A method and system for multiple planning and force adjustment of grinding allowance in robotics

Abstract

This invention relates to a method and system for multiple planning and force adjustment of welding allowance in robotic machining. The invention includes real-time monitoring of weld allowance changes to obtain real-time actual weld allowance data; establishing a material removal rate model based on historical grinding data and material properties; wherein the historical grinding data includes the material removal amount of materials with different properties under different grinding parameters; the material removal rate model is used to predict the material removal amount based on the actual grinding force and material properties; the grinding parameters include grinding speed, grinding path, magnitude and direction of grinding force and / or the tilt angle of the grinding tool; when there is a deviation between the actual weld allowance data and the target allowance, the grinding parameters are dynamically adjusted based on the material removal rate model; future grinding needs are predicted based on the current grinding state, and the grinding parameters are adjusted in advance. This invention improves the consistency of welding allowance processing and grinding accuracy, and enhances grinding quality and efficiency.

Description

Technical Field

[0001] This invention relates to the field of robotic polishing technology, and in particular to a method and system for multiple planning and force adjustment of robotic polishing allowance. Background Technology

[0002] In modern industrial manufacturing, robotic polishing technology is gradually replacing traditional manual polishing methods to improve production efficiency and product quality. With the continuous development of industrial automation, robotic polishing technology has been widely applied in various fields such as automotive manufacturing, aerospace, and machinery manufacturing. However, traditional robotic polishing methods still have the following drawbacks in practical applications:

[0003] Unstable control of grinding allowance; traditional grinding methods usually set the grinding allowance based on experience. If the weld contour of the workpiece is uneven or the equipment vibrates, it is easy to cause inconsistent allowance, which will cause deviations in subsequent processing or assembly.

[0004] The grinding precision is not high; traditional grinding is mostly based on a pre-programmed fixed trajectory, which cannot achieve real-time adaptive adjustment to local differences in the weld, making it difficult to guarantee the surface roughness, flatness and other indicators after grinding.

[0005] Material removal rate is difficult to predict; the material removal rate varies depending on the material properties, grinding force, grinding speed and environmental conditions. Without accurate prediction of the material removal rate, over-grinding or under-grinding is very likely to occur. Summary of the Invention

[0006] Therefore, the present invention provides a method and system for multiple planning and force adjustment of the grinding allowance in a robot, which improves the consistency of grinding allowance processing and grinding accuracy, and enhances grinding quality and efficiency.

[0007] To solve the above-mentioned technical problems, the present invention provides a method for multiple planning and force adjustment of machining allowance in robotic polishing, comprising:

[0008] Real-time monitoring of changes in weld allowance yields real-time data on actual weld allowance.

[0009] A material removal rate model is established based on historical polishing data and material properties; wherein, the historical polishing data includes the amount of material removed by materials with different properties under different polishing parameters; the material removal rate model is used to predict the amount of material removed based on the actual polishing force and material properties; the polishing parameters include polishing speed, polishing path, magnitude and direction of polishing force and / or tilt angle of polishing tool;

[0010] When there is a deviation between the actual weld allowance data and the target allowance, the grinding parameters are dynamically adjusted based on the material removal rate model.

[0011] Predict future polishing needs based on the current polishing status and adjust polishing parameters in advance.

[0012] In one embodiment of the present invention, real-time monitoring of changes in weld allowance to obtain real-time actual weld allowance data includes:

[0013] Deploy a vision sensor, which includes at least one high-resolution camera, to ensure coverage of the entire processing area of ​​the weld.

[0014] The vision sensor acquires real-time images of the weld area, and preprocesses the real-time images, including noise reduction, contrast enhancement, and brightness adjustment, to improve image quality and reduce the impact of ambient light changes on monitoring.

[0015] In one embodiment of the present invention, a material removal rate model is established based on historical polishing data and material properties, including:

[0016] Collect data on the amount of material removed during grinding of materials with different properties under different grinding parameters in the past;

[0017] The material properties include hardness, brittleness, elastic modulus, and thermal conductivity.

[0018] In one embodiment of the present invention, the amount of material removed is expressed as:

[0019] R removal =k·F·t,

[0020] Among them, R removal denoted by , k represents the material removal amount, F represents the material removal rate coefficient, and t represents the grinding force.

[0021] In one embodiment of the present invention, when there is a deviation between the actual weld allowance data and the target allowance, the grinding parameters are dynamically adjusted based on the material removal rate model, including:

[0022] An interpolation algorithm is used to smoothly adjust the grinding trajectory and parameters, avoiding abrupt changes during the grinding process and ensuring the continuity and stability of the processing.

[0023] When the deviation exceeds the tolerance range, a global path replanning is triggered, and the grinding path is recalculated based on the actual situation of the current weld to ensure accurate coverage of the processing;

[0024] Adjust the posture of the grinding tool in real time, including the tilt angle and feed depth, to ensure the best contact between the tool and the weld surface and avoid over-grinding or under-grinding.

[0025] After adjusting the grinding parameters, continue to monitor the changes in weld allowance to verify the effectiveness of the parameter update;

[0026] If the deviation does not decrease significantly, repeat the path update process or continue to adjust the processing parameters to adapt to the changes in weld shape until the deviation is within the tolerance range.

[0027] In one embodiment of the present invention, predicting future polishing needs based on the current polishing state and adjusting polishing parameters in advance includes:

[0028] Multi-dimensional data is acquired in real time during the weld grinding process. This multi-dimensional data is obtained through a combination of a vision sensor, a current and voltage sensor, a sound sensor, and a force sensor. The vision sensor acquires the weld's morphology and surface features, and uses a deep learning algorithm to extract geometric features, allowance distribution, and surface condition from the weld images acquired by the vision sensor. The current and voltage sensor acquires current and voltage signals, and processes these signals using frequency domain analysis to identify abnormal operating conditions, including sudden changes in spindle load. The sound sensor acquires sound signals, and uses Mel-frequency cepstral coefficients to extract features from the sound signals to detect tool wear and vibration issues. The force sensor monitors the uniformity of the grinding force and analyzes the smoothness and deviation of the grinding trajectory in conjunction with displacement data.

[0029] The multi-dimensional data is fused using a deep neural network to obtain a weld grinding quality prediction model. The weld grinding quality prediction model outputs quality indicators in real time, including surface roughness, allowance consistency, and defect probability. Based on the weld grinding quality prediction model, future grinding needs are predicted, and grinding parameters are adjusted in advance.

[0030] In one embodiment of the present invention, it further includes:

[0031] Multiple cameras were installed around the sanding work area to monitor the sanding process.

[0032] This invention also provides a robot grinding allowance multiple planning and force adjustment system, comprising:

[0033] The weld allowance monitoring module is used to monitor changes in weld allowance in real time and obtain real-time actual weld allowance data.

[0034] The material removal rate model building module is used to build a material removal rate model based on historical polishing data and material properties; wherein, the historical polishing data includes the material removal amount of materials with different material properties under different polishing parameters; the material removal rate model is used to predict the material removal amount based on the actual polishing force and material properties; the polishing parameters include polishing speed, polishing path, magnitude and direction of polishing force and / or tilt angle of polishing tool;

[0035] The grinding parameter adjustment module is used to dynamically adjust the grinding parameters based on the material removal rate model when there is a deviation between the actual weld allowance data and the target allowance.

[0036] The prediction module is used to predict future polishing needs based on the current polishing status and adjust polishing parameters in advance.

[0037] In one embodiment of the present invention, a safety monitoring module is also included, which is used to monitor abnormal situations during the polishing process and automatically stop the polishing operation when a potential risk is detected.

[0038] The technical solution of the present invention has the following advantages compared with the prior art:

[0039] The present invention discloses a method and system for multiple planning and force adjustment of machining allowance in robotics. It captures high-definition images of the weld area in real time using a high-resolution camera, and improves image contrast and brightness by combining software algorithms. This enables more accurate identification of weld contours and allowances, achieving high-precision machining. It also provides uniform illumination under different lighting conditions, reduces shadows and reflections, significantly improves image quality, and provides more reliable data for subsequent image processing and analysis.

[0040] This invention establishes a material removal rate model based on historical data and material properties, which can predict the amount of material removed under different grinding forces. This provides a scientific basis for optimizing grinding forces and process parameters, and dynamically optimizes parameters such as grinding speed, path, force magnitude and direction, and tool tilt angle to adapt to different material properties and weld characteristics, thereby achieving consistent processing of allowances.

[0041] This invention improves the consistency and quality of finished products by automatically adjusting the grinding force, reducing deviations, and updating the grinding path in real time to adapt to changes in weld shape. The user interface allows operators to monitor the grinding process and manually adjust grinding parameters when necessary, enhancing operational flexibility and system adaptability. Attached Figure Description

[0042] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0043] Figure 1 This is a flowchart of the robot grinding allowance multiple planning and force adjustment method of the present invention. Detailed Implementation

[0044] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0045] In this invention, when directions (up, down, left, right, front, and back) are described, it is only for the convenience of describing the technical solution of this invention, and does not indicate or imply that the technical features referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0046] In this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," "exceeding," etc., are understood to exclude the stated number; "above," "below," "within," etc., are understood to include the stated number. In the description of this invention, the terms "first" and "second" are used only to distinguish technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0047] In this invention, unless otherwise explicitly defined, the terms "setting," "installing," and "connecting" should be interpreted broadly. For example, they can refer to a direct connection or an indirect connection through an intermediate medium; a fixed connection, a detachable connection, or an integrally formed connection; a mechanical connection, an electrical connection, or a connection capable of mutual communication; or the internal connection of two components or the interaction between two components. Those skilled in the art can reasonably determine the specific meaning of the above terms in this invention based on the specific content of the technical solution.

[0048] Example 1

[0049] Reference Figure 1 As shown, the present invention provides a method for multiple planning and force adjustment of machining allowance in robotic polishing, comprising:

[0050] S1. Monitor the changes in weld allowance in real time to obtain real-time actual weld allowance data.

[0051] It is understandable that weld allowance refers to the excess material thickness on the weld surface relative to the target weld shape (or ideal profile).

[0052] Specifically, by deploying vision sensors at appropriate locations within the robotic grinding system, including at least one high-resolution camera, the entire processing area of ​​the weld is ensured to be covered. The camera features a high frame rate and high resolution to capture subtle changes in allowance.

[0053] The vision sensor acquires real-time images of the weld area, and preprocesses the real-time images, including noise reduction, contrast enhancement, and brightness adjustment, to improve image quality and reduce the impact of ambient light changes on monitoring.

[0054] Furthermore, image processing techniques such as edge detection and image segmentation (e.g., CannyEdge, thresholding) or deep learning-based convolutional neural networks (CNNs) can be used to extract the geometric features of the weld seam, thereby obtaining the allowance distribution of the weld seam area. Utilizing a high-resolution camera combined with deep learning algorithms can maintain high recognition accuracy even with complex weld seam shapes or partial occlusion. It should be noted that the vision sensor can be adjusted according to the workpiece size and the size of the grinding area, as long as the camera can cover the corresponding processing area.

[0055] S2. Establish a material removal rate model based on historical polishing data and material properties; wherein, the historical polishing data includes the amount of material removed by materials with different material properties under different polishing parameters; the material removal rate model is used to predict the amount of material removed based on the actual polishing force and material properties; the polishing parameters include polishing speed, polishing path, magnitude and direction of polishing force and / or tilt angle of polishing tool.

[0056] Specifically, a material removal rate model is established based on historical polishing data and material properties, including:

[0057] Collect data on the amount of material removed during grinding of materials with different properties under different grinding parameters in the past;

[0058] The material properties include hardness H, brittleness B, elastic modulus E, and thermal conductivity λ.

[0059] The amount of material removed is expressed as:

[0060] R removal =k·F·t,

[0061] Among them, R removal denoted by , k represents the material removal amount, F represents the material removal rate coefficient, and t represents the grinding force.

[0062] It should be noted that when establishing the material removal rate model, multiple linear regression and neural networks are used to establish the mapping relationship between the material removal rate coefficient k and material properties (hardness H, brittleness B, elastic modulus E, thermal conductivity λ, etc.) and grinding parameters. During actual grinding, the expected removal amount (target margin) can be estimated in real time based on the known properties of the material and the grinding force.

[0063] For example, actual grinding experiment data under different materials and grinding parameters can be collected. For common materials (such as carbon steel, stainless steel, aluminum alloy, etc.), grinding experiments can be conducted under several sets of grinding parameters (different grinding speeds, grinding paths, magnitude and direction of grinding force, and / or tilt angle of grinding tools, etc.). Before and after each set of experiments, the change in material mass or thickness is measured, and the material properties such as hardness H, brittleness B, elastic modulus E, and thermal conductivity λ are recorded. At the same time, the grinding time t and the applied force F are recorded.

[0064] In addition, the changes in material mass or thickness before and after grinding are measured using methods such as weighing, 3D measurement, or laser thickness measurement. Appropriate quantitative indicators are selected for material properties based on experience or literature (e.g., hardness can be measured using HV or HRC values; brittleness can be measured using fracture toughness, etc.). Measurement data that is clearly inconsistent with reality (sensor error, operational error, etc.) is also considered. If certain material properties are missing, interpolation or deletion of the corresponding records can be used, but it is essential to ensure that the remaining data coverage is sufficiently broad. To reduce the impact of dimensional differences on the model, features such as hardness, brittleness, elastic modulus, and thermal conductivity can be standardized (e.g., Z-Score standardization) or normalized (e.g., mapping values ​​to the [0,1] interval).

[0065] The dataset is divided into training, validation, and test sets in ratios of 6:2:2 and 7:2:1 to ensure that the model has enough data to learn from, while reserving some data for independent validation and testing.

[0066] S3. When there is a deviation between the actual weld seam allowance data and the target allowance, the grinding parameters are dynamically adjusted based on the material removal rate model.

[0067] Specifically, interpolation algorithms (such as B-splines) are used to smoothly adjust the grinding trajectory and parameters, avoiding abrupt changes during the grinding process and ensuring the continuity and stability of the process. In industrial grinding, the continuity of the trajectory and force parameters is crucial; abrupt changes can cause processing instability or lead to problems such as vibration and over-grinding. B-splines use a series of control points to smoothly interpolate the curve, allowing for gradual adjustments to local areas while maintaining the overall shape, avoiding drastic jumps.

[0068] When the deviation exceeds the tolerance range, a global path replanning is triggered, and the grinding path is recalculated based on the actual situation of the current weld to ensure the continuity and stability of grinding.

[0069] During the dynamic adjustment process, the posture of the grinding tool is adjusted in real time, including the tilt angle and feed depth, to ensure the best contact between the tool and the weld surface and avoid over-grinding or under-grinding.

[0070] After adjusting the grinding parameters, continue to monitor the changes in weld allowance to verify the effectiveness of the parameter update;

[0071] If the deviation does not decrease significantly, repeat the path update process or continue to adjust the processing parameters to adapt to the changes in weld shape until the deviation is within the tolerance range.

[0072] For example, the deviation between the actual weld allowance data and the target allowance is compared. If the deviation is within an acceptable tolerance range, the existing grinding strategy is maintained; if it exceeds a threshold, dynamic adjustment is required. Thresholds can be set based on material characteristics, grinding accuracy requirements, or quality inspection standards, such as ±0.05mm, ±0.1mm, or ±0.2mm. If the deviation exceeds this threshold, it is considered that a significant process adjustment or path replanning is needed. Simultaneously, first-level, second-level, or third-level thresholds can be set: when the deviation is within the first-level threshold, the current grinding parameters are maintained; when it exceeds the second-level threshold, interpolation correction is performed; and when it exceeds the third-level threshold (maximum tolerance range), global path replanning is triggered.

[0073] Based on the material removal rate model, as well as the current grinding force, grinding time, and material properties, the potential material removal amount is estimated in real time. When a deviation is detected, reverse calculation is required to increase or decrease the grinding force or adjust the path within the range of variation to bring the weld allowance closer to the target value.

[0074] Interpolation algorithms or global path replanning can be used to fine-tune or significantly modify parameters such as grinding trajectory or grinding force; by continuously monitoring weld allowance and processing effect, once the new parameters take effect, it is necessary to verify whether the deviation has been reduced.

[0075] The original grinding path is discretized into several control points, and the pose (position and attitude) of some key points is corrected based on the actual weld contour and allowance information.

[0076] The B-spline algorithm is used to reconstruct the grinding trajectory on the updated control point sequence; keyframes can also be set for interpolation of grinding force, tool tilt angle, etc., to ensure that the force smoothly transitions from the original value to the new value.

[0077] When local adjustments fail to control the deviation within a reasonable range, or when the weld shape changes significantly (e.g., local weld protrusions / depressions), a global path recalculation is triggered. This includes reconstructing the overall weld contour, which can be achieved using laser scanning or vision sensors to capture the current weld's 3D morphology; on the new weld surface model, the grinding path is globally replanned using methods such as offline programming software, feasible region search, or CAD / CAM-based algorithms; and, combined with a material removal rate model, appropriate grinding forces, feed rates, and grinding times are assigned to different parts.

[0078] When adjusting the posture of the grinding tool in real time, the tilt angle is adjusted to maintain a reasonable contact angle between the grinding tool and the weld surface, thereby improving the removal efficiency.

[0079] The amount of material removed is affected by adjusting the feed depth. Through multi-axis linkage of the robot's end effector, the grinding tool can achieve a combination of rotation and translation in three-dimensional space.

[0080] After each attitude update, the removal speed of weld seam allowance, surface quality, and grinding force uniformity are detected by combining vision and force sensors under the new attitude.

[0081] If the deviation is not significantly reduced, continue the process until it is within the tolerance range.

[0082] S4. Predict future polishing needs based on the current polishing status and adjust polishing parameters in advance.

[0083] Specifically, multi-dimensional data is acquired in real time during the weld grinding process. This multi-dimensional data is obtained through a combination of a vision sensor, a current and voltage sensor, a sound sensor, and a force sensor. The vision sensor acquires the weld's morphology and surface features, using deep learning algorithms to extract geometric features, allowance distribution, and surface condition from the weld images acquired by the vision sensor. The current and voltage sensor acquires current and voltage signals, which are processed using frequency domain analysis to identify abnormal operating conditions, including sudden changes in spindle load. The sound sensor acquires sound signals, extracting features from the sound signals using Mel-frequency cepstral coefficients (MFCC) to detect tool wear and vibration issues. The force sensor monitors the uniformity of the grinding force, combining displacement data to analyze the smoothness and deviation of the grinding trajectory.

[0084] The multi-dimensional data is fused using a deep neural network to obtain a weld grinding quality prediction model. The weld grinding quality prediction model outputs quality indicators in real time, including surface roughness, allowance consistency, and defect probability. Based on the weld grinding quality prediction model, future grinding needs are predicted, and grinding parameters are adjusted in advance.

[0085] The deep neural network is a multi-input structure, including:

[0086] A visual subnetwork is used to extract weld morphology features from image or point cloud data;

[0087] The audio sub-network is used to process audio features such as MFCC;

[0088] Current, voltage, and force data sub-networks are used to process the timing signals of the grinding spindle and grinding force;

[0089] After fusing the above features at higher or middle layers of the network, the predicted results such as weld surface roughness, allowance consistency, and defect probability are output.

[0090] In addition, during the grinding process, if a sudden surge in current, tool breakage, or obvious abnormal noise is detected, an emergency stop command is immediately triggered, and update commands for parameters such as path or grinding force are sent to the robot controller via the industrial communication protocol to ensure safety.

[0091] For example, a vision sensor (high-resolution industrial camera) is mounted on a fixed bracket at the robot's end effector or grinding station, covering the weld area; current and voltage sensors are integrated into the grinding machine spindle or connected to the spindle drive motor (servo system); sound sensors (industrial microphone array or single-point microphone) are located near the grinding station or mounted at the robot's end effector to collect grinding sounds in real time and determine tool wear, vibration, or abnormal sounds; audio signals are typically collected at a sampling rate of 16kHz to 48kHz. A force sensor (six-dimensional force / torque sensor or end effector force sensor) is installed between the robot's end effector flange and the grinding machine to detect the magnitude, direction, and uniformity of the grinding force; the host computer performs data acquisition, preprocessing, and deep learning model inference, and sends grinding parameter adjustment commands to the robot controller.

[0092] Before and during weld grinding, an industrial camera acquires images of the weld surface at 10–30 fps (or higher). The images or point cloud data are sent to a host computer in real time for preprocessing (denoising, correction, ROI extraction, etc.). While the grinder is operating, the spindle current / voltage is recorded multiple times per second. If a sudden abnormal increase or decrease in current or voltage is detected, it is sent to the host computer via event marking for analysis (e.g., whether there is overload or idling). A sound sensor continuously records audio, periodically (e.g., every 1 or 2 seconds) capturing an audio frame. This audio data is transmitted to the host computer via a dedicated interface for feature extraction (MFCC, etc.) to determine if grinding is stable and whether the tool is excessively worn or vibrating abnormally. A force sensor continuously outputs the current grinding force (Fx, Fy, Fz), sending it to the host computer every sampling period (e.g., 50–200 Hz) for calculation of grinding trajectory deviation, force uniformity, etc., to help determine whether the clamping force needs adjustment or the path needs to be replanned.

[0093] A unified timestamp or industrial communication protocol (EtherCAT, PROFINET, ROS2, etc.) is used to ensure that the data from all sensors are aligned at the same time. The host computer merges multi-source data such as vision, current, voltage, sound, and force according to time sequence or frame number to form multimodal data.

[0094] A multi-input deep neural network is used, including a vision sub-network for processing image features; a sound sub-network for extracting MFCC features and then processing them through a one-dimensional convolutional or LSTM network; and current, voltage, and force data sub-networks for processing time-series signals through fully connected layers or LSTM branches. Feature fusion is performed at high or intermediate layers of the network to finally output the quality prediction results. Among them, surface roughness (Ra, Rz, etc.) is predicted by network regression or classified into grades (excellent, acceptable, need improvement, etc.); the consistency of the allowance is represented by a score in the range of [0,1] or [0,100%] to indicate whether the current weld allowance is uniform; the defect probability includes the probability of occurrence of defects such as holes, pinholes, overheating / cracks, etc., which is output through network classification.

[0095] By collecting a large amount of labeled data: each data point includes image clips, current and voltage waveforms, sound clips, and force sensor data, and is associated with actual measured surface roughness, margin distribution, defect records, etc.; after labeling these multimodal data, supervised learning is performed to train a deep neural network model.

[0096] When grinding a section of weld, the deep learning network comprehensively predicts the weld morphology, tool condition, and force distribution of subsequent workstations. If it determines that "there is an 80% probability of defects in the weld area 10cm behind," it intervenes in advance. Before entering this high-risk or thicker area, the grinding speed is reduced or the feed path is changed. If the thickness of the material in the area is too thick, the grinding force can be appropriately increased (within the material's tolerance). If the sound sensor and deep learning determine that the tool is severely worn, the system is reminded to replace the grinding tool after the next work cycle. If a serious anomaly is detected (such as a sharp current surge or tool breakage), an emergency stop can be triggered immediately to ensure human and machine safety. Once the model provides adjustment suggestions, the host computer issues new instructions to the robot controller through the industrial communication protocol, including updating the trajectory and adjusting the grinding force or posture parameters. After the robot executes the instructions, it continues to verify the adjustment effect through various sensors, achieving continuous closed-loop optimization.

[0097] Example 2

[0098] Based on the same inventive concept, this embodiment provides a robot grinding allowance multiple planning and force adjustment system. The principle of solving the problem is similar to that of the robot grinding allowance multiple planning and force adjustment method, and the repeated parts will not be described again.

[0099] This embodiment provides a robot grinding allowance multiple planning and force adjustment system, including: The present invention also provides a robot grinding allowance multiple planning and force adjustment system, including:

[0100] The weld allowance monitoring module is used to monitor changes in weld allowance in real time and obtain real-time actual weld allowance data.

[0101] The material removal rate model building module is used to build a material removal rate model based on historical polishing data and material properties; wherein, the historical polishing data includes the material removal amount of materials with different material properties under different polishing parameters; the material removal rate model is used to predict the material removal amount based on the actual polishing force and material properties; the polishing parameters include polishing speed, polishing path, magnitude and direction of polishing force and / or tilt angle of polishing tool;

[0102] The grinding parameter adjustment module is used to dynamically adjust the grinding parameters based on the material removal rate model when there is a deviation between the actual weld allowance data and the target allowance.

[0103] The prediction module is used to predict future polishing needs based on the current polishing status and adjust polishing parameters in advance.

[0104] The safety monitoring module is used to monitor abnormal situations during the polishing process and automatically stop the polishing operation when a potential risk is detected.

[0105] Security monitoring modules utilize cameras, sensors (such as infrared sensors, motion sensors, and sound sensors) and other devices (intrusion detection systems and access control devices) to monitor and record the status and activities of specific areas in real time to prevent theft and identify abnormal behavior.

[0106] In addition, it includes a user interface module that allows operators to monitor the polishing process and manually adjust polishing parameters when necessary. The user interface is connected to the robot control system through a real-time data exchange interface and integrates digital input / output functions to receive simple on / off control commands such as start / stop signals and emergency stop buttons from users.

[0107] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0109] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0110] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0111] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for multiple planning and force adjustment of machining allowance in robotic polishing, characterized in that, include: Real-time monitoring of changes in weld allowance yields real-time data on actual weld allowance. A material removal rate model is established based on historical polishing data and material properties; wherein, the historical polishing data includes the amount of material removed by materials with different properties under different polishing parameters; the material removal rate model is used to predict the amount of material removed based on the actual polishing force and material properties; the polishing parameters include polishing speed, polishing path, magnitude and direction of polishing force and / or tilt angle of polishing tool; When there is a deviation between the actual weld allowance data and the target allowance, the grinding parameters are dynamically adjusted based on the material removal rate model. Predict future polishing needs based on the current polishing status and adjust polishing parameters in advance; A material removal rate model is established based on historical polishing data and material properties, including: Collect data on the amount of material removed during grinding of materials with different properties under different grinding parameters in the past; The material properties include hardness, brittleness, elastic modulus, and thermal conductivity; The amount of material removed is expressed as: R_removal = k·F·t, Where R_removal represents the amount of material removed, k represents the material removal rate coefficient, F represents the grinding force, and t represents the grinding time; When there is a deviation between the actual weld allowance data and the target allowance, the grinding parameters are dynamically adjusted based on the material removal rate model, including: An interpolation algorithm is used to smoothly adjust the grinding trajectory and parameters, avoiding abrupt changes during the grinding process and ensuring the continuity and stability of the processing. When the deviation exceeds the tolerance range, a global path replanning is triggered, and the grinding path is recalculated based on the actual situation of the current weld to ensure accurate coverage of the processing; Adjust the posture of the grinding tool in real time, including the tilt angle and feed depth, to ensure the best contact between the tool and the weld surface and avoid over-grinding or under-grinding. After adjusting the grinding parameters, continue to monitor the changes in weld allowance to verify the effectiveness of the parameter update; If the deviation does not decrease significantly, repeat the path update process or continue to adjust the processing parameters to adapt to the changes in weld shape until the deviation is within the tolerance range. Predict future sanding needs based on the current sanding status and adjust sanding parameters in advance, including: Multi-dimensional data is acquired in real time during the weld grinding process. This multi-dimensional data is obtained through a combination of a vision sensor, a current and voltage sensor, a sound sensor, and a force sensor. The vision sensor acquires the weld's morphology and surface features, and uses a deep learning algorithm to extract geometric features, allowance distribution, and surface condition from the weld images acquired by the vision sensor. The current and voltage sensor acquires current and voltage signals, and processes these signals using frequency domain analysis to identify abnormal operating conditions, including sudden changes in spindle load. The sound sensor acquires sound signals, and uses Mel-frequency cepstral coefficients to extract features from the sound signals to detect tool wear and vibration issues. The force sensor monitors the uniformity of the grinding force and analyzes the smoothness and deviation of the grinding trajectory in conjunction with displacement data. The multi-dimensional data is fused using a deep neural network to obtain a weld grinding quality prediction model. The weld grinding quality prediction model outputs quality indicators in real time, including surface roughness, allowance consistency, and defect probability. Based on the weld grinding quality prediction model, future grinding needs are predicted, and grinding parameters are adjusted in advance.

2. The method for multiple planning and force adjustment of machining allowance in robotic polishing according to claim 1, characterized in that, Real-time monitoring of weld allowance changes yields real-time actual weld allowance data, including: Deploy a vision sensor, which includes at least one high-resolution camera, to ensure coverage of the entire processing area of ​​the weld. The vision sensor acquires real-time images of the weld area, and preprocesses the real-time images, including noise reduction, contrast enhancement, and brightness adjustment, to improve image quality and reduce the impact of ambient light changes on monitoring.

3. The method for multiple planning and force adjustment of machining allowance in robotic polishing according to claim 1, characterized in that, Also includes: Multiple cameras were installed around the sanding work area to monitor the sanding process.

4. A robotic grinding allowance multiple planning and force adjustment system, characterized in that, include: The weld allowance monitoring module is used to monitor changes in weld allowance in real time and obtain real-time actual weld allowance data. The material removal rate model building module is used to build a material removal rate model based on historical polishing data and material properties; wherein, the historical polishing data includes the material removal amount of materials with different material properties under different polishing parameters; the material removal rate model is used to predict the material removal amount based on the actual polishing force and material properties; the polishing parameters include polishing speed, polishing path, magnitude and direction of polishing force and / or tilt angle of polishing tool; The grinding parameter adjustment module is used to dynamically adjust the grinding parameters based on the material removal rate model when there is a deviation between the actual weld allowance data and the target allowance. The prediction module is used to predict future polishing needs based on the current polishing status and adjust polishing parameters in advance.

5. The robotic grinding allowance multiple planning and force adjustment system according to claim 4, characterized in that, It also includes a safety monitoring module, which monitors abnormal situations during the polishing process and automatically stops the polishing operation when a potential risk is detected.

Images