid incorrect compensation caused by outputting too early or too late.

[0109] In this embodiment, by decomposing the control signal into segmented control signals that match the defect type of each segment and outputting them sequentially in spatial order, the compensation action is performed only in the segment where the defect is located, avoiding over-compensation or miscompensation in defect-free segments. The parameters of the segmented control signals are set differently according to the defect type of the segment. Burrs, edge collapses, and tears correspond to different compensation parameters. By utilizing the spatial distribution positioning and segmented control capabilities of machine intelligence, segmented differentiated compensation is achieved, making the compensation parameters accurately match the actual defect type of each segment. Compared with the open-loop processing method of using uniform compensation parameters for the entire edge in the prior art, the targeting and effectiveness of the compensation are further improved.

[0110] For the same purpose, please refer to Figure 10 In one embodiment, an intelligent dynamic edge adjustment system based on the processing of metal foil for intensifying screens is provided, comprising: The pose acquisition module is used to acquire a smooth pose deviation sequence, which represents the relative pose relationship between the tool head and the metal foil during the metal foil processing.

[0111] The normal vector acquisition module is used to obtain the normal vector direction between the tool head and the metal foil surface.

[0112] The vertical angle calculation module is used to calculate the change in vertical angle between the tool head axis and the normal vector based on the spatial rotation matrix components and the normal vector direction in the smooth pose deviation sequence.

[0113] The cutting depth difference acquisition module is used to obtain the cutting depth difference between the actual cutting depth value and the theoretical cutting depth value of the tool head cutting into the metal foil, based on the change in vertical angle and the parameter value of the metal foil.

[0114] The error accumulation feature generation module is used to accumulate the spatial translation coordinate components in the smoothed pose deviation sequence in response to the cutting depth difference value being greater than a preset difference threshold, and generate an error accumulation feature vector. The error accumulation feature vector is used to represent the edge deformation trend of the metal foil.

[0115] The defect prediction module is used to classify the cumulative error feature vector and predict the edge defect category and corresponding spatial distribution location of the metal foil.

[0116] The control signal output module is used to output control signals for the tool head rotation axis based on the vertical angle change, edge defect type, and corresponding spatial distribution location.

[0117] The specific implementation methods and working principles of each module included in the above system correspond to the steps in the aforementioned method embodiments, and will not be repeated here.

[0118] For those skilled in the art, the above division of functional modules is merely an exemplary logical division. In practical applications, modules can be merged, split, or recombined according to specific needs. Each module can be integrated into a single processor or distributed across different processors or computing nodes. The operations performed by each module, the algorithmic steps involved, and the data processing flow adopted are all fully and clearly described in the foregoing method embodiments. Based on reading the foregoing method embodiments, those skilled in the art can implement the method steps into corresponding functional modules without creative effort and deploy them on general-purpose or special-purpose computing devices using programming languages ​​and software development tools.

[0119] This application also provides an electronic device, which includes one or more processors and memory resources represented by memory for storing instructions executable by the processor, such as application programs. The application programs stored in the memory may include one or more modules, each corresponding to a set of instructions.

[0120] The electronic device may also include a power supply component configured to perform power management of the electronic device, a wired or wireless network interface configured to connect the electronic device to a network, and an input / output (I / O) interface. The electronic device can be operated based on operating devices stored in memory, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0121] In one embodiment, a computer device, which may be a server, is also provided. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device stores data. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with external terminals via a network connection.

[0122] In one embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. The display unit of the computer device is used to form a visually visible image and may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0123] This application also provides a non-transitory computer-readable storage medium, which, when the instructions in the storage medium are executed by the processor of the electronic device, enables the electronic device to perform one of the above embodiments.

[0124] This application may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information may be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to: phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0125] It should be noted that although the steps of the method in this application are described in a specific order in the accompanying drawings, this does not require or imply that these steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps, such as omitting certain steps, combining multiple steps into one step, and / or breaking down one step into multiple steps, should all be considered part of this application.

[0126] It should be understood that this application is not limited to the detailed data of the methods presented in this specification. This application is capable of other implementations and can be implemented and performed in various ways. The foregoing variations and modifications fall within the scope of this application. It should be understood that the disclosure and definition of this application extend to all alternative combinations of two or more individual features mentioned or apparent in the text and / or drawings. All these different combinations constitute multiple alternative aspects of this application. The embodiments described in this specification illustrate the best known mode for implementing this application and will enable those skilled in the art to utilize this application.

Claims

1. A method for intelligent dynamic edge adjustment based on intensifying screen metal foil processing, characterized in that, include: Obtain a smooth pose deviation sequence and the direction of the normal vector between the tool head and the metal foil surface. The smooth pose deviation sequence is used to represent the relative pose relationship between the tool head and the metal foil during the metal foil processing. Based on the spatial rotation matrix components in the smoothed pose deviation sequence and the direction of the normal vector between the tool head and the metal foil surface, calculate the change in the vertical angle between the tool head axis and the normal vector. Based on the change in vertical angle and the parameter value of the metal foil, the difference in cutting depth between the actual cutting depth of the tool head into the metal foil and the theoretical cutting depth is obtained. In response to the cutting depth difference value being greater than a preset difference threshold, the spatial translation coordinate components in the smoothed pose deviation sequence are accumulated to generate an error accumulation feature vector, which is used to represent the edge deformation trend of the metal foil. The error accumulation feature vector is classified to obtain the predicted metal foil edge defect category and its corresponding spatial distribution location; Based on the vertical angle change, the edge defect type, and the corresponding spatial distribution location, a control signal for the tool head rotation axis is output.

2. The method according to claim 1, characterized in that, The process of obtaining the smooth pose deviation sequence includes: Acquire the actual pose data of the tool head during the processing of the metal foil edge, as well as the preset theoretical settings; By comparing the differences between the actual pose data and the theoretically set data in the spatial rotation matrix and spatial translation coordinates, an initial pose deviation sequence is obtained. The attitude deviation frequency components in the initial pose deviation sequence are subjected to state estimation and noise filtering to obtain the smooth pose deviation sequence.

3. The method according to claim 1, characterized in that, The acquisition tool head and the normal vector direction of the metal foil surface include: Obtain the 3D point cloud coordinates of the metal foil surface; The local plane equation of the three-dimensional point cloud coordinates is fitted using the least squares method; Extract the direction of the tool head and the normal vector of the metal foil surface from the local plane equation.

4. The method according to claim 1, characterized in that, The parameters of the metal foil include its thickness. The step of obtaining the difference in cutting depth between the actual cutting depth of the tool head and the theoretical cutting depth based on the change in vertical angle and the parameter values ​​of the metal foil includes: Based on the mapping relationship between the vertical angle change and the metal foil thickness, the actual cutting depth of the tool head into the metal foil is calculated. Obtain the preset theoretical cutting depth value; The cutting depth difference value is calculated based on the difference between the actual cutting depth value and the theoretical cutting depth value.

5. The method according to claim 1, characterized in that, The step of accumulating the spatial translation coordinate components in the smoothed pose deviation sequence to generate an error accumulation feature vector includes: Extract the spatial translation coordinate deviation components of each trajectory point in the edge processing path from the smooth pose deviation sequence; The error accumulation feature vector is obtained by summing the spatial translation coordinate deviation components of each trajectory point.

6. The method according to claim 1, characterized in that, The step of classifying the accumulated error feature vector to obtain the predicted metal foil edge defect category and its corresponding spatial distribution location includes: A classifier is constructed using the support vector machine algorithm; The error accumulation feature vector is input into the classifier, and the geometric distance between the error accumulation feature vector and the classification hyperplane is calculated. Based on the numerical distribution of the geometric distance, the contour shape of each segment of the metal foil edge is respectively mapped to one or more of burrs, collapsed edges, or tears; The corresponding results are used as the predicted edge defect categories of the metal foil and their corresponding spatial distribution locations.

7. The method according to claim 1, characterized in that, The step of outputting a control signal for the tool head rotation axis based on the vertical angle change, the edge defect type, and the corresponding spatial distribution location includes: Based on the edge defect category and the vertical angle change, a preset inter-axis coupling relationship table is queried to extract the inter-axis coupling degree of each rotation axis to the current defect. The inter-axis coupling degree is configured to represent the degree of influence of each rotation axis on the current defect. Based on the inter-axis coupling degree of each rotation axis, a compensation priority is assigned to each rotation axis; Inverse kinematics is used to perform pose inverse solving on the rotation axis with the highest compensation priority, and the joint correction compensation amount is calculated. The control signal for each rotation axis of the tool head is output according to the joint correction compensation amount.

8. The method according to claim 7, characterized in that, The process of assigning compensation priorities to each rotating axis based on the inter-axis coupling degree of each rotating axis includes: Calculate the coupling weight of each rotation axis to the current edge defect category; The compensation order is set for each rotation axis in descending order of coupling weight.

9. The method according to claim 1, characterized in that, The control signals for the output tool head rotation axis include: Based on the edge defect category and the corresponding spatial distribution location, the control signal is decomposed into segmented control signals corresponding to each processing path segment, and the parameters of each segmented control signal are matched with the defect type of the segment in which it is located. According to the spatial sequence of the processing path segments, the control signal of each segment is output to the corresponding rotary axis in sequence.

10. A smart dynamic edge adjustment system for metal foil processing based on intensifying screens, characterized in that, include: The pose acquisition module is used to acquire a smooth pose deviation sequence, which represents the relative pose relationship between the tool head and the metal foil during the metal foil processing. The normal vector acquisition module is used to obtain the normal vector direction between the tool head and the metal foil surface; The vertical angle calculation module is used to calculate the change in vertical angle between the tool head axis and the normal vector based on the spatial rotation matrix components in the smooth pose deviation sequence and the normal vector direction. The cutting depth difference acquisition module is used to acquire the cutting depth difference between the actual cutting depth value and the theoretical cutting depth value of the tool head cutting into the metal foil based on the vertical angle change and the parameter value of the metal foil. The error accumulation feature generation module is used to accumulate the spatial translation coordinate components in the smooth pose deviation sequence in response to the cutting depth difference value being greater than a preset difference threshold, and generate an error accumulation feature vector. The error accumulation feature vector is used to represent the edge deformation trend of the metal foil. The defect prediction module is used to classify the error accumulation feature vector and predict the edge defect category and corresponding spatial distribution location of the metal foil. The control signal output module is used to output a control signal for the tool head rotation axis based on the vertical angle change, the edge defect type, and the corresponding spatial distribution position.

Images