A machining system and machining method for hub ring production

Abstract

This invention discloses a processing system and method for producing wheel hub steel rims, belonging to the field of intelligent manufacturing equipment technology for automotive parts. The system includes a processing platform and loading, welding, flattening, grinding, and unloading stations arranged along the processing flow. It also includes an integrated production control center that communicates with each station via an industrial fieldbus. The loading station is equipped with an automatic loading mechanism, a conveying device, and an infrared position sensor; a flexible clamping and transfer module is provided between the welding and grinding stations; the welding station is equipped with an arc-surface centering clamping mechanism and an automatic welding device with a detection module; the grinding station is equipped with a dual-end-face synchronous grinding device and a grinding force feedback component. The control center includes data acquisition, visual perception, time-series collaborative scheduling, and process parameter adaptive optimization units. This invention achieves fully automated production of wheel hub steel rims, featuring high-precision time-series collaboration across multiple stations, adaptive optimization of welding parameters, and dual-end-face synchronous precision grinding.

Description

Technical Field

[0001] This invention belongs to the field of intelligent manufacturing equipment technology for automotive parts, specifically a processing system and method for producing wheel hub steel rims. Background Technology

[0002] As a key load-bearing component and safety element of the automotive chassis system, the processing quality of steel wheel rims directly affects the overall vehicle's driving safety and reliability. Currently, the production process of steel wheel rims typically includes multiple steps such as rolling, welding, flaring, flattening, end face grinding, and inspection. With the increasing demands for lightweight and mass production in automobiles, existing wheel rim processing production lines still have significant shortcomings in terms of automation and intelligence.

[0003] Insufficient precision in multi-process collaborative control makes it difficult to match production cycle time. In existing production lines, stations such as feeding, clamping, welding, and grinding are mostly managed by independent controllers. There is a lack of a unified timing coordination mechanism between stations, which often leads to waiting or conflicts when workpieces are transferred between stations, resulting in low equipment utilization and difficulty in optimizing the overall production cycle time.

[0004] Welding process parameters rely on manual experience, resulting in poor quality consistency. In the welding process of wheel rims and spokes, the setting of welding current, voltage, and welding torch movement speed largely depends on the operator's experience. When there are slight deviations in the incoming workpiece or changes in the welding heat accumulation state, it is impossible to adjust the process parameters in real time, which easily leads to defects such as incomplete welding, over-welding, or uneven welds.

[0005] The precision and consistency of end face grinding are difficult to guarantee. After welding, the two ends of the wheel hub steel rim need to be finely ground to ensure assembly accuracy. Existing grinding equipment mostly processes one end face sequentially or drives both ends independently. It is difficult to ensure that the grinding amount on both sides is strictly consistent, and there is a lack of real-time grinding force feedback and compensation mechanism, resulting in excessive deviations in the flatness and symmetry of the end face, which affects subsequent assembly.

[0006] The entire process has a low level of automation and involves a lot of manual intervention. In existing systems, the transfer of workpieces between different workstations still largely relies on manual labor or simple robotic arms. The clamping and positioning accuracy is greatly affected by human factors and poses safety hazards, making it difficult to meet the needs of intelligent manufacturing with large-scale and high consistency requirements.

[0007] Therefore, there is an urgent need for a fully automated machining system and corresponding control methods that can achieve high-precision collaborative control of multiple workstations, adaptive optimization of welding process parameters, and synchronous precision grinding of both end faces. Summary of the Invention

[0008] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a processing system and processing method for the production of wheel hub steel rims.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a processing system for producing wheel hub steel rims, comprising a processing platform and, sequentially arranged along the processing flow, a loading station, a welding station, a flattening station, a grinding station, and a unloading station on the processing platform; and an integrated production control center, which is communicatively connected to the execution units and sensing units in the loading station, welding station, flattening station, grinding station, and unloading station via an industrial fieldbus or industrial Ethernet, forming a closed-loop control network for the entire process; wherein: the loading station includes an automatic loading mechanism, a conveying device, and a positioning detection component; the conveying device is provided with multiple limit plates spaced apart along the conveying direction, and the positioning detection component... The system includes infrared position sensors installed at the entrance of each workstation to detect the position status of the wheel hub steel rim on the conveying device in real time and feed the position signal back to the integrated production control center; a flexible clamping and transfer module is provided between the welding station and the grinding station; the flexible clamping and transfer module includes a gantry-type transfer rail mounted above the processing platform, a lifting drive mechanism installed on the transfer rail, and a clamping execution mechanism suspended by the lifting drive mechanism; the lifting drive mechanism includes a first servo motor and a first ball screw pair driven by the first servo motor; the clamping execution mechanism includes a lifting frame fixedly connected to the nut seat of the first ball screw pair, and a first ball screw pair horizontally mounted on the lifting frame. The system includes a linear drive element and a second linear drive element, a front push plate driven by the first linear drive element, and a clamping plate driven by the second linear drive element. The front push plate and the clamping plate are arranged opposite each other, forming a clamping gap adapted to the thickness of the wheel hub steel ring to achieve flexible clamping and precise handling of the wheel hub steel ring. The welding station includes a welding positioning table, an arc-shaped centering clamping mechanism, and an automatic welding device. The arc-shaped centering clamping mechanism includes multiple arc-shaped clamping blocks arranged circumferentially around the center of the welding positioning table and a third linear drive element that drives the arc-shaped clamping blocks to retract radially and synchronously. The automatic welding device includes a second servo motor, a second ball screw pair driven by the second servo motor, and a component mounted on... The second ball screw pair has a movable welding head on its sliding seat; the welding head is equipped with a welding power supply and welding parameter real-time detection module, which is used to collect current, voltage and arc sound signals during the welding process; the grinding station includes a positioning table, a double-end face synchronous grinding device and a grinding force feedback component; the double-end face synchronous grinding device includes a dual-axis drive motor, a first horizontal screw and a second horizontal screw synchronously driven by the dual-axis drive motor through a bevel gear set, a first moving seat and a second moving seat respectively threaded with the first horizontal screw and the second horizontal screw, and a first rotary polishing mechanism and a second rotary polishing mechanism respectively installed on the first moving seat and the second moving seat;The first horizontal lead screw and the second horizontal lead screw have opposite thread directions, so that the first moving seat and the second moving seat move synchronously towards each other or backwards in the horizontal direction under the drive of the dual-axis drive motor; both the first rotary polishing mechanism and the second rotary polishing mechanism include a rotary drive motor, a polishing spindle driven by the rotary drive motor, and a grinding head mounted on the end of the polishing spindle; the grinding force feedback component includes a pressure sensor mounted on the first moving seat and / or the second moving seat, used to detect the contact pressure between the grinding head and the end face of the wheel hub steel rim in real time; the integrated production control center includes a data acquisition unit, a vision perception unit, a time-series collaborative scheduling unit, and a process parameter adaptive optimization unit. The data acquisition unit is used to receive and preprocess signals from each sensing unit. The vision perception unit includes an industrial area array camera mounted above the welding station and the grinding station. The time-series collaborative scheduling unit is used to issue action commands to each execution unit based on a preset station cycle time table. The process parameter adaptive optimization unit is used to perform fusion analysis of multi-dimensional parameters of the welding process and generate welding parameter fine-tuning commands.

[0010] Compared with the prior art, the beneficial effects of the present invention are: This invention achieves fully automated production of wheel hub steel rims, eliminating manual intervention. Through an integrated production control center, the various workstations, including automatic feeding, flexible clamping and transfer, welding, flattening, grinding, and unloading, are uniformly scheduled. This integrates the originally scattered and independent processes into a continuous automated production flow, significantly improving production efficiency and reducing labor costs and safety hazards.

[0011] This invention achieves high-precision timing coordination across multiple workstations, optimizing production cycle time. The integrated production control center, based on real-time trigger signals from infrared position sensors and image recognition results from visual perception units, performs millisecond-level coordinated scheduling of each execution unit through a preset workstation cycle time table. This achieves seamless integration of loading, transfer, and processing actions, minimizing equipment idle time and waiting time.

[0012] This invention achieves adaptive optimization of welding process parameters, improving weld quality consistency. By integrating the temporal characteristics of welding electrical signals with the visual characteristics of the molten pool image, a real-time evaluation and feedback mechanism for welding heat input is constructed. This mechanism enables dynamic fine-tuning of the welding torch movement speed and welding current during the welding process, effectively suppressing welding defects caused by fluctuations in the incoming workpiece material or heat accumulation effects, and ensuring uniform weld formation and stable penetration depth.

[0013] This invention achieves synchronous precision grinding of both end faces, ensuring the accuracy and consistency of end face machining. It employs a structure where a single motor synchronously drives two counter-rotating lead screws via a bevel gear set, ensuring that the grinding heads on both sides feed synchronously with strictly equal speeds and strokes. Combined with real-time grinding force feedback and closed-loop compensation, it achieves a high degree of consistency in the grinding amount on both ends of the wheel hub steel rim, effectively guaranteeing the flatness of the end faces and the accuracy of subsequent assembly.

[0014] The flexible clamping and transfer module of this invention has strong adaptability and high positioning accuracy. Through the cooperation of a ball screw lifting mechanism driven by a servo motor and a clamping plate driven by dual cylinders, it can adapt to the clamping requirements of wheel hub steel rims of different specifications. The clamping action is smooth, and the positioning repeatability is high, avoiding collisions and deformation of the workpiece during the transfer process.

[0015] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. The embodiments of this application will provide a detailed description and understanding of the application. Attached Figure Description

[0016] Figure 1 This is a top view of the overall system layout of an embodiment of the present invention, showing the spatial relationship between the processing platform, conveying device, loading station, welding station, flattening station, grinding station, unloading station and flexible clamping and transfer module. Figure 2 This is a schematic diagram of the functional architecture and signal flow of the integrated production control center described in this invention, showing the communication connection relationship between the data acquisition unit, visual perception unit, time-series collaborative scheduling unit, process parameter adaptive optimization unit and each workstation execution unit and sensing unit. Figure 3 This is a structural schematic diagram of the flexible clamping and transfer module described in this invention, showing the assembly relationship of the gantry transfer track, the first servo motor, the first ball screw pair, the lifting frame, the first linear drive element, the front push plate, the second linear drive element, and the clamping plate. Figure 4 This is a schematic diagram of the welding station of the present invention, showing the layout of the welding positioning table, the arc-shaped clamping block, the third linear drive element, the second servo motor, the second ball screw pair and the movable welding head; Figure 5 This is a schematic diagram of the structure of the dual-end-face synchronous polishing device of the present invention, showing the mechanical structure of the dual-axis drive motor, bevel gear set, first horizontal lead screw, second horizontal lead screw, first moving seat, second moving seat, first rotary polishing mechanism, second rotary polishing mechanism and pressure sensor; Figure 6This is a schematic diagram of the data processing flow of the adaptive optimization unit for process parameters described in this invention, illustrating the logical flow of welding electrical signal temporal feature extraction, molten pool image visual feature extraction, feature-level fusion, quality prediction, and parameter decision-making.

[0017] In the diagram: 1. Processing platform; 2. Conveying device; 3. Limiting plate; 4. Loading station; 5. Welding station; 6. Flattening station; 7. Grinding station; 8. Unloading station; 9. Infrared position sensor; 10. Flexible clamping and transfer module; 11. Gantry transfer track; 12. First servo motor; 13. First ball screw pair; 14. Lifting frame; 15. First linear drive element; 16. Front push plate; 17. Second linear drive element; 18. Clamping plate; 19. Welding positioning table; 20. Arc-shaped clamping block; 21. Third linear drive element; 22. Second servo motor; 23. Second ball screw pair; 24. Welding head; 25. Welding power supply; 26. Real-time welding parameter detection module; 27. 28. Grinding positioning stage; 29. ​​Dual-axis drive motor; 30. Bevel gear set; 31. First horizontal lead screw; 32. Second horizontal lead screw; 33. First moving seat; 34. Second moving seat; 35. First rotary polishing mechanism; 36. Second rotary polishing mechanism; 37. Rotary drive motor; 38. Polishing spindle; 39. Grinding head; 40. Pressure sensor; 41. Rotary worktable; 42. Integrated production control center; 43. Data acquisition unit; 44. Visual perception unit; 45. Time-series collaborative scheduling unit; 46. Process parameter adaptive optimization unit; 47. Industrial area array camera; 48. Feature extraction submodule; 49. State fusion submodule; 50. Parameter decision submodule; 61. Quality prediction model. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 See Figures 1 to 6 This embodiment takes the automated processing of a certain type of steel passenger car wheel hub rim (made of low carbon steel, with a rim diameter of about 16 inches) as an example to illustrate the specific implementation of the present invention.

[0020] System Overall Layout and Initialization: Before system startup, the integrated production control center 41 performs communication handshakes and status self-checks with each workstation execution unit via industrial Ethernet. After the self-check is completed, the control center loads preset process recipe parameters, including the operating speed of the conveyor, the lifting stroke and transfer trajectory coordinates of the flexible clamping and transfer module, the reference value of the welding head moving speed, the reference value of the welding current, the feed speed of the double-end grinding device, and the target threshold of the grinding force.

[0021] S1 Implementation Details: Intelligent Feeding and Initial Positioning: The automatic feeding mechanism places the wheel hub steel rim blank horizontally between two adjacent limiting plates 3 of the conveying device 2. The conveying device 2 is driven by a variable frequency motor and conveys forward at a constant speed of 0.3 m / s. A first infrared position sensor 9 (through-beam photoelectric sensor) is installed 200 mm in front of the entrance of welding station 5. When the leading edge of the wheel hub steel rim blocks the sensor beam, the sensor outputs a high-level trigger signal to the data acquisition unit 42 of the control center. After receiving the trigger signal, the control center sends a deceleration command to the variable frequency motor, and the conveying device 2 decelerates to zero speed within a braking distance of 200 mm, ensuring that the wheel hub steel rim stops precisely at the feeding waiting position of welding station 5.

[0022] After the conveyor 2 stops, the control center triggers the first industrial area array camera 46 (resolution 2048-2048, equipped with a telecentric lens) mounted above the welding station 5 to acquire images. The vision perception unit 43 performs edge detection and Hough circle transformation on the acquired images to identify the outer circle contour and center pixel coordinates of the wheel hub steel rim. Combining this with pre-calibrated camera intrinsic parameters and hand-eye calibration matrix, the pixel coordinates are converted into physical coordinates (X, Y) in the processing platform coordinate system. This coordinate information serves as the gripping reference for the subsequent flexible clamping and transfer module 10.

[0023] S2 Implementation Details: Flexible Clamping and Precise Transfer: The control center sends the workpiece coordinates (X, Y) to the servo controller of the flexible clamping and transfer module 10. The horizontal transfer motor on the gantry transfer track 11 drives the clamping actuator to move along the X and Y directions to directly above the target coordinates. The first servo motor 12 (with brake) starts, driving the first ball screw pair 13 (lead 5mm, accuracy C5 grade) to rotate forward, and the lifting frame 14 descends at a speed of 50mm / s. When the proximity switch installed at the lower end of the lifting frame 14 detects that the distance to the upper surface of the wheel hub steel ring reaches the set value (e.g., 10mm), the first servo motor 12 stops.

[0024] The first linear drive element 15 uses a double-acting cylinder (40mm diameter), with its piston rod end connected to the front push plate 16. The control center outputs a signal to the solenoid valve, and the cylinder front push plate 16 extends horizontally with a gas pressure of 0.5MPa. The initial gap between the front push plate 16 and the clamping plate 18 is slightly larger than the wheel hub rim thickness (e.g., a gap of 8mm and a rim thickness of 6mm). After the front push plate 16 is in place, the second linear drive element 17 (single-acting cylinder, 32mm diameter, with built-in spring return) actuates, and its piston rod pushes the clamping plate 18 to translate towards the front push plate 16, clamping the wheel hub rim between the front push plate 16 and the clamping plate 18. The clamping force is determined by the cylinder pressure and the clamping area. In this embodiment, the clamping force is set to approximately 200N, which is sufficient to overcome the inertial force during the transfer process and will not cause plastic deformation of the rim.

[0025] After clamping, the first servo motor 12 reverses, and the lifting frame 14 rises at a speed of 80 mm / s, detaching from the conveying device 2. The horizontal transfer motor drives the clamping actuator to move along a preset trajectory to directly above the welding positioning table 19. After the control center performs minor position compensation based on the center coordinates of the welding positioning table 19, the lifting frame 14 descends, precisely placing the wheel hub steel ring onto the support surface of the welding positioning table 19. After placement, the first linear drive element 15 and the second linear drive element 17 reset sequentially, the clamping plate releases the wheel hub steel ring, and one transfer cycle is completed. In this embodiment, the total transfer cycle from gripping to placement is approximately 8 seconds.

[0026] S3 Implementation Details: Adaptive Welding Process. After the wheel hub steel rim is placed on the welding positioning table 19, the control center outputs a clamping signal to the third linear drive element 21 (the third cylinder, three sets in total, evenly distributed circumferentially). The three sets of arc-shaped clamping blocks 20 (the inner arc surface curvature matches the outer circle of the wheel hub steel rim) synchronously retract radially under the cylinder drive, clamping the wheel hub steel rim from the outer circumference to the center, achieving automatic centering and clamping. The centering accuracy can reach 0.1mm.

[0027] The welding head 24 employs a gas metal arc welding (GMAW) torch, which is mounted on a sliding seat of the second ball screw assembly 23. A second servo motor 22 (with an absolute encoder) drives the second ball screw assembly 23 (10mm lead) to rotate, causing the welding torch to move at a preset reference speed (e.g., 600mm / min) along the annular weld seam where the wheel hub rim and spokes meet. The welding power supply 25 is a digital inverter welding machine with a remote current adjustment interface.

[0028] During the welding process, the welding parameter real-time detection module 26 collects the welding current and arc voltage in real time through a Hall current sensor and a voltage dividing resistor network, with a sampling frequency of 10 kHz; at the same time, the first industrial area array camera 46 continuously collects images of the molten pool area at a frequency of 30 frames per second. The data acquisition unit 42 anti-aliasing filters the current and voltage signals and then down-samples them to 100 Hz, calculates the root mean square current Irms, root mean square voltage Urms within each 100 ms window, and the short-time energy Earc of the arc sound signal (collected by a microphone), and constructs a time-domain feature vector Ft = [Irms, Urms, Earc].

[0029] The visual perception unit 43 preprocesses the molten pool area image (grayscaling, contrast enhancement), extracts the molten pool length Lp, molten pool width Wp, and the average gray value Gm of the molten pool center, and constructs an image feature vector Fi = [Lp, Wp, Gm].

[0030] The state fusion sub-module 48 in the process parameter adaptive optimization unit 45 aligns Ft and Fi according to the time stamp and then splices them to generate a fusion feature Ffusion = [Ft, Fi]. The parameter decision sub-module 49 has a quality prediction model 50 built based on support vector regression (SVR) or a shallow neural network. This model is trained with historical welding data (including weld cross-section metallographic data under different heat input conditions), with Ffusion as the input and the predicted weld penetration value Hpre and the predicted weld width value Wpre as the output.

[0031] The control center presets a target penetration range [Hmin, Hmax] and a target weld width range [Wmin, Wmax]. When Hpre < Hmin or Wpre < Wmin, it is determined that the heat input is insufficient, and the parameter decision sub-module 49 outputs a current increase instruction (step adjustment amount I = +10 A) and a welding torch deceleration instruction (step adjustment amount v = -20 mm / min); conversely, when Hpre > Hmax or Wpre > Wmax, it is determined that the heat input is excessive, and a current decrease instruction and a welding torch acceleration instruction are output. The adjustment instructions are sent by the control center to the welding power source 25 and the second servo motor 22 to form a closed-loop control of the welding quality. In this embodiment, the parameter adjustment period is 500 ms, and the welding heat input can be stabilized within the target window within 3 adjustment cycles.

[0032] After welding is completed, the control center controls the welding torch to return to the origin, the third cylinder resets to release the arc-shaped clamping block 20, and the flexible clamping and transfer module 10 re-clamps the welded wheel steel ring and places it back on the conveying device 2 to enter the next process.

[0033] S4 Implementation Details: Flattening and Shaping: Conveying device 2 transports the wheel hub steel rim to flattening station 6. The flattening device uses a pair of hydraulic rollers arranged vertically. The control center adjusts the gap between the upper and lower rollers according to the target thickness value in the current production formula through servo electric cylinders installed at both ends of the upper roller. After the wheel hub steel rim enters the gap, the hydraulic system rolls the wheel hub steel rim at a set pressure (e.g., 50MPa) to eliminate warping deformation generated during welding and forming, ensuring the axial flatness of the wheel body.

[0034] S5 Implementation Details: Simultaneous Precision Grinding of Both End Faces: The flattened wheel hub steel rim is conveyed to the grinding station 7 by the conveyor device 2. After the second infrared position sensor is triggered, the conveyor device 2 stops. The flexible clamping and transfer module 10 transports the wheel hub steel rim to the grinding positioning table 27. The bottom of the grinding positioning table 27 is equipped with a rotary worktable 40 driven by a fourth servo motor. The center of the rotary worktable 40 is equipped with an expansion-type centering clamp, which achieves centering and clamping from the inner hole of the wheel hub steel rim.

[0035] The dual-end synchronous grinding device is activated. The output shaft of the dual-axis drive motor 28 (a single three-phase asynchronous motor with a power of 3kW) transmits power synchronously to the first horizontal lead screw 30 and the second horizontal lead screw 31 through a pair of bevel gear sets 29 with a 1:1 transmission ratio. The first horizontal lead screw 30 has a right-hand thread, and the second horizontal lead screw 31 has a left-hand thread, and both have the same lead (e.g., a lead of 8mm). When the dual-axis drive motor 28 rotates in the forward direction, the first moving seat 32 and the second moving seat 33 move towards each other at the same speed; when rotating in the reverse direction, they retract synchronously in opposite directions.

[0036] The first rotary polishing mechanism 34 and the second rotary polishing mechanism 35 are symmetrically arranged. The grinding heads 38 are cup-shaped diamond grinding wheels, each driven by its own rotary drive motor 36 (high-speed variable frequency motor, speed adjustable from 0-12000rpm). The control center controls the dual-axis drive motor 28 to drive the two moving seats to move towards each other at a feed speed of 100mm / min. The pressure sensor 39 (range 0-500N, accuracy 0.1%FS) mounted on the second moving seat 33 detects the contact pressure Pr between the right grinding head 38 and the right end face of the wheel hub steel ring in real time; the pressure sensor 39 mounted on the first moving seat 32 detects the contact pressure Pl on the left side in real time.

[0037] When both Pr and Pl reach the preset grinding force threshold (e.g., 80N), the control center determines that the grinding heads 38 on both sides have simultaneously contacted the workpiece end face. It then controls the dual-axis drive motor 28 to stop feeding and starts the rotary drive motors 36 on both sides, setting the grinding head 38 speed to 8000rpm. Simultaneously, the fourth servo motor drives the rotary table 40 to rotate the wheel hub steel ring at a constant speed of 30rpm. The grinding heads 38 on both sides complete the full circumference grinding of the entire end face during one revolution of the wheel hub steel ring.

[0038] During the grinding process, if the pressure sensor 39 detects a deviation in Pr or Pl due to grinding wheel wear or fluctuations in the workpiece end face shape (e.g., a deviation exceeding 5N), the control center immediately compensates for the feed position on the corresponding side by finely adjusting the rotation angle of the dual-axis drive motor 28 (incremental adjustment, with each step angle corresponding to a feed amount of 0.01mm), so that the grinding force returns to the target threshold. Since the feed mechanisms on both sides are synchronously driven by the same motor through the bevel gear set 29, the theoretical feed amounts on both sides remain strictly consistent, fundamentally ensuring the symmetry of the grinding amounts on both end faces. In this embodiment, the surface roughness Ra after end face grinding can reach 0.8μm, and the parallelism of the two end faces is better than 0.05mm.

[0039] After grinding, the dual-axis drive motor 28 reverses, and the two moving seats move out synchronously in opposite directions. The rotary drive motor 36 slows down and stops. The expansion-type centering fixture releases, and the flexible clamping and transfer module 10 transports the finished wheel hub steel ring to the unloading station 8.

[0040] S6 Implementation Details: Material Unloading and Data Archiving: The material unloading station 8 is equipped with a six-axis articulated robot. The robot's end is equipped with a vacuum suction cup or pneumatic gripper to take the finished wheel hub steel ring from the clamping mechanism of the flexible clamping and transfer module 10 and place it on different finished product trays according to good / defective products.

[0041] The control center writes all the data of this processing process, including the workpiece identification code, the arrival timestamps of each station, the welding process current / voltage / speed timing curves, the key frames of the molten pool image, the flattening roller gap setting value, the grinding force curve of the grinding process, and the final quality judgment result, into the database to form a digital production traceability file that uniquely corresponds to the wheel hub steel ring, which facilitates subsequent quality traceability and process optimization.

[0042] In this embodiment, the entire production line can achieve fully automated cyclic operation under the unified scheduling of the control center. Operators can monitor the status of each workstation in real time, modify process formula parameters, or perform manual single-step interventions through the human-machine interface (HMI). The system also supports integration with the factory's MES system to realize automatic production plan distribution and capacity data reporting.

[0043] Example 2 The main difference between this embodiment and Embodiment 1 is that the first linear drive element 15 and the second linear drive element 17 in the flexible clamping and transfer module 10 use electric cylinders instead of pneumatic cylinders to achieve higher clamping position control accuracy. In this embodiment, both the first linear drive element 15 and the second linear drive element 17 use servo electric cylinders (stroke 100mm, repeatability 0.02mm), and the precise displacement control of the front push plate 16 and the clamping plate 18 is achieved by driving the ball screw through a servo motor.

[0044] In step S2, the control center sends a position command to the servo electric cylinder driver. The front pusher plate 16 extends forward to the set position (e.g., 30mm from the initial position) at a speed of 50mm / s. Subsequently, the clamping plate 18 moves towards the wheel hub rim at a speed of 30mm / s until the clamping force reaches the set value (controlled by feedback from the pressure sensor built into the electric cylinder). The clamping force is set to 250N. The closed-loop position control of the servo electric cylinder enables the adjustment accuracy of the clamping gap to reach 0.05mm, which can accommodate wheel hub rims with rim thicknesses ranging from 5mm to 12mm.

[0045] The advantages of the servo electric cylinder solution compared to the pneumatic cylinder solution are: (1) the clamping force can be precisely controlled and is stable without fluctuations, avoiding changes in clamping force caused by air pressure fluctuations; (2) the clamping position can be precisely programmed to adapt to rapid changes in workpieces of different specifications; (3) no additional pneumatic system (air compressor, pipeline, solenoid valve, etc.) is required, reducing equipment maintenance costs. In this embodiment, the transfer cycle is about 10 seconds, which is slightly longer than the pneumatic cylinder solution, but the clamping accuracy and stability are significantly improved, making it suitable for high-precision machining scenarios.

[0046] Example 3 The main difference between this embodiment and Embodiment 1 is that the adaptive optimization unit 45 for process parameters at the welding station uses a deep learning model instead of a support vector regression model. In this embodiment, the parameter decision submodule 49 incorporates a fusion prediction model based on convolutional neural networks (CNN) and long short-term memory networks (LSTM).

[0047] Specifically, the image sequence of the molten pool area acquired by the visual perception unit 43 is input to the CNN branch. The CNN network contains three convolutional layers (with kernel sizes of 55, 33, and 33, and filter numbers of 32, 64, and 128, respectively) and two fully connected layers, used to extract high-dimensional visual features of the molten pool image. The temporal feature vector Ft=[Irms,Urms,Earc] uploaded by the data acquisition unit 42 is input to the LSTM branch. The LSTM network contains two LSTM layers, each with 64 hidden units, used to capture the temporal dynamic characteristics of the welding electrical signal. The outputs of the CNN branch and the LSTM branch are concatenated at the feature level and then input to the regression head composed of two fully connected layers, outputting the predicted weld depth Hpre and the predicted weld width Wpre, respectively.

[0048] The model training dataset contains 500 sets of welding experimental data under different welding process parameters (welding current 100A-250A, welding speed 400mm / min-800mm / min, workpiece thickness 2mm-5mm). Each set of data includes welding electrical signal timing data, molten pool image sequences, and corresponding metallographic measurement results of the weld cross-section (penetration depth, weld width). The model was trained for 200 epochs using the Adam optimizer, with an initial learning rate of 0.001, a learning rate decay coefficient of 0.95 / 10 epochs, and a batch size of 16.

[0049] In this embodiment, the root mean square error (RMSE) of the weld penetration prediction model is 0.08 mm, and the RMSE of the weld width prediction is 0.12 mm, representing improvements of 35% and 28% respectively compared to the SVR model in Embodiment 1. The parameter adjustment cycle remains 500 ms, but due to the higher model prediction accuracy, the convergence speed of parameter adjustment is faster, typically stabilizing the heat input within the target window within two adjustment cycles. This embodiment is suitable for high-end wheel rim production scenarios where extremely high welding quality consistency is required.

[0050] Example 4 The main difference between this embodiment and Embodiment 1 lies in the different drive layout schemes used in the dual-end synchronous grinding device of the grinding station. In this embodiment, the first horizontal lead screw 30 and the second horizontal lead screw 31 are driven by independent servo motors, and the precise synchronization of the feed motion on both sides is achieved through the synchronous control algorithm of the control center.

[0051] Specifically, the first horizontal lead screw 30 is driven by the fifth servo motor, and the second horizontal lead screw 31 is driven by the sixth servo motor. Both servo motors are equipped with absolute encoders (23-bit resolution). The control center synchronously sends position commands to the two servo drives via industrial Ethernet at a refresh rate of 1kHz. The timing coordination scheduling unit 44 has a built-in synchronization control algorithm and adopts a master-slave synchronization strategy: the fifth servo motor is the master axis, and the sixth servo motor is the slave axis. The slave axis tracks the position commands of the master axis in real time and compensates for the position deviation on both sides through a cross-coupling controller.

[0052] The control law of the cross-coupled controller is as follows: Let the actual positions on both sides be x1 and x2, respectively. The position synchronization error = x1 - x2. The cross-coupled controller output compensation amount u_c = Kp + Kidt, where Kp is the proportional gain (Kp = 2.0 in this embodiment) and Ki is the integral gain (Ki = 0.5 in this embodiment). The compensation amount is superimposed on the position commands of the master and slave axes respectively, realizing dynamic synchronous compensation of the feed motion on both sides.

[0053] In this embodiment, the position synchronization error on both sides can be controlled within 0.01mm, which is more accurate than the mechanical synchronization scheme in Embodiment 1. Furthermore, the advantage of the independent drive scheme is that the feed speed of the grinding heads 38 on both sides can be adjusted independently. When the initial morphology of the two end faces differs significantly, the control center can achieve adaptive balancing of the grinding amount on both sides through differentiated speed control. This embodiment is suitable for workpieces with significant differences in end face morphology or process scenarios requiring differentiated grinding.

[0054] Example 5 The main difference between this embodiment and Embodiment 1 is that the loading station 4 uses a vision-guided robotic automatic loading system instead of the traditional vibratory feeder loading mechanism. In this embodiment, the loading station 4 is equipped with a six-axis articulated loading robot, with a vacuum suction end effector installed at the robot's end. A second industrial area array camera (resolution 40963072, equipped with a 25mm industrial lens) is mounted above the loading area for three-dimensional position recognition and attitude detection of the wheel hub steel rim blanks stacked in the loading tray.

[0055] In step S1, the loading pallet is transported manually or by an automated guided vehicle (AGV) to the positioning fixture in the loading area. A second industrial area scan camera performs a comprehensive scan of the wheel rim blanks within the pallet. The vision perception unit 43 performs instance segmentation processing on the acquired images, identifying the position, orientation, and stacking layer number of each wheel rim. Based on the identification results, the control center plans the gripping sequence and gripping posture of the loading robot, prioritizing the gripping of the top-layer workpiece to avoid interference.

[0056] During robot gripping, the control center sends the target coordinates (X, Y, Z, Rx, Ry, Rz) to the robot controller. The robot end effector moves along a straight path to a position 100mm above the target location, then descends to the workpiece surface at a speed of 50mm / s. After the vacuum-adsorption end effector contacts the workpiece, the vacuum generator activates, establishing a vacuum of -0.06MPa within 0.3 seconds, achieving an adsorption force of over 500N. The robot lifts the workpiece from the tray along a straight path, moves it above the gap of the limit plate 3 of the conveyor device 2, and descends at a speed of 30mm / s to place the workpiece. The vacuum is then released, completing one loading cycle.

[0057] In this embodiment, the advantages of the vision-guided robotic loading system compared to the traditional vibratory feeder are: (1) it can adapt to wheel rims of different diameters (14-20 inches) and thicknesses without the need to replace the vibratory feeder track; (2) it can handle multiple loading trays simultaneously, achieving continuous and uninterrupted material supply; (3) visual detection can identify the workpiece posture, ensuring that the workpiece is placed on the conveyor 2 in the correct horizontal posture, avoiding positioning deviations caused by incorrect posture. The loading cycle time of this embodiment is about 15 seconds / piece, and the visual recognition time is about 2 seconds, which can meet the needs of medium-batch production.

[0058] Example 6 The main difference between this embodiment and Embodiment 1 is that the communication architecture of the integrated production control center 41 adopts Time-Sensitive Networking (TSN) instead of traditional industrial Ethernet to meet the requirements of higher precision time-series coordinated control. In this embodiment, the execution units and sensing units of each workstation are connected to the integrated production control center 41 through industrial Ethernet interfaces supporting the TSN protocol.

[0059] The TSN network employs the IEEE 802.1Qbv Time-Aware Shaper (TAS) mechanism. The timing coordination scheduling unit 44 in the control center acts as the network's Grand Master, synchronizing time with each slave node via the IEEE 802.1AS Precision Time Protocol (gPTP), achieving a time synchronization accuracy of 50ns. The network cycle is set to 1ms, and each cycle is divided into the following time slots: 0-200s for the control center to issue motion commands to the execution units; 200-500s for the sensing units to upload data (encoder position, sensor signals, image data, etc.) to the control center; 500-800s for data processing and decision-making within the control center; and 800-1000s as a protection bandwidth for the transmission of non-time-sensitive data.

[0060] The timing coordination scheduling unit 44 sends synchronous motion commands to each execution unit during the command issuance time slot of each network cycle, ensuring that each execution unit receives and executes commands within the same network cycle. For example, during the process of the flexible clamping and transfer module 10 moving the wheel hub steel ring from the conveying device 2 to the welding positioning table 19, the control center sends coordinated motion commands to the first servo motor 12, the horizontal transfer motor, the first linear drive element 15, and the second linear drive element 17 in each cycle. The difference in action delay between each execution unit is controlled within 1ms, achieving true millisecond-level multi-axis coordination.

[0061] In this embodiment, the introduction of the TSN network improves the timing coordination accuracy between workstations from 5ms in traditional industrial Ethernet to within 1ms, and reduces the overall production cycle fluctuation from 8% to within 2%. Furthermore, the deterministic transmission characteristics of the TSN network ensure that the high-frequency data (10kHz sampling) from the real-time welding parameter detection module 26 and the image data from the visual perception unit 43 can be transmitted to the control center within strict time limits, avoiding data loss and latency jitter problems caused by network congestion in traditional Ethernet. This embodiment is suitable for high-speed production lines with extremely high timing coordination accuracy requirements.

[0062] Example 7 The main difference between this embodiment and Embodiment 1 is that the flattening station 6 uses a multi-point independently controllable hydraulic flattening system instead of a traditional roller flattening device. In this embodiment, the flattening station 6 is equipped with twelve sets of circumferentially distributed independent hydraulic pressure heads. Each set of pressure heads is driven by an independent servo hydraulic cylinder, and the pressing force and pressing position can be adjusted independently.

[0063] The flattening and positioning platform is equipped with a centering clamping mechanism for precise positioning of the wheel hub rim. The twelve sets of hydraulic pressure heads are divided into six inner rings and six outer rings, corresponding to the spoke and rim areas of the wheel hub rim, respectively. The control center, based on the current wheel hub rim model and flattening process formula, sends the target pressing position and target pressing force to each servo hydraulic cylinder. The target pressing force for the six inner ring pressure heads is set to 80kN, and the target pressing force for the six outer ring pressure heads is set to 120kN. The pressing speed of each pressure head is controlled at 5mm / s.

[0064] During the flattening process, the force sensors built into each servo hydraulic cylinder provide real-time feedback on the actual pressing force. The control center adjusts the opening of the hydraulic valves through a PID closed-loop control algorithm, so that the actual pressing force of each pressure head tracks the target value. When the actual pressing force of all twelve pressure heads reaches more than 95% of the target value and remains there for 2 seconds, the flattening is considered complete. The advantages of the multi-point independent controllable flattening system in this embodiment compared with the traditional roller flattening are: (1) it can apply differentiated flattening forces to the spoke and rim areas to adapt to the deformation requirements of different structural stiffness areas; (2) the independent control of the twelve pressure heads can achieve precise correction of local warping; (3) the flatness after flattening can reach 0.03mm, which is better than the 0.08mm of roller flattening. This embodiment is suitable for the production of high-end wheel hub steel rims with extremely high flatness requirements.

[0065] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

[0066] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A processing system for producing wheel hub steel rims, comprising a processing platform and, sequentially arranged along the processing flow, a loading station, a welding station, a flattening station, a grinding station, and a unloading station on the processing platform, characterized in that, It also includes an integrated production control center, which communicates with the execution units and sensing units in the loading station, welding station, flattening station, grinding station, and unloading station via industrial fieldbus or industrial Ethernet, forming a closed-loop control network for the entire process. Specifically: the loading station includes an automatic loading mechanism, a conveying device, and a positioning detection component; the conveying device has multiple limit plates spaced along the conveying direction; the positioning detection component includes infrared position sensors installed at the entrance of each station to detect the position status of the wheel hub steel rim on the conveying device in real time and feed the position signal back to the integrated production control center; a flexible clamp is provided between the welding station and the grinding station. The flexible clamping and transfer module includes a gantry-type transfer rail mounted above the processing platform, a lifting drive mechanism installed on the transfer rail, and a clamping execution mechanism suspended by the lifting drive mechanism. The lifting drive mechanism includes a first servo motor and a first ball screw pair driven by the first servo motor. The clamping execution mechanism includes a lifting frame fixedly connected to the nut seat of the first ball screw pair, a first linear drive element and a second linear drive element horizontally mounted on the lifting frame, a front push plate driven by the first linear drive element, and a clamping plate driven by the second linear drive element. The front push plate and the clamping plate are arranged opposite to each other, forming a space between them and the wheel hub steel rim. A clamping gap adapted to the thickness is used to achieve flexible clamping and precise handling of the wheel hub steel ring; the welding station includes a welding positioning table, an arc-shaped centering clamping mechanism, and an automatic welding device; the arc-shaped centering clamping mechanism includes multiple arc-shaped clamping blocks arranged circumferentially around the center of the welding positioning table and a third linear drive element that drives the arc-shaped clamping blocks to retract radially and synchronously; the automatic welding device includes a second servo motor, a second ball screw pair driven by the second servo motor, and a movable welding head mounted on a sliding seat of the second ball screw pair; the welding head is equipped with a welding power supply and welding parameter real-time detection module, which is used to collect current, voltage, and arc sound during the welding process. The grinding station includes a positioning table, a dual-end-face synchronous grinding device, and a grinding force feedback component. The dual-end-face synchronous grinding device includes a dual-axis drive motor, a first horizontal lead screw and a second horizontal lead screw synchronously driven by the dual-axis drive motor through a bevel gear set, a first moving seat and a second moving seat respectively threaded with the first horizontal lead screw and the second horizontal lead screw, and a first rotary polishing mechanism and a second rotary polishing mechanism respectively mounted on the first moving seat and the second moving seat. The threads of the first horizontal lead screw and the second horizontal lead screw have opposite directions, so that the first moving seat and the second moving seat move synchronously towards each other or synchronously away from each other in the horizontal direction under the drive of the dual-axis drive motor.Both the first and second rotary polishing mechanisms include a rotary drive motor, a polishing spindle driven by the rotary drive motor, and a grinding head mounted on the end of the polishing spindle. The grinding force feedback component includes a pressure sensor mounted on the first and / or second moving base for real-time detection of the contact pressure between the grinding head and the end face of the wheel hub steel rim. The integrated production control center includes a data acquisition unit, a vision perception unit, a time-series collaborative scheduling unit, and a process parameter adaptive optimization unit. The data acquisition unit receives and preprocesses signals from each sensing unit. The vision perception unit includes an industrial area array camera mounted above the welding station and the grinding station. The time-series collaborative scheduling unit issues action commands to each execution unit based on a preset station cycle time table. The process parameter adaptive optimization unit performs fusion analysis on multi-dimensional parameters of the welding process and generates welding parameter fine-tuning commands.

2. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, The adaptive optimization unit for process parameters includes: a feature extraction submodule, used to extract the effective values ​​of welding current, effective values ​​of voltage, and short-time energy features of arc sound signals from the multi-dimensional parameter sequence of the welding process, and construct a time-domain feature vector; simultaneously, it extracts the geometric contour features and gray-scale distribution features of the molten pool from the image sequence of the molten pool area, and constructs an image feature vector; a state fusion submodule, used to perform feature-level fusion by aligning the time-domain feature vector and the image feature vector according to timestamps, and generate a fused feature representing the current welding thermal state; and a parameter decision submodule, which has a built-in quality prediction model trained based on historical welding process data, used to predict the current welding quality level according to the fused feature, and output speed adjustment commands for the second servo motor and / or current adjustment commands for the welding power supply according to preset mapping rules.

3. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, In the clamping execution mechanism of the flexible clamping and transfer module, the clamping gap width between the front push plate and the clamping plate is adjusted by the coordinated action of the first linear drive element and the second linear drive element to adapt to the clamping requirements of wheel hub steel rings of different thicknesses; the gantry transfer track is equipped with a horizontal transfer motor, which drives the lifting drive mechanism and the clamping execution mechanism to move in the horizontal direction, so as to realize the precise handling of wheel hub steel rings between various workstations.

4. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, The inner curvature of the multiple arc-shaped clamping blocks in the arc-shaped centering clamping mechanism is adapted to the outer circular contour of the wheel hub steel ring to be processed. When the third linear drive element drives each arc-shaped clamping block to synchronously retract radially, it clamps the workpiece from the outer circumference of the wheel hub steel ring to the center, realizing automatic centering and clamping of the wheel hub steel ring with a centering accuracy of 0.1mm.

5. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, In the dual-end-face synchronous grinding device, the lead of the first horizontal lead screw and the lead screw of the second horizontal lead screw are the same and the screw threads are opposite. Under the synchronous drive of the dual-axis drive motor through the bevel gear set, the first rotary polishing mechanism and the second rotary polishing mechanism feed towards each other or withdraw in opposite directions at strictly equal speeds and strokes, ensuring the symmetry of the grinding amount on both ends of the wheel hub steel ring.

6. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, The integrated production control center's timing coordination scheduling unit has a built-in programmable logic controller and industrial computer. Based on a preset workstation cycle timetable, combined with the trigger signals from the infrared position sensors and the image recognition results from the visual perception unit, it issues action commands to the flexible clamping and transfer module, the automatic welding device, and the dual-end synchronous grinding device, thereby achieving seamless connection between material flow and processing actions between each workstation.

7. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, The visual perception unit performs edge detection and Hough circle transformation on the images acquired by the industrial area array camera, identifies the outer circle contour and center pixel coordinates of the wheel hub steel ring, and combines the pre-calibrated camera intrinsic parameters and hand-eye calibration matrix to convert the pixel coordinates into physical coordinates in the processing platform coordinate system, establishes the workpiece coordinate system, and provides a gripping reference for the flexible clamping and transfer module.

8. The processing system for producing wheel hub steel rims according to claim 1, characterized in that, The bottom of the positioning table of the grinding station is equipped with a rotary worktable driven by a fourth servo motor. The center of the rotary worktable is equipped with an expansion-type centering clamp for centering and clamping from the inner hole of the wheel hub steel ring. During the double-end face grinding process, the rotary worktable drives the wheel hub steel ring to rotate at a constant speed around its axis, and works with the grinding heads on both sides to achieve synchronous grinding of the entire end face.

9. A method for machining wheel hub steel rims based on the machining system according to any one of claims 1 to 8, characterized in that, Includes the following steps: S1. Intelligent Feeding and Initial Positioning: The automatic feeding mechanism places the wheel hub steel rim blank to be processed onto the conveying device. The wheel hub steel rim is conveyed forward under the constraint of the limiting plate. When the infrared position sensor detects that the wheel hub steel rim has reached the station entrance, it generates an arrival signal. After receiving the arrival signal, the integrated production control center controls the conveying device to stop and triggers the vision perception unit to acquire images of the wheel hub steel rim, identify its model, center coordinates and azimuth angle, and establish the workpiece coordinate system; S2. Flexible Clamping and Precise Transfer: The integrated production control center issues a transfer command to the flexible clamping and transfer module according to the workpiece coordinate system information. The first servo motor of the flexible clamping and transfer module drives the first ball screw pair to rotate, driving the lifting frame and clamping execution mechanism to descend to the preset height. The first linear drive element pushes the front push plate to move horizontally forward, and the second linear drive... The moving element drives the clamping plate to move closer to the front push plate, completing the flexible clamping of the wheel hub steel ring. Then, the wheel hub steel ring is transported to the target station and placed precisely. S3, Adaptive welding process: After the wheel hub steel ring is centered and clamped on the welding positioning table by the arc surface centering clamping mechanism, the integrated production control center starts the automatic welding device. The second servo motor drives the second ball screw pair to drive the welding head to move along the edge of the wheel hub steel ring to be welded at a uniform speed for welding. During the welding process, the welding parameter real-time detection module collects welding current and voltage signals in real time, the visual perception unit collects the image sequence of the molten pool area in real time, and the process parameter adaptive optimization unit performs fusion analysis on the signals and image sequences. When the heat input is detected to deviate from the preset window, a fine adjustment command is generated. The welding heat input is returned to the target range by adjusting the speed of the second servo motor and / or the welding current. S4. Flattening and Shaping: The conveying device transports the welded wheel hub steel ring to the flattening station. The flattening device automatically adjusts the roller gap between the upper and lower pressure rollers according to the thickness setting value issued by the integrated production control center, applying the set flattening force to the wheel hub steel ring to complete the flatness shaping of the wheel body. S5. Double-End Face Synchronous Precision Grinding: After the wheel hub steel ring arrives at the grinding station, the flexible clamping and transfer module transports the wheel hub steel ring to the grinding positioning table and centers and clamps it. The integrated production control center starts the double-end face synchronous grinding device. The dual-axis drive motor synchronously drives the first and second horizontal lead screws to rotate through the bevel gear set, so that the first and second moving seats drive their respective rotating polishing mechanisms to move synchronously towards each other until the grinding heads on both sides are respectively against the two end faces of the wheel hub steel ring. Contact; the pressure sensor detects the contact pressure in real time. When the contact pressure on both sides reaches the preset grinding force threshold, the rotary drive motor starts and drives the grinding head to rotate at high speed. At the same time, the rotary worktable drives the wheel hub steel ring to rotate at a uniform speed, realizing full-circumference synchronous grinding of both ends. During the grinding process, the integrated production control center dynamically compensates for grinding force fluctuations based on the real-time feedback from the pressure sensor. S6, Unloading and Data Archiving: After grinding, the two sets of rotary polishing mechanisms of the double-end synchronous grinding device move out in opposite directions synchronously. The flexible clamping and transfer module transports the finished wheel hub steel ring to the unloading station. The integrated production control center associates and stores all process parameters, quality inspection data and equipment operation status logs collected during this processing, forming a complete production traceability file.

10. The wheel hub steel rim processing method according to claim 9, characterized in that, The adaptive optimization process of process parameters in step S3 specifically includes: a feature extraction submodule extracts the effective values ​​of welding current, effective values ​​of voltage, and short-time energy features of arc sound signals from the multi-dimensional parameter sequence of the welding process to construct a time-domain feature vector; simultaneously, it extracts the geometric contour features and gray-scale distribution features of the molten pool from the image sequence of the molten pool area to construct an image feature vector; a state fusion submodule aligns the time-domain feature vector and the image feature vector by timestamp and performs feature-level fusion to generate a fused feature; a parameter decision submodule predicts the current welding quality level based on the fused feature and outputs speed adjustment commands for the second servo motor and / or current adjustment commands for the welding power supply according to preset mapping rules; when the predicted penetration depth or weld width is lower than the target lower limit, it outputs a current increase command and a welding torch deceleration command; when the predicted penetration depth or weld width is higher than the target upper limit, it outputs a current decrease command and a welding torch acceleration command, with an adjustment cycle of 500ms.

Images