Robot applied to filter bag keel welding and control method thereof

Abstract

The invention provides a robot applied to filter bag keel welding and a control method of the robot. The robot comprises a dynamic clamping assembly, a track planning assembly and a quality detection assembly. The method comprises the steps that the clamping posture and force are adjusted in real time through the multi-degree-of-freedom clamping mechanism, filter bag keel materials with different diameters and surface states are matched, and positioning is achieved; before welding, a to-be-welded position of the filter bag keel material is scanned through three-dimensional contour detection, and a welding path is generated; welding parameters are adjusted in real time through a welding seam monitoring sensor, and welding deviation caused by filter bag keel material tolerance is compensated; in the welding process, the welding seam quality is evaluated in real time through multi-mode sensing; and automatically pausing and reporting a correction instruction when a defect welding seam is met. High-precision and high-quality full-automatic welding is completed, the technological process of precise positioning, intelligent execution and quality closed-loop verification is formed through the three assemblies, it is ensured that welding of the filter bag keel is stable and controllable, and the overall production efficiency and product consistency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial robot control, and in particular to a robot used for welding filter bag keels and a control method thereof. Background Art

[0002] Amidst the rapid growth of the environmental protection equipment manufacturing industry, filter bags, as core components of bag filters, face a significant impact on the service life of the entire system. Traditionally, workers hunch over, holding welding torches and repeating these actions thousands of times in an environment permeated with metal dust. This work environment is not only inefficient, requiring an average of 30 minutes of welding time per keel, but also carries serious occupational health risks. According to industry association statistics from 2023, the incidence of lumbar spine disease among welding workers is as high as 47%, and complaints of filter bag damage due to welding defects account for 68% of all quality issues.

[0003] Prior art 1, application number CN201310121417.9, discloses a filter bag surface cleaning device for a stand-alone dust collector used in welding robots. The device comprises a programmable logic controller (PLC), an intermediate relay, an electromagnetic pulse valve, and a fan. The intermediate relay is connected to the electromagnetic pulse valve, which is mounted on the fan's blower duct, which extends into the filter bag. The PLC output controls the electromagnetic pulse valve via the intermediate relay. The PLC activates the electromagnetic pulse valves one by one, allowing compressed air to flow through the fan nozzle to clean the filter bags. The filter bags suddenly expand, and the reverse airflow rapidly pulls dust attached to the bag surface away from the bag and into the ash hopper. While this device purifies the fumes generated during the welding process, the dust is discharged through the ash discharge valve. In addition, the equipment is very convenient for subsequent maintenance and modification, with low modification investment cost. Only the program needs to be changed, which is basically cost-free. Moreover, the system works stably, is not prone to problems, and is simple and convenient to operate. However, the curvature of the clamping arc cannot be automatically adjusted according to the different diameters of the filter bag keel, resulting in material eccentricity or slippage; Surface state sensitivity problem: the existing clamping device has no adaptive ability to surface roughness and material stiffness, which can easily cause surface indentation or clamping failure.

[0004] The existing technology is unable to automatically adjust the curvature of the clamping arc according to the different diameters of the filter bag keel, resulting in material eccentricity or slippage. Therefore, the present invention provides a robot for filter bag keel welding and a control method thereof. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a robot for welding filter bag keels, comprising:

[0006] The dynamic clamping component is responsible for adjusting the clamping posture and force in real time through the multi-degree-of-freedom clamping mechanism to match the filter bag keel materials with different diameters and surface conditions to achieve positioning;

[0007] The trajectory planning component is responsible for scanning the welded position of the filter bag keel material using 3D contour detection before welding to generate the welding path; and adjusts the welding parameters in real time through the weld monitoring sensor to compensate for welding deviations caused by the tolerance of the filter bag keel material;

[0008] The quality inspection component is responsible for real-time evaluation of weld quality during the welding process using multimodal sensing. It automatically pauses and reports correction instructions when a defective weld is encountered.

[0009] Optional, dynamic clamping assembly, including:

[0010] The contact sensing module is responsible for detecting the initial contact force of the touch pressure sensor of the multi-degree-of-freedom clamping mechanism. It uses the strain gauge array to establish the surface topology map of the filter bag keel material, including the diameter characteristic parameters or curvature radius distribution and the surface state parameters or roughness coefficient and material stiffness gradient, forming the clamping reference feature set;

[0011] The dynamic fitting module is responsible for performing three-stage fitting of radial fitting, circumferential fitting, and axial compensation on the clamping mechanism based on the benchmark feature set to obtain clamping parameters including opening and closing degree, pressure distribution, and preload force;

[0012] The real-time coupling module is responsible for transmitting the clamping parameters to the trajectory planning component in real time as boundary conditions for weld path calculation. At the same time, it continuously monitors changes in the clamping state. When the characteristic offset of the filter bag keel material exceeds the threshold due to thermal deformation, the clamping parameter recalculation closed loop is triggered.

[0013] Optionally, radial fitting dynamically adjusts the jaw opening according to the diameter characteristic parameter so that the curvature error between the clamping arc surface and the material meets the standard;

[0014] The circumferential fitting adjusts the clamping pressure distribution according to the roughness coefficient, adopting the surface contact mode in the smooth section and switching to the lattice contact mode in the texture section;

[0015] Axial compensation adjusts the axial preload through the material stiffness gradient data to prevent the filter bag keel material from axial movement in the welding heat affected zone.

[0016] Optional contact sensing module, including:

[0017] The initial contact submodule is responsible for ensuring that the multi-degree-of-freedom clamping mechanism contacts the surface of the filter bag keel material through the touch pressure sensor array. The strain gauge group collects the three-dimensional force distribution of the contact points to form an initial contact force field matrix. The extreme point distribution reflects the macroscopic curvature characteristics of the filter bag keel material surface. The main curvature direction is extracted through singular value decomposition to obtain the reference curvature radius distribution in the diameter characteristic parameter.

[0018] The surface topology quantification submodule is responsible for constructing the microscopic morphology characteristics of the filter bag keel material surface based on the spatial gradient changes of the initial contact force field and the dynamic response data of the high-frequency sampling strain gauge. The roughness coefficient of the filter bag keel material surface state parameters is quantified based on the correlation analysis between the initial contact force fluctuation frequency and amplitude. At the same time, the material stiffness gradient value is derived from the slope change rate of the force attenuation curve to form the original feature vector set.

[0019] The feature fusion submodule is responsible for inputting the diameter feature parameters and surface state parameters into the feature fusion device. It first performs Gaussian kernel smoothing on the curvature radius distribution to eliminate local distortion. Then, it maps the roughness coefficient to the curvature distribution curve through adaptive weighting to form a topological map with texture features. The stiffness gradient value is embedded in the depth dimension of the topological map as an axial compensation factor.

[0020] The benchmark calibration submodule is responsible for matching and verifying the topological map with the preset filter bag keel material database, and correcting the validity range of the characteristic parameters through residual analysis; when the coupling error between the curvature radius distribution and the roughness coefficient is less than the set threshold, the current parameter combination is locked as the clamping benchmark feature set. Each coordinate point in the topological map contains three types of normalized data: radial curvature, circumferential texture and axial stiffness.

[0021] Optional, initial contact submodule, including:

[0022] The force field feature base construction unit is responsible for constructing the original contact force field matrix using the three-dimensional force data collected by the touch pressure sensor array. Its row vectors correspond to the distribution of spatial sampling points, and the column vectors record the contact force amplitude at each sampling time step. The extreme point groups in the matrix that significantly deviate from the mean constitute the characteristic observation set of the surface macroscopic curvature.

[0023] The mechanical feature space mapping unit is responsible for performing singular value decomposition on the original contact force field matrix to obtain a set of eigenvectors representing the main directions of the force;

[0024] The curvature extreme value trajectory extraction unit is responsible for obtaining a cluster of characteristic curves along the two main curvature directions by tracing the spatial evolution path of the force field extreme points within the characteristic reference plane. The density distribution of the extreme points on each characteristic curve reflects the severity of the curvature change in that direction, and the inverse of the extreme point spacing constitutes the discrete curvature sampling sequence.

[0025] The continuous curvature field reconstruction unit is responsible for performing smooth interpolation of the discrete curvature sampling sequence based on energy minimization to generate a continuous curvature function distributed along the main curvature direction; during the interpolation process, the curvature mutation characteristics that match the distribution of the initial force field extreme points are automatically retained; and the continuous curvature function is converted into a set of curvature radius distribution parameters, where the anisotropic characteristics are characterized by the ratio of the radii of the two main directions.

[0026] Optionally, the two orthogonal vectors with the largest eigenvalues ​​in the eigenvector group of the mechanical feature space mapping unit correspond to the mechanical response modes in the maximum curvature direction and the minimum curvature direction of the filter bag keel surface, respectively, and the plane formed by the two main direction vectors is the curvature feature reference plane.

[0027] Optional, dynamic fitting module, including:

[0028] The radial geometry adaptation submodule is responsible for establishing a jaw motion model based on the curvature radius distribution parameters provided by the contact sensing module, extracting the characteristic radius extreme value sequence in the main curvature direction, and driving the multi-link mechanism to generate the corresponding variable diameter motion trajectory. During the motion process, the matching degree of the clamping arc pressure gradient and the reference curvature distribution is compared in real time, and incremental compensation is used to gradually approach the optimal opening and closing degree, thereby determining the jaw posture parameters that minimize the overall curvature deviation.

[0029] The circumferential contact optimization submodule is responsible for dynamically configuring the contact mode within the radial geometric framework based on the spatial distribution characteristics of the roughness coefficient. For low-roughness areas, a continuous and uniform clamping pressure field is generated, and an integral contact pad is used for the corresponding surface. For high-roughness sections, a discrete lattice force pattern is converted, and the layout density of each contact point is negatively correlated with the local roughness amplitude. The optimized pressure distribution forms a conformal mapping between the surface texture characteristics and the contact stress field.

[0030] The axial load balancing submodule is responsible for superimposing preload compensation on the basis of radial-circumferential constraints according to the axial distribution law of the stiffness gradient value. A stiffness piecewise function is constructed along the keel axis, and the thermal expansion compensation amount is independently calculated in each stiffness interval. The preload loading curve follows the stiffness gradient change trend, adopting a linear growth strategy in the low stiffness area and implementing saturation limiting control in the high stiffness area.

[0031] Optional trajectory planning component, including:

[0032] The initial module of 3D contour modeling is responsible for dynamically clamping the components through the multi-degree-of-freedom clamping mechanism to position the filter bag keel material. It then triggers the 3D contour detection program to perform a spiral scan along the axial direction of the filter bag keel, collect surface topology data, and generate a complete 3D point cloud model that includes diameter fluctuations, seam misalignment, and local deformation.

[0033] The welding path dynamic planning module is responsible for extracting the geometric features of the area to be welded based on the contour modeling results of the complete 3D point cloud model; the geometric features are input into the path generator to construct the reference spiral trajectory, and then the dynamic correction vector derived from the geometric deviation is superimposed;

[0034] The real-time welding deviation compensation module is responsible for capturing the molten pool morphology characteristics, molten pool morphology characteristics and thermal deformation drift at the sampling rate of the weld monitoring sensor.

[0035] Optional quality inspection components, including:

[0036] The welding synchronous data acquisition module is responsible for capturing the radiation characteristic distribution of the weld area after welding, recording the dynamic behavior of the molten pool and the shape evolution characteristics of the solidification trajectory; capturing high-frequency vibration signals during the welding process, extracting the transient energy characteristics of bubble formation or material cracking; monitoring the axial temperature gradient of the welding heat-affected zone, and establishing a correlation model between heat input and material micro-deformation;

[0037] The multi-feature fusion analysis module is responsible for comparing the melting width fluctuation with the reference weld geometry data provided by the 3D contour modeling module. If the axial deviation exceeds the preset threshold, it is determined to be a risk of lack of fusion. The burst energy pulse peak of the transient energy feature is associated with the seam misalignment compensation amount in the path planning module. If the energy peak phase is consistent with the compensated azimuth angle deviation, a porosity defect warning is triggered. Combining the temperature gradient drop rate with the clamping pressure data fed back by the dynamic clamping component, if the temperature in the high-pressure area drops abnormally slowly, it is identified as uneven heat conduction caused by clamping deformation.

[0038] The defect closed-loop decision module is responsible for generating three types of quality criteria: axial deviation, energy peak, and abnormal temperature drop. When any of the criteria exceeds the limit, the welding process is suspended and the trajectory planning component is requested to replan the local path, and the optimized clamping pressure value is fed back to the dynamic clamping component.

[0039] The present invention provides a control method for a robot used for welding filter bag keels, comprising the following steps:

[0040] Through the multi-degree-of-freedom clamping mechanism, the clamping posture and force are adjusted in real time to match the filter bag keel materials with different diameters and surface conditions to achieve positioning;

[0041] Before welding, 3D contour detection is used to scan the weld position of the filter bag keel material to generate the welding path; and the welding parameters are adjusted in real time through the weld monitoring sensor to compensate for the welding deviation caused by the tolerance of the filter bag keel material;

[0042] During the welding process, multimodal sensing is used to evaluate the weld quality in real time; when a defective weld is encountered, the process is automatically paused and correction instructions are reported.

[0043] The dynamic clamping assembly of this invention adaptively secures keel materials of varying diameters and surface conditions, ensuring accurate initial weld position and laying the foundation for precise welding. The trajectory planning component automatically plans the welding path based on the scanned weld contour and, in conjunction with weld monitoring data, adjusts the welding gun's posture and parameters in real time, proactively compensating for material tolerances and ensuring weld consistency. The quality inspection component monitors the weld state through real-time sensing. Upon detecting an unqualified weld, it immediately interrupts the operation and generates correction instructions, reducing the need for subsequent rework.

[0044] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0045] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0047] Figure 1 This is a block diagram of a robot used for welding filter bag keels in Example 1 of the present invention;

[0048] Figure 2 This is a block diagram of the dynamic clamping component in Example 2 of the present invention;

[0049] Figure 3 This is a block diagram of the trajectory planning component in Example 7 of the present invention;

[0050] Figure 4 This is a block diagram of the quality detection component in Example 8 of the present invention;

[0051] Figure 5 This is a flow chart of a control method for a robot used for welding filter bag keels in Example 9 of the present invention;

[0052] Figure 6 This is a schematic structural diagram of a robot used for welding filter bag keels in Example 10 of the present invention. DETAILED DESCRIPTION

[0053] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0054] The terms used in the embodiments of the present application are for the purpose of describing specific embodiments only and are not intended to limit the embodiments of the present application. The singular forms "a", "the" and "the" used in the embodiments of the present application are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0055] When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0056] Example 1: Figure 1 As shown, an embodiment of the present invention provides a robot for welding filter bag keels, comprising:

[0057] The dynamic clamping component is responsible for adjusting the clamping posture and force in real time through the multi-degree-of-freedom clamping mechanism to match the filter bag keel materials with different diameters and surface conditions to achieve positioning;

[0058] The trajectory planning component is responsible for scanning the welded position of the filter bag keel material using 3D contour detection before welding to generate the welding path; and adjusts the welding parameters in real time through the weld monitoring sensor to compensate for welding deviations caused by the tolerance of the filter bag keel material;

[0059] The quality inspection component is responsible for real-time evaluation of weld quality during the welding process using multimodal sensing. It automatically pauses and reports correction instructions when a defective weld is encountered.

[0060] The working principles and beneficial effects of the above technical solution are as follows: The dynamic clamping assembly of this embodiment uses a multi-degree-of-freedom clamping mechanism to adjust the clamping posture and force in real time to match filter bag keel materials of varying diameters and surface conditions for positioning. Before welding, the trajectory planning assembly uses three-dimensional contour detection to scan the welded position of the filter bag keel material and generate a welding path. Weld monitoring sensors adjust welding parameters in real time to compensate for welding deviations caused by filter bag keel material tolerances. During welding, the quality inspection assembly uses multimodal sensing to assess weld quality in real time. When a defective weld is encountered, the assembly automatically pauses and reports correction instructions. The dynamic clamping assembly of the above solution adaptively fixes keel materials of varying diameters and surface conditions, ensuring the accuracy of the initial weld position and laying the foundation for precise welding. The trajectory planning assembly automatically plans the welding path based on the scanned weld contour and, in combination with weld monitoring data, adjusts the welding gun posture and parameters in real time to proactively compensate for material tolerances and ensure weld consistency. The quality inspection assembly monitors the weld condition through real-time sensing. Upon detecting a defective weld, it immediately interrupts the operation and generates correction instructions, reducing the need for subsequent rework.

[0061] In summary, this embodiment achieves high-precision, high-quality fully automatic welding: the three components form a process flow of precise positioning, intelligent execution, and closed-loop quality verification, ensuring stable and controllable welding of the filter bag keel and improving overall production efficiency and product consistency.

[0062] Example 2: Figure 2 As shown, based on Example 1, the dynamic clamping assembly provided by the embodiment of the present invention includes:

[0063] The contact sensing module is responsible for detecting the initial contact force of the touch pressure sensor of the multi-degree-of-freedom clamping mechanism. It uses the strain gauge array to establish the surface topology map of the filter bag keel material, including diameter characteristic parameters (curvature radius distribution) and surface state parameters (roughness coefficient, material stiffness gradient), forming the clamping benchmark feature set;

[0064] The dynamic fitting module is responsible for performing three-stage fitting of radial fitting, circumferential fitting, and axial compensation on the clamping mechanism based on the benchmark feature set to obtain clamping parameters including opening and closing degree, pressure distribution, and preload force;

[0065] Among them, radial fitting dynamically adjusts the jaw opening and closing according to the diameter characteristic parameters, so that the curvature error between the clamping arc surface and the material meets the standard;

[0066] The circumferential fitting adjusts the clamping pressure distribution according to the roughness coefficient, adopting the surface contact mode in the smooth section and switching to the lattice contact mode in the texture section;

[0067] Axial compensation adjusts the axial preload force through the material stiffness gradient data to prevent the filter bag keel material from axial movement in the welding heat affected zone;

[0068] The real-time coupling module is responsible for transmitting the clamping parameters to the trajectory planning component in real time as boundary conditions for weld path calculation. At the same time, it continuously monitors changes in the clamping state. When the characteristic offset of the filter bag keel material exceeds the threshold due to thermal deformation, the clamping parameter recalculation closed loop is triggered.

[0069] The working principle and beneficial effects of the above technical solution are as follows: the touch pressure sensor of the multi-degree-of-freedom clamping mechanism of the contact sensing module of this embodiment detects the initial contact force, and establishes the surface topology map of the filter bag keel material through the strain gauge array, which includes the diameter characteristic parameters (curvature radius distribution) and the surface state parameters (roughness coefficient, material stiffness gradient), and constitutes the clamping reference feature set; the dynamic fitting module is based on the reference feature set, and the clamping mechanism performs radial fitting, circumferential fitting and axial compensation three-stage fitting to obtain the clamping parameters including opening and closing degree, pressure distribution and preload force; wherein, the radial fitting is based on the diameter characteristic parameters dynamic fitting. Dynamically adjust the opening and closing degree of the jaws to ensure that the error between the clamping arc surface and the material curvature meets the standard; circumferential fitting adjusts the clamping pressure distribution according to the roughness coefficient, adopts the surface contact mode in the smooth section, and switches to lattice contact in the texture section; axial compensation adjusts the axial preload through the material stiffness gradient data to prevent the filter bag keel material from axial movement in the welding heat affected zone; the real-time coupling module transmits the clamping parameters to the trajectory planning component in real time as the boundary condition for the weld path calculation; at the same time, it continuously monitors the changes in the clamping state. When the characteristic offset of the filter bag keel material exceeds the threshold due to thermal deformation, the clamping parameter recalculation closed loop is triggered. The above solution is based on the topological map constructed by the contact perception module (curvature radius distribution / roughness coefficient / stiffness gradient ternary parameters), and the dynamic fitting module realizes three-dimensional mechanical coupling: radial curvature matching ensures the jaw envelope fit, circumferential roughness responsive pressure distribution, and axial stiffness compensation triple adaptation, which solves the problem of nonlinear deformation compensation of the filter bag keel special-shaped surface during thermal processing. A high-speed control loop established through a real-time coupling module uses clamping parameters as rigid constraints for trajectory planning. It simultaneously monitors characteristic offsets caused by thermal deformation, enabling control of relative material displacement during welding and effectively suppressing creep effects in the 300-450°C heat-affected zone. A three-stage fitting algorithm and trajectory planning component form a dual spatial-mechanical feedback mechanism: radial opening ensures basic positioning accuracy, circumferential pressure distribution optimizes the friction coupling coefficient, and axial preload compensates for thermally induced stiffness degradation. These three factors work together to reduce welding path tracking errors.

[0070] Example 3: Based on Example 2, the contact sensing module provided by the embodiment of the present invention includes:

[0071] The initial contact submodule is responsible for ensuring that the multi-degree-of-freedom clamping mechanism contacts the surface of the filter bag keel material through the touch pressure sensor array. The strain gauge group collects the three-dimensional force distribution of the contact points to form an initial contact force field matrix. The extreme point distribution reflects the macroscopic curvature characteristics of the filter bag keel material surface. The main curvature direction is extracted through singular value decomposition to obtain the reference curvature radius distribution in the diameter characteristic parameter.

[0072] The surface topology quantification submodule is responsible for constructing the microscopic morphology characteristics of the filter bag keel material surface based on the spatial gradient changes of the initial contact force field and the dynamic response data of the high-frequency sampling strain gauge. The roughness coefficient of the filter bag keel material surface state parameters is quantified based on the correlation analysis between the initial contact force fluctuation frequency and amplitude. At the same time, the material stiffness gradient value is derived from the slope change rate of the force attenuation curve to form the original feature vector set.

[0073] The feature fusion submodule is responsible for inputting the diameter feature parameters and surface state parameters into the feature fusion device. It first performs Gaussian kernel smoothing on the curvature radius distribution to eliminate local distortion. Then, it maps the roughness coefficient to the curvature distribution curve through adaptive weighting to form a topological map with texture features. The stiffness gradient value is embedded in the depth dimension of the topological map as an axial compensation factor.

[0074] The benchmark calibration submodule is responsible for matching and verifying the topological map with the preset filter bag keel material database, and correcting the validity range of the characteristic parameters through residual analysis; when the coupling error between the curvature radius distribution and the roughness coefficient is less than the set threshold, the current parameter combination is locked as the clamping benchmark feature set. Each coordinate point in the topological map contains three types of normalized data: radial curvature, circumferential texture and axial stiffness.

[0075] The working principle and beneficial effects of the above technical solution are as follows: the multi-degree-of-freedom clamping mechanism of the initial contact submodule of this embodiment contacts the surface of the filter bag keel material through the touch pressure sensor array, and the strain gauge group collects the three-dimensional force distribution of the contact point to form an initial contact force field matrix. The extreme point distribution reflects the macroscopic curvature characteristics of the surface of the filter bag keel material, and the main curvature direction is extracted by singular value decomposition to obtain the reference curvature radius distribution in the diameter characteristic parameter; the surface topology quantization submodule is based on the spatial gradient change of the initial contact force field and the dynamic response data of the high-frequency sampling strain gauge to construct the microscopic morphology characteristics of the surface of the filter bag keel material; according to the correlation analysis of the initial contact force fluctuation frequency and amplitude, the roughness coefficient in the surface state parameters of the filter bag keel material is quantified; at the same time, the material is deduced by the slope change rate of the force attenuation curve. The stiffness gradient value constitutes the original feature vector set; the feature fusion submodule inputs the diameter feature parameters and the surface state parameters into the feature fusion device, first performs Gaussian kernel smoothing on the curvature radius distribution to eliminate local distortion, and then maps the roughness coefficient to the curvature distribution curve through adaptive weighting to form a topological map with texture features. The stiffness gradient value is embedded in the depth dimension of the topological map as an axial compensation factor; the benchmark calibration submodule matches and verifies the topological map with the preset filter bag keel material database, and corrects the validity range of the feature parameters through residual analysis; when the coupling error between the curvature radius distribution and the roughness coefficient is less than the set threshold, the current parameter combination is locked as the clamping benchmark feature set. Each coordinate point in the topological map contains three types of normalized data: radial curvature, circumferential texture and axial stiffness. The contact sensing module of the above scheme contacts the surface of the filter bag keel material through the touch pressure sensor array of the multi-degree-of-freedom clamping mechanism. The strain gauge group collects the three-dimensional force distribution to generate an initial contact force field matrix. The main curvature direction is extracted through singular value decomposition using the macroscopic characteristics of the extreme point distribution, and the reference curvature radius distribution is obtained. Based on this mechanical signal, the spatial gradient and high-frequency dynamic response data are simultaneously analyzed, and the roughness coefficient and stiffness gradient values ​​are quantitatively analyzed to form an original vector set containing the surface microtopological features. The curvature distribution is smoothed with a Gaussian kernel through a feature fusion to eliminate local distortion. The roughness features are adaptively mapped to the curvature curve, and the stiffness gradient is superimposed to construct a three-dimensional topological map. Finally, based on material database matching and verification, a normalized reference feature set containing radial curvature, circumferential texture, and axial stiffness is formed. This fully characterizes the geometric morphology and mechanical properties of the clamped object and provides the dynamic fitting module with precise constraints covering multiple degrees of freedom. This ensures that the clamping process can adaptively adjust the opening and closing degree, contact mode, and preload parameters to achieve stable and reliable clamping under material thermal deformation conditions. The entire process realizes closed-loop transformation from physical contact information to time-varying clamping parameters, enabling the clamping system to have real-time dynamic adaptation capabilities for complex curved surface materials.

[0076] Example 4: Based on Example 3, the initial contact submodule provided by this embodiment of the present invention includes:

[0077] The force field feature base construction unit is responsible for constructing the original contact force field matrix using the three-dimensional force data collected by the touch pressure sensor array. Its row vectors correspond to the distribution of spatial sampling points, and the column vectors record the contact force amplitude at each sampling time step. The extreme point groups in the matrix that significantly deviate from the mean constitute the characteristic observation set of the surface macroscopic curvature.

[0078] The mechanical feature space mapping unit is responsible for performing a singular value decomposition operation on the original contact force field matrix to obtain a set of eigenvectors representing the main directions of force. The two orthogonal vectors with the largest eigenvalues ​​correspond to the mechanical response modes in the directions of maximum and minimum curvature of the filter bag keel surface, respectively. The plane formed by the two main direction vectors is the curvature feature reference plane.

[0079] The curvature extreme value trajectory extraction unit is responsible for obtaining a cluster of characteristic curves along the two main curvature directions by tracing the spatial evolution path of the force field extreme points within the characteristic reference plane. The density distribution of the extreme points on each characteristic curve reflects the severity of the curvature change in that direction, and the inverse of the extreme point spacing constitutes the discrete curvature sampling sequence.

[0080] The continuous curvature field reconstruction unit is responsible for performing smooth interpolation of the discrete curvature sampling sequence based on energy minimization to generate a continuous curvature function distributed along the main curvature direction; during the interpolation process, the curvature mutation characteristics that match the distribution of the initial force field extreme points are automatically retained; and the continuous curvature function is converted into a set of curvature radius distribution parameters, where the anisotropic characteristics are characterized by the ratio of the radii of the two main directions.

[0081] The working principle and beneficial effects of the above technical solution are as follows: the force field feature base construction unit of this embodiment uses the three-dimensional force data collected by the touch pressure sensor array to construct the original contact force field matrix, whose row vectors correspond to the distribution positions of the spatial sampling points, and the column vectors record the contact force amplitudes at each sampling time step; the extreme point group that significantly deviates from the mean in the matrix constitutes a characteristic observation set of the surface macroscopic curvature; the mechanical feature space mapping unit performs a singular value decomposition operation on the original contact force field matrix to obtain a set of characteristic vectors representing the main directions of the force; the two orthogonal vectors with the largest eigenvalues ​​correspond to the mechanical response modes in the directions of maximum and minimum curvature of the filter bag keel surface, respectively, and the plane formed by the two main direction vectors is the curvature. Characteristic reference plane; the curvature extreme value trajectory extraction unit obtains a cluster of characteristic curves along the two main curvature directions by tracing the spatial evolution path of the force field extreme points within the characteristic reference plane; the density distribution of extreme value points on each characteristic curve reflects the severity of the curvature change in that direction, and the inverse of the extreme value point spacing constitutes a discrete curvature sampling sequence; the continuous curvature field reconstruction unit performs smooth interpolation based on energy minimization on the discrete curvature sampling sequence to generate a continuous curvature function distributed along the main curvature direction; the curvature mutation characteristics that match the initial force field extreme value point distribution are automatically retained during the interpolation process; the continuous curvature function is converted into a curvature radius distribution parameter set, where the anisotropic characteristics are characterized by the radius ratio of the two main directions. The initial contact submodule of the above scheme achieves parametric modeling of the macroscopic curvature characteristics of the filter bag keel surface through systematic mechanical-geometric field interaction processing. The original force field matrix containing the dynamic characteristics of the contact force is formed through the spatial-temporal synchronous acquisition of the net value data of the touch pressure sensor. The mechanical principal direction extraction algorithm based on eigenvalue decomposition maps the three-dimensional force distribution into a curvature characteristic plane with clear physical meaning. The extreme value trajectory tracking mechanism implemented in the characteristic reference plane coordinate system ensures the strict correspondence between the curvature feature extraction process and the physical properties of the material surface. The continuity reconstruction method combined with the energy optimization criterion converts the discrete mechanical observations into a set of geometric parameters that can be processed analytically. The final output curvature radius distribution parameter set not only retains the spatial gradient characteristics of the original contact force field, but also fully characterizes the anisotropic properties of the macroscopic curvature of the material surface through the quantization of the principal direction ratio. This forms a closed-loop conversion link from contact mechanical signals to geometric characteristic parameters, and its output parameters can directly support the microscopic feature fusion calculation of the subsequent surface topology quantization module.

[0082] Example 5: Based on Example 3, the feature fusion submodule provided in this embodiment of the present invention includes:

[0083] The curvature basis optimization unit is responsible for receiving the parameters of the curvature radius distribution and using an adaptive bandwidth Gaussian kernel to perform non-uniform smoothing on the discrete curvature values. The smoothing process retains the main curvature direction weights determined by the singular value decomposition of the initial contact force field, uses a larger bandwidth along the maximum curvature direction to maintain macroscopic features, and uses a smaller bandwidth along the minimum curvature direction to enhance detail retention. The corrected curvature distribution forms a topological basis mesh.

[0084] The roughness feature mapping unit is responsible for spatially registering the roughness coefficient sequence according to the curvature feature plane. It dynamically adjusts the roughness weight allocation strategy based on the curvature gradient change rate. It uses linear weighted mapping in areas with gentle curvature changes and enables nonlinear interpolation to ensure texture continuity in areas with sudden curvature changes. The generated composite curve carries both geometric curvature and physical roughness information.

[0085] The stiffness dimension nested unit is responsible for discretizing the stiffness gradient value hierarchically along the axial direction, with each level corresponding to a specific stiffness interval. A three-dimensional cylindrical coordinate system is established on the curvature-roughness composite curve, and the corresponding level is selected for axial projection according to the local stiffness value to form a three-dimensional topological unit with depth information.

[0086] The working principle and beneficial effects of the above technical solution are as follows: The curvature basis optimization unit of this embodiment receives the parameters of the curvature radius distribution and uses an adaptive bandwidth Gaussian kernel to perform non-uniform smoothing on the discrete curvature values. The smoothing process retains the principal curvature direction weights determined by the singular value decomposition of the initial contact force field, uses a larger bandwidth along the maximum curvature direction to maintain macroscopic features, and uses a smaller bandwidth along the minimum curvature direction to enhance detail preservation. The modified curvature distribution forms a topological basis mesh. The roughness feature mapping unit spatially aligns the roughness coefficient sequence according to the curvature feature plane, dynamically adjusts the roughness weight allocation strategy based on the curvature gradient change rate, uses linear weighted mapping in areas with gentle curvature changes, and enables nonlinear interpolation to ensure texture continuity in areas with sudden curvature changes. The generated composite curve carries both geometric curvature and physical roughness information. The stiffness dimension nesting unit discretizes the stiffness gradient values ​​hierarchically along the axial direction, with each level corresponding to a specific stiffness interval. A three-dimensional cylindrical coordinate system is established on the curvature-roughness composite curve, and the corresponding level is selected according to the local stiffness value for axial projection, forming a three-dimensional topological unit with depth information. The feature fusion submodule of the above scheme realizes the unified characterization of the multi-dimensional features of the filter bag keel surface through hierarchical parameter coupling; the curvature base optimization unit establishes the benchmark framework of the macro-morphology, and retains the geometric constraints obtained by mechanical feature decomposition through adaptive smoothing in the main direction; the roughness feature mapping unit embeds the micro-texture information into the curvature base grid, maintains feature continuity through gradient-sensitive weight allocation, and ensures the spatial consistency of geometric curvature and physical roughness; the stiffness dimension nesting unit expands the axial stiffness dimension on the two-dimensional curvature-roughness plane, and constructs a three-dimensional parameter space using hierarchical discretization of stiffness gradients to form a complete feature body covering radial curvature, circumferential texture, and axial stiffness. The final generated three-dimensional topological unit satisfies the following requirements: curvature distribution dominates macro-morphology modeling, roughness coefficient modifies local microstructure, and stiffness gradient constrains material response characteristics. The three are spatially aligned and parameter coupled to form an integrated feature model with multi-physical field characterization capabilities.

[0087] Example 6: Based on Example 2, the dynamic fitting module provided by the embodiment of the present invention includes:

[0088] The radial geometry adaptation submodule is responsible for establishing a jaw motion model based on the curvature radius distribution parameters provided by the contact sensing module, extracting the characteristic radius extreme value sequence in the main curvature direction, and driving the multi-link mechanism to generate the corresponding variable diameter motion trajectory. During the motion process, the matching degree of the clamping arc pressure gradient and the reference curvature distribution is compared in real time, and incremental compensation is used to gradually approach the optimal opening and closing degree, thereby determining the jaw posture parameters that minimize the overall curvature deviation.

[0089] The circumferential contact optimization submodule is responsible for dynamically configuring the contact mode within the radial geometric framework based on the spatial distribution characteristics of the roughness coefficient. For low-roughness areas, a continuous and uniform clamping pressure field is generated, and an integral contact pad is used for the corresponding surface. For high-roughness sections, a discrete lattice force pattern is converted, and the layout density of each contact point is negatively correlated with the local roughness amplitude. The optimized pressure distribution forms a conformal mapping between the surface texture characteristics and the contact stress field.

[0090] The axial load balancing submodule is responsible for superimposing preload compensation on the basis of radial-circumferential constraints according to the axial distribution law of the stiffness gradient value. A stiffness piecewise function is constructed along the keel axis, and the thermal expansion compensation amount is independently calculated in each stiffness interval. The preload loading curve follows the stiffness gradient change trend, adopting a linear growth strategy in the low stiffness area and implementing saturation limiting control in the high stiffness area.

[0091] The working principle and beneficial effects of the above technical solution are as follows: the radial geometry adaptation submodule of this embodiment establishes a jaw motion model based on the parameters of the curvature radius distribution provided by the contact sensing module, extracts the characteristic radius extreme value sequence in the main curvature direction, and drives the multi-link mechanism to generate the corresponding variable diameter motion trajectory; during the movement, the matching degree of the clamping arc surface pressure gradient and the reference curvature distribution is compared in real time, and the incremental compensation is used to gradually approach the optimal opening and closing degree, and the jaw posture parameters that minimize the overall curvature deviation are determined; the circumferential contact optimization submodule implements the dynamic configuration of the contact mode in the radial geometric framework according to the spatial distribution characteristics of the roughness coefficient: for the low roughness area, the dynamic configuration of the contact mode is realized; A continuous and uniform clamping pressure field is achieved, with an integral contact pad for the corresponding curved surface. High-roughness sections are converted to a discrete lattice force pattern, with the layout density of each contact point negatively correlated with the local roughness amplitude. The optimized pressure distribution ensures a conformal mapping between the surface texture characteristics and the contact stress field. The axial load balancing submodule superimposes preload compensation on the radial-circumferential constraint based on the axial distribution of the stiffness gradient. A piecewise stiffness function is constructed along the keel axis, independently calculating thermal expansion compensation within each stiffness interval. The preload loading curve follows the stiffness gradient variation trend, adopting a linear growth strategy in low-stiffness regions and saturation limiting control in high-stiffness regions. The radial geometry adaptation submodule of the above scheme achieves topological tracking of the profile of the irregular cross-section by establishing a curvature radius-driven kinematic model. The extraction of a characteristic radius extreme value sequence ensures that the variable diameter motion trajectory accurately matches the main curvature characteristics of the clamped object. The introduction of an incremental compensation mechanism enables the system to continuously correct the pose parameters through closed-loop feedback, ultimately converging the geometric deviation between the actual clamping arc surface and the target surface to within the allowable threshold. The circumferential contact optimization submodule establishes a multimodal contact response mechanism based on the spatial heterogeneity of the surface morphology. By intelligently switching between a continuous uniform pressure field and discrete point-matrix force application, it achieves dynamic coupling between the contact stress field and the surface texture characteristics. The self-adaptive pressure distribution strategy effectively eliminates the local stress concentration phenomenon caused by traditional rigid clamping. The axial load balance submodule adopts a segmented stiffness matching method to construct a gradient compensation system for the axial preload. The stiffness interval division ensures the positioning accuracy of the thermal deformation compensation, while the differentiated loading strategy (the organic combination of linear growth and saturation limiting) maintains the force flow balance of the entire clamping system.

[0092] Example 7: Figure 3 As shown, based on Example 1, the trajectory planning component provided by the embodiment of the present invention includes:

[0093] The initial module of 3D contour modeling is responsible for dynamically clamping the components through the multi-degree-of-freedom clamping mechanism to position the filter bag keel material. It then triggers the 3D contour detection program to perform a spiral scan along the axial direction of the filter bag keel, collect surface topology data, and generate a complete 3D point cloud model that includes diameter fluctuations, seam misalignment, and local deformation.

[0094] The welding path dynamic planning module is responsible for extracting the geometric features of the area to be welded based on the contour modeling results of the complete 3D point cloud model; the geometric features are input into the path generator to construct the reference spiral trajectory, and then the dynamic correction vector derived from the geometric deviation is superimposed;

[0095] In the axial dimension, the periodic variation trend of the filter bag keel material diameter is identified, and the welding interval is divided according to the extreme points of curvature. In the circumferential dimension, an azimuth compensation parameter table is established based on the seam misalignment. The correction vector includes: an axial component for adjusting the Z-axis displacement of the welding gun to track diameter fluctuations; a tangential component for adjusting the rotation phase of the welding gun to match the seam misalignment.

[0096] Real-time welding deviation compensation module, responsible for the sampling rate of the weld monitoring sensor to capture the molten pool morphology characteristics, molten pool morphology characteristics and thermal deformation drift;

[0097] The molten pool morphology characteristics are extracted through neural networks to extract the key parameters of weld width and weld depth; the thermal deformation drift is compared with the coordinate offset of the current welding point and the three-dimensional contour model.

[0098] The working principle and beneficial effects of the above technical solution are as follows: the dynamic clamping component of the 3D contour modeling initial module of this embodiment positions the filter bag keel material through a multi-degree-of-freedom clamping mechanism, then triggers a 3D contour detection program to perform a spiral scan along the axial direction of the filter bag keel, collects surface topology data, and generates a complete 3D point cloud model that includes diameter fluctuations, seam misalignment, and local deformation. The dynamic welding path planning module extracts the geometric features of the area to be welded based on the contour modeling results of the complete 3D point cloud model; the geometric features are input into the path generator to construct a reference spiral trajectory, and then the dynamic correction vector derived from the geometric deviation is superimposed. In the axial dimension, the periodic variation trend of the filter bag keel material diameter is identified, and the welding interval is divided according to the curvature extreme points. In the circumferential dimension, an azimuth compensation parameter table is established based on the seam misalignment. The correction vector includes: an axial component: adjusting the Z-axis displacement of the welding gun to track diameter fluctuations; a tangential component: adjusting the welding gun rotation phase to match the seam misalignment. The real-time welding deviation compensation module captures the weld pool morphology characteristics, weld pool morphology characteristics, and thermal deformation drift at the weld monitoring sensor sampling rate; the molten pool morphology characteristics are extracted through a neural network to extract the key parameters of weld width and depth; the thermal deformation drift is compared with the coordinate offset of the current weld point and the 3D contour model. The trajectory planning component of the above solution realizes intelligent trajectory control of the filter bag keel welding process through three precisely coordinated sub-modules. The 3D contour modeling initial module obtains the complete geometric characteristics of the filter bag keel material through 3D spiral scanning, and establishes an accurate 3D point cloud model that includes diameter fluctuations, seam misalignment, and local deformation, providing a high-precision reference for welding path planning. The dynamic welding path planning module quantifies material deviations (diameter fluctuation and seam misalignment) in both the axial and circumferential dimensions, constructs a reference trajectory, and superimposes adaptive correction vectors (Z-axis displacement compensation and rotational phase adjustment) to ensure the welding gun accurately tracks the target weld contour. The real-time welding deviation compensation module integrates weld pool morphology monitoring and thermal deformation drift detection to dynamically adjust welding parameters (such as penetration control) and trajectory compensation (such as coordinate offset correction) to suppress random errors and systematic deviations during the process.

[0099] In summary, this embodiment implements closed-loop path control for measurement, planning, and correction based on the analysis of the geometric features of the three-dimensional model, balancing preset accuracy with real-time deviation correction capabilities. Axial compensation ensures smooth tracking of diameter changes, circumferential compensation eliminates the effects of seam misalignment, and thermal deformation correction maintains welding stability under all operating conditions. The linkage between molten pool parameter feedback and path planning forms a continuously optimized process chain, making the system robust against material tolerances and processing disturbances. Ultimately, a full-process welding control system is formed, based on precise modeling, centered on dynamic correction, and guaranteed by real-time monitoring.

[0100] Example 8: Figure 4 As shown, based on Example 1, the quality detection component provided by the embodiment of the present invention includes:

[0101] The welding synchronous data acquisition module is responsible for capturing the radiation characteristic distribution of the weld area after welding, recording the dynamic behavior of the molten pool and the shape evolution characteristics of the solidification trajectory; capturing high-frequency vibration signals during the welding process, extracting the transient energy characteristics of bubble formation or material cracking; monitoring the axial temperature gradient of the welding heat-affected zone, and establishing a correlation model between heat input and material micro-deformation;

[0102] The multi-feature fusion analysis module is responsible for comparing the melting width fluctuation with the reference weld geometry data provided by the 3D contour modeling module. If the axial deviation exceeds the preset threshold, it is determined to be a risk of lack of fusion. The burst energy pulse peak of the transient energy feature is associated with the seam misalignment compensation amount in the path planning module. If the energy peak phase is consistent with the compensated azimuth angle deviation, a porosity defect warning is triggered. Combining the temperature gradient drop rate with the clamping pressure data fed back by the dynamic clamping component, if the temperature in the high-pressure area drops abnormally slowly, it is identified as uneven heat conduction caused by clamping deformation.

[0103] The defect closed-loop decision module is responsible for generating three types of quality criteria: axial deviation, energy peak, and abnormal temperature drop. When any of the criteria exceeds the limit, the welding process is suspended and the trajectory planning component is requested to replan the local path, and the optimized clamping pressure value is fed back to the dynamic clamping component.

[0104] The working principle and beneficial effects of the above technical solution are as follows: the welding synchronous data acquisition module of this embodiment captures the radiation characteristic distribution of the weld area after welding, records the dynamic behavior of the molten pool and the shape evolution characteristics of the solidification trajectory; captures the high-frequency vibration signal during the welding process, and extracts the transient energy characteristics of bubble formation or material cracking; monitors the axial temperature gradient of the welding heat-affected zone, and establishes a correlation model between heat input and material micro-deformation; the multi-feature fusion analysis module compares the melting width fluctuation with the reference weld geometry data provided by the three-dimensional contour modeling module. If the axial deviation exceeds the preset threshold, it is determined to be a risk of unfusion; The peak value of the sudden energy pulse of the transient energy characteristic is associated with the seam misalignment compensation in the path planning module. If the energy peak phase is consistent with the compensation azimuth deviation, a porosity defect warning is triggered. Combining the temperature gradient drop rate with the clamping pressure data fed back by the dynamic clamping component, if the temperature in the high-pressure area drops abnormally, it is identified as uneven heat conduction caused by clamping deformation. The defect closed-loop decision module generates three types of quality criteria: axial deviation, energy peak, and abnormal temperature drop. If any of these criteria exceeds the limit, the welding process is suspended and a request is made to the trajectory planning component for local path replanning. The optimized clamping pressure value is fed back to the dynamic clamping component. The quality inspection component of this solution acquires the weld radiation characteristics, high-frequency vibration signals, and temperature gradient distribution through the welding synchronous data acquisition module**, forming a raw data set of the dynamic behavior of the molten pool, defect signals, and heat-affected state. The multi-feature fusion analysis module establishes a mapping relationship, comparing melt width fluctuations with the baseline geometry data of the 3D contour modeling to determine the adequacy of fusion. The transient energy pulse peak is correlated with the path planning seam misalignment compensation to identify the tendency of porosity defects. The abnormal temperature drop is cross-analyzed with the dynamic clamping pressure data to locate the thermal conductivity imbalance caused by clamping deformation. The defect closed-loop decision module makes quality judgments based on the fusion analysis results and intervenes in the welding process in two ways: when the axial deviation, energy peak, or temperature gradient exceeds the limit, a local trajectory replanning instruction is triggered, and the trajectory planning component executes the corrective welding. When the temperature and pressure correlation indicates uneven clamping, the dynamic clamping component's pressure distribution strategy is adjusted to avoid conductive interference in subsequent weld segments.

[0105] Example 9: Figure 5 As shown, based on Examples 1 to 8, the control method of the robot for welding filter bag keels provided in the embodiment of the present invention includes the following steps:

[0106] S100: Through the multi-degree-of-freedom clamping mechanism, the clamping posture and force are adjusted in real time to match the filter bag keel materials with different diameters and surface conditions to achieve positioning;

[0107] S200: Before welding, 3D contour detection is used to scan the weld position of the filter bag keel material to generate the welding path; and the weld monitoring sensor is used to adjust the welding parameters in real time to compensate for the welding deviation caused by the tolerance of the filter bag keel material;

[0108] S300: During the welding process, multimodal sensing is used to evaluate weld quality in real time; when a defective weld is encountered, the system automatically pauses and reports correction instructions.

[0109] The working principle and beneficial effects of the above technical solution are as follows: First, this embodiment uses a multi-degree-of-freedom clamping mechanism to adjust the clamping posture and force in real time to match filter bag keel materials with different diameters and surface conditions to achieve positioning; secondly, before welding, three-dimensional contour detection is used to scan the welded position of the filter bag keel material to generate a welding path; and the welding parameters are adjusted in real time through the weld monitoring sensor to compensate for the welding deviation caused by the tolerance of the filter bag keel material; finally, during the welding process, multi-modal sensing is used to evaluate the weld quality in real time; when a defective weld is encountered, it automatically pauses and reports correction instructions. The combination of adaptive dynamic clamping with three-dimensional contour scanning and path planning in the above solution can maintain a high-precision positioning reference even when the metal material has diameter fluctuations, surface unevenness or local deformation, ensuring that the weld trajectory generation strictly matches the actual physical contour and avoiding welding path deviation due to clamping posture deviation. Path planning and real-time compensation rely on the stable clamping state provided. The geometric features obtained by three-dimensional scanning (such as axial diameter changes and circumferential seam misalignment) are used to dynamically adjust welding parameters (such as welding gun Z-axis tracking and rotational phase compensation), thereby eliminating systematic errors introduced by material tolerances in the initial stage of welding. Multimodal quality detection is closely linked to real-time sensor data (molten pool morphology, temperature field distribution), and combined with clamping feedback (such as whether the pressure distribution is balanced) to establish a real-time closed-loop assessment of weld quality. If a defect is detected, the system recalculates and corrects the trajectory based on the path planning capability, and optimizes the clamping force to form a full-process adaptive control chain of "clamping-scanning-welding-detection-adjustment".

[0110] Example 10: Figure 6 As shown, based on Examples 1 to 8, the structure of the robot for filter bag keel welding provided by the embodiments of the present invention includes: a robot base 1, a first rotation axis node 2, a second rotation axis node 3, a three-dimensional contour detection scanning probe 4, a transmission shaft 5, and a multi-degree-of-freedom clamping mechanism 6;

[0111] A multi-axis robotic arm is installed on the robot base 1 through the first rotation axis node 2 and the second rotation axis node 3. A three-dimensional contour detection scanning probe 4 is installed on the robotic arm at the end of the second rotation axis node 3. A controller is installed in the shell above the three-dimensional contour detection scanning probe 4 to store the control program of the robot; the robotic arm at the end of the second rotation axis node 3 is connected to one end of the transmission shaft 5, and the other end of the transmission shaft 5 is connected to the multi-degree-of-freedom clamping mechanism 6. The transmission shaft 5 is mechanically connected to each end of the multi-degree-of-freedom clamping mechanism 6.

[0112] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention's equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A robot used for welding filter bag keels, characterized in that: Include: The dynamic clamping component is responsible for adjusting the clamping posture and force in real time through the multi-degree-of-freedom clamping mechanism to match the filter bag keel materials with different diameters and surface conditions to achieve positioning; The trajectory planning component is responsible for scanning the welded position of the filter bag keel material using 3D contour detection before welding to generate the welding path; and adjusts the welding parameters in real time through the weld monitoring sensor to compensate for welding deviations caused by the tolerance of the filter bag keel material; The quality inspection component is responsible for real-time evaluation of weld quality during the welding process using multimodal sensing. It automatically pauses and reports correction instructions when a defective weld is encountered.

2. The robot for welding filter bag keels according to claim 1, characterized in that: Dynamic clamping assembly, including: The contact sensing module is responsible for detecting the initial contact force of the touch pressure sensor of the multi-degree-of-freedom clamping mechanism. It uses the strain gauge array to establish the surface topology map of the filter bag keel material, including the diameter characteristic parameters or curvature radius distribution and the surface state parameters or roughness coefficient and material stiffness gradient, forming the clamping reference feature set; The dynamic fitting module is responsible for performing three-stage fitting of radial fitting, circumferential fitting, and axial compensation on the clamping mechanism based on the benchmark feature set to obtain clamping parameters including opening and closing degree, pressure distribution, and preload force; The real-time coupling module is responsible for transmitting the clamping parameters to the trajectory planning component in real time as boundary conditions for weld path calculation. At the same time, it continuously monitors changes in the clamping state. When the characteristic offset of the filter bag keel material exceeds the threshold due to thermal deformation, the clamping parameter recalculation closed loop is triggered.

3. The robot for welding filter bag keels according to claim 2, characterized in that: in, Radial fitting dynamically adjusts the jaw opening and closing according to the diameter characteristic parameters, so that the curvature error between the clamping arc surface and the material meets the standard; The circumferential fitting adjusts the clamping pressure distribution according to the roughness coefficient, adopting the surface contact mode in the smooth section and switching to the lattice contact mode in the texture section; Axial compensation adjusts the axial preload through the material stiffness gradient data to prevent the filter bag keel material from axial movement in the welding heat affected zone.

4. The robot for welding filter bag keels according to claim 1, characterized in that: Contact sensing module, including: The initial contact submodule is responsible for ensuring that the multi-degree-of-freedom clamping mechanism contacts the surface of the filter bag keel material through the touch pressure sensor array. The strain gauge group collects the three-dimensional force distribution of the contact points to form an initial contact force field matrix. The extreme point distribution reflects the macroscopic curvature characteristics of the filter bag keel material surface. The main curvature direction is extracted through singular value decomposition to obtain the reference curvature radius distribution in the diameter characteristic parameter. The surface topology quantification submodule is responsible for constructing the microscopic morphology characteristics of the filter bag keel material surface based on the spatial gradient change of the initial contact force field and the dynamic response data of the high-frequency sampling strain gauge; Based on the correlation analysis between the initial contact force fluctuation frequency and amplitude, the roughness coefficient of the filter bag keel material surface state parameters is quantified; at the same time, the material stiffness gradient value is derived through the slope change rate of the force attenuation curve to form the original feature vector set; The feature fusion submodule is responsible for inputting the diameter feature parameters and surface state parameters into the feature fusion device. It first performs Gaussian kernel smoothing on the curvature radius distribution to eliminate local distortion. Then, it maps the roughness coefficient to the curvature distribution curve through adaptive weighting to form a topological map with texture features. The stiffness gradient value is embedded in the depth dimension of the topological map as an axial compensation factor. The benchmark calibration submodule is responsible for matching and verifying the topological map with the preset filter bag keel material database, and correcting the validity range of the characteristic parameters through residual analysis; when the coupling error between the curvature radius distribution and the roughness coefficient is less than the set threshold, the current parameter combination is locked as the clamping benchmark feature set. Each coordinate point in the topological map contains three types of normalized data: radial curvature, circumferential texture and axial stiffness.

5. The robot for welding filter bag keels according to claim 4, characterized in that: Initial contact submodule, including: The force field feature base construction unit is responsible for constructing the original contact force field matrix using the three-dimensional force data collected by the touch pressure sensor array. Its row vectors correspond to the distribution positions of the spatial sampling points, and the column vectors record the contact force amplitude at each sampling time step. The extreme point group that significantly deviates from the mean in the matrix constitutes the characteristic observation set of the surface macroscopic curvature; The mechanical feature space mapping unit is responsible for performing singular value decomposition on the original contact force field matrix to obtain a set of eigenvectors representing the main directions of the force; The curvature extreme value trajectory extraction unit is responsible for obtaining the characteristic curve clusters along the two main curvature directions by tracing the spatial evolution path of the force field extreme value points within the characteristic reference plane; The density distribution of extreme points on each characteristic curve reflects the degree of curvature change in that direction, and the inverse of the extreme point spacing constitutes a discrete curvature sampling sequence; The continuous curvature field reconstruction unit is responsible for performing smooth interpolation of the discrete curvature sampling sequence based on energy minimization to generate a continuous curvature function distributed along the main curvature direction; the interpolation process automatically retains the curvature mutation characteristics that match the distribution of the initial force field extreme points; The continuous curvature function is converted into a set of curvature radius distribution parameters, where the anisotropic characteristics are characterized by the ratio of the radii in the two main directions.

6. The robot for welding filter bag keels according to claim 5, characterized in that: The two orthogonal vectors with the largest eigenvalues ​​in the eigenvector group of the mechanical feature space mapping unit correspond to the mechanical response modes in the maximum curvature direction and the minimum curvature direction of the filter bag keel surface, respectively. The plane formed by the two main direction vectors is the curvature feature reference plane.

7. The robot for welding filter bag keels according to claim 2, characterized in that: Dynamic fitting module, including: The radial geometry adaptation submodule is responsible for establishing a jaw motion model based on the curvature radius distribution parameters provided by the contact sensing module, extracting the characteristic radius extreme value sequence in the main curvature direction, and driving the multi-link mechanism to generate the corresponding variable diameter motion trajectory. During the motion process, the matching degree of the clamping arc pressure gradient and the reference curvature distribution is compared in real time, and incremental compensation is used to gradually approach the optimal opening and closing degree, thereby determining the jaw posture parameters that minimize the overall curvature deviation. The circumferential contact optimization submodule is responsible for dynamically configuring the contact mode within the radial geometric framework based on the spatial distribution characteristics of the roughness coefficient. For low-roughness areas, a continuous and uniform clamping pressure field is generated, and an integral contact pad is used for the corresponding surface. For high-roughness sections, a discrete lattice force pattern is converted, and the layout density of each contact point is negatively correlated with the local roughness amplitude. The optimized pressure distribution forms a conformal mapping between the surface texture characteristics and the contact stress field. The axial load balancing submodule is responsible for superimposing preload compensation on the basis of radial-circumferential constraints according to the axial distribution law of the stiffness gradient value. A stiffness piecewise function is constructed along the keel axis, and the thermal expansion compensation amount is independently calculated in each stiffness interval. The preload loading curve follows the stiffness gradient change trend, adopting a linear growth strategy in the low stiffness area and implementing saturation limiting control in the high stiffness area.

8. The robot for welding filter bag keels according to claim 1, characterized in that: Trajectory planning components, including: The initial module of 3D contour modeling is responsible for dynamically clamping the components through the multi-degree-of-freedom clamping mechanism to position the filter bag keel material. It then triggers the 3D contour detection program to perform a spiral scan along the axial direction of the filter bag keel, collect surface topology data, and generate a complete 3D point cloud model that includes diameter fluctuations, seam misalignment, and local deformation. The welding path dynamic planning module is responsible for extracting the geometric features of the area to be welded based on the contour modeling results of the complete 3D point cloud model; the geometric features are input into the path generator to construct the reference spiral trajectory, and then the dynamic correction vector derived from the geometric deviation is superimposed; The real-time welding deviation compensation module is responsible for capturing the molten pool morphology characteristics, molten pool morphology characteristics and thermal deformation drift at the sampling rate of the weld monitoring sensor.

9. The robot for welding filter bag keels according to claim 1, characterized in that: Quality inspection components, including: The welding synchronous data acquisition module is responsible for capturing the radiation characteristic distribution of the weld area after welding, recording the dynamic behavior of the molten pool and the shape evolution characteristics of the solidification trajectory; capturing the high-frequency vibration signal during the welding process, and extracting the transient energy characteristics of bubble formation or material cracking; Monitor the axial temperature gradient in the heat-affected zone of welding and establish a correlation model between heat input and material micro-deformation; The multi-feature fusion analysis module is responsible for comparing the melting width fluctuation with the reference weld geometry data provided by the 3D contour modeling module. If the axial deviation exceeds the preset threshold, it is determined to be a lack of fusion risk. The peak value of the sudden energy pulse of the transient energy feature is associated with the seam misalignment compensation in the path planning module. If the energy peak phase is consistent with the compensation azimuth deviation, a porosity defect warning is triggered. Combining the temperature gradient drop rate with the clamping pressure data fed back by the dynamic clamping assembly, if the temperature in the high-pressure area drops abnormally slowly, it is identified as uneven heat conduction caused by clamping deformation. The defect closed-loop decision module is responsible for generating three types of quality criteria: axial deviation, energy peak, and abnormal temperature drop. When any of the criteria exceeds the limit, the welding process is suspended and the trajectory planning component is requested to replan the local path, and the optimized clamping pressure value is fed back to the dynamic clamping component.

10. A control method for a robot used for welding filter bag keels, characterized in that: The following steps are involved: Through the multi-degree-of-freedom clamping mechanism, the clamping posture and force are adjusted in real time to match the filter bag keel materials with different diameters and surface conditions to achieve positioning; Before welding, 3D contour detection is used to scan the weld position of the filter bag keel material to generate the welding path; and the welding parameters are adjusted in real time through the weld monitoring sensor to compensate for the welding deviation caused by the tolerance of the filter bag keel material; During the welding process, multimodal sensing is used to evaluate the weld quality in real time; when a defective weld is encountered, the process is automatically paused and correction instructions are reported.

Images