Jumper laser welding device and method

Abstract

The invention provides a jumper wire laser welding device and method, and the device comprises a linear transmission mechanism which is used for achieving the sequential circulation of all stations; the cable clamp directional conveying mechanism is used for automatically and directionally conveying the cable clamp to a clamping station; the rotary clamping mechanism is configured on the linear transmission mechanism, forms a space matching relation with the cable clamp directional conveying mechanism, and is used for completing spinning assembly of the cable clamp and the jumper wire to form an assembly body; the press-fit mechanism comprises a press-fit assembly and a pressing force detector electrically connected with the press-fit assembly, and is used for performing force-controlled press-fit on the assembly body; the insulating layer processing mechanism is arranged at a downstream station of the pressing mechanism and is used for stripping an insulating layer of the crimping part of the assembly body; the intelligent welding mechanism is arranged at a downstream station of the insulating layer processing mechanism and used for generating a welding path through image recognition and executing laser welding; unmanned operation of the whole process from assembling to welding is achieved, the reject ratio is controlled while the comprehensive efficiency is improved, and the production cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of jumper laser welding technology, and specifically to a jumper laser welding device and method. Background Technology

[0002] In the field of power connector manufacturing, the welding quality of jumpers and cable clamps directly affects the long-term reliability of electrical equipment. Currently, the industry faces three major technological bottlenecks: The process is fragmented: Traditional processes use a manual step-by-step operation mode, including discrete processes such as cable clamp placement, manual crimping, insulation peeling and welding. This is not only inefficient (the processing time for a single piece exceeds 3 minutes), but also results in poor product consistency due to manual intervention (the defect rate is as high as 8%). Insufficient precision: Conventional crimping mechanisms lack force control detection, which can easily lead to crimping that is too tight (damaging the cable) or too loose (poor contact). Lack of intelligent features: Welding paths rely on manual presets and cannot adapt to assembly position deviations, and welding quality is significantly affected by the operator's experience.

[0003] Although existing technologies have proposed automated welding solutions, they have the following drawbacks: 1) They adopt a split equipment layout, requiring manual transfer between processes; 2) The insulation treatment is separated from the welding station, increasing production costs and the risk of errors; 3) The vision system is only used for coarse positioning, and the welding path still needs to be preset.

[0004] The above background information is disclosed only to assist in understanding the inventive concept and technical solution of this invention. It does not necessarily belong to the prior art of this patent application, nor does it necessarily provide technical teachings. In the absence of clear evidence, the novelty and inventiveness of the above application shall be deemed to be incomplete. Summary of the Invention

[0005] To address the technical problems of traditional automated welding solutions, such as dispersed processes and low efficiency, insufficient precision in pressure control, poor adaptability of welding paths, and reliance on manual handling for insulation layer processing, this invention proposes a jumper laser welding device and method. By constructing a closed-loop production line through a linear transmission mechanism, it integrates force-controlled pressure bonding and adaptive welding, realizing unmanned operation throughout the entire process from assembly to welding. This not only improves overall efficiency but also controls the defect rate and significantly reduces production costs.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: On one hand, the present invention provides a jumper laser welding apparatus, comprising: Linear transmission mechanism, used to realize sequential flow of each workstation; A cable clamp directional conveying mechanism is used to automatically and directionally convey cable clamps to the clamping station; A rotary clamping mechanism is configured on the linear transmission mechanism and forms a spatial cooperation relationship with the cable clamp directional conveying mechanism. It is used to complete the spin-forming assembly of the cable clamp and the jumper to form an assembly. A pressing mechanism, comprising a pressing assembly and a pressing force detector electrically connected thereto, is used for force-controlled pressing of an assembly; An insulation layer processing mechanism is located downstream of the pressing mechanism and is used to peel off the insulation layer of the crimped part of the assembly. An intelligent welding mechanism, located downstream of the insulation layer processing mechanism, is used to generate a welding path through image recognition and perform laser welding.

[0007] This invention proposes a jumper laser welding device and method, which realizes unmanned operation of the entire process from assembly to welding, improves overall efficiency, controls the defect rate, and significantly reduces production costs.

[0008] As a preferred technical solution, the intelligent welding mechanism includes: A visual positioning unit is used to acquire feature images of the pressing end face; The deep learning processing unit uses deep learning to extract contours. The path planning unit automatically generates laser scanning trajectories based on contour features; The program generation module generates a processing program based on laser scanning trajectory data and welding parameters; The galvanometer control unit converts the processing program into laser control commands to complete the laser welding.

[0009] As a preferred technical solution, the deep learning processing unit includes: The image acquisition module is used to acquire the original image of the crimped area after the insulation layer has been removed; The annotation processing module is configured to annotate the indented edges in the original image and generate an annotated training dataset. The model training module adopts the FCN algorithm architecture, which trains the labeled dataset by configuring a deep learning environment to generate a deep learning model; The model conversion module converts deep learning models into ONNX universal format files; The edge extraction module loads the ONNX model to perform semantic segmentation on the new input image and outputs the outer contour data of the pressing area. The FCN algorithm adopts an end-to-end fully convolutional network structure; The training dataset contains at least 300 labeled samples.

[0010] As a preferred technical solution, the linear transmission mechanism includes: The support frame consists of a base, vertical plates, and layered fixed and / or movable load-bearing plates; A clamping array, consisting of wire clamps evenly spaced on each support plate, wherein the wire clamps hold jumper wires. The driving component includes: a group of parallel guide rails, a stepper motor disposed between the guide rails, and a slider connected to the movable support plate and the guide rails. The stepper motor is used to realize the dynamic alignment of the wire clamp with each preset station.

[0011] As a preferred technical solution, the cable clamp directional conveying mechanism includes: Vibratory feeder is used for orientation correction and serialized output of cable clamps; The directional conveyor has its input end connected to the output end of the vibratory feeder, and is used to convey the sorted cable clamps to the clamping station.

[0012] As a preferred technical solution, the rotary clamping mechanism includes: support A linear guide component, which is mounted on and connected to the bracket; A rotary drive component is slidably mounted on the linear guide component; A rotary clamping actuator is connected to the rotary drive, wherein the linear guide guides the rotary drive to move the rotary clamping actuator between the clamping station and the assembly station, and the rotary drive drives the rotary clamping actuator to perform clamping and rotary assembly actions.

[0013] As a preferred technical solution, the pressing assembly includes: Pressing drive component, A pressing actuator that is connected to the pressing drive; The controller, wherein the clamping force detector is electrically connected to the controller, and the controller is electrically connected to the clamping drive, is used to control the working state of the clamping drive based on the detected clamping force.

[0014] As a preferred technical solution, the insulation layer processing mechanism includes: Insulation layer treatment drive component, A cutting machine is connected to the insulation layer treatment drive unit. The cutting machine has circumferentially arranged cutting blades. The insulation layer treatment drive unit drives the cutting machine to rotate so that the cutting blades can cut and peel off the insulation layer of the assembly crimping part.

[0015] On the other hand, the present invention also provides a jumper laser welding method, which is produced using the jumper laser welding apparatus as described in any of the preceding claims, and includes the following steps: S1 delivers the jumper wire to the cable clamp installation station via a linear transmission mechanism; S2 achieves the coiling of cable clamps and jumpers by forming a spatial coordination relationship between the cable clamp directional conveying mechanism, the rotary clamping mechanism, and the linear transmission mechanism. S3 performs pressure-controlled pressing operations on the assembly at the pressing station; S4 is the process of peeling off the insulation layer of the assembly after pressing at the insulation layer peeling station. At the welding station, the S5 generates a welding path for the assembly with the insulation layer stripped through image recognition and performs laser welding.

[0016] As a preferred technical solution, step S5 specifically includes the following steps: Images of the compression cross-section are acquired using a visual positioning unit; The outer contour data was obtained by segmenting the model using the FCN algorithm. Automatically generate laser galvanometer control commands based on contour features; Complete the laser welding operation; The establishment of the FCN algorithm model includes: Collect and label no fewer than 300 sets of sample images after insulation layer removal; Configure a deep learning training environment for model training; Output a generic model file in ONNX format; Deploy the model into the welding control system.

[0017] The jumper laser welding device and method provided by this invention have the following beneficial effects: 1) The present invention provides a jumper laser welding device and method, which realizes unmanned operation of the entire process from assembly to welding, improves the overall efficiency, controls the defect rate, and significantly reduces the production cost.

[0018] 2) The patch cord laser welding device and method provided by the present invention rely on segmented manual operation (such as assembly, pressing and welding) in traditional processes, which leads to process interruption and poor consistency. The present invention connects each station in series through a linear transmission mechanism and combines it with a rotary clamping mechanism to realize the automatic assembly of cable clamps and patch cords, ensuring continuous flow throughout the process and significantly improving production efficiency and stability. Conventional pressing processes lack real-time monitoring, which can easily lead to defects such as excessive tightness (damaging cables) or excessive looseness (poor electrical contact). This solution uses a pressing mechanism with a pressing force detector to achieve force-controlled pressing of the assembly, ensuring the reliability of the pressing quality. Existing equipment relies on preset paths and cannot cope with assembly position deviations, resulting in low welding accuracy. This solution generates welding paths in real time through the image recognition function of an intelligent welding mechanism, combined with laser welding execution, to adapt to assembly requirements at various angles, thereby improving welding accuracy and yield. The insulation layer peeling process usually requires separate operation, which increases costs and the risk of errors. This solution integrates the insulation layer processing mechanism into the downstream station, automatically peeling off the insulation layer of the crimping part, eliminating process breaks and enhancing the degree of automation. In summary, this solution, through the synergistic action of a linear transmission mechanism, a force-controlled pressing mechanism, an insulation layer processing mechanism, and an intelligent welding mechanism, solves the bottleneck problems of insufficient automation, difficulty in quality control, and lack of process intelligence in jumper welding; it achieves unmanned operation throughout the entire process from assembly to welding, improving overall efficiency while controlling the defect rate and significantly reducing production costs. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of a jumper laser welding device provided by the present invention; Figure 2 A schematic diagram of the structure of a jumper laser welding device provided by the present invention from another perspective; Figure 3 This is a structural schematic diagram of a jumper laser welding device provided by the present invention from another perspective. Figure 4 This is a partial structural schematic diagram of a jumper laser welding device provided by the present invention; Figure 5 A flowchart of a jumper laser welding method provided by the present invention; Among them, 1-linear transmission mechanism; 2-cable clamp directional conveying mechanism; 3-rotary clamping mechanism; 4-pressing mechanism; 5-insulation layer treatment mechanism; 6-intelligent welding mechanism; 7-visual positioning unit; 8-base; 9-vertical plate; 10-bearing plate; 11-cable clamp; 12-jump wire; 13-guide rail; 14-stepper motor; 15-slider; 16-vibratory feeder; 17-directional transmission component; 18-input end of directional transmission component; 19-output end of vibratory feeder; 20-bracket; 21-linear guide component; 22-rotary drive component; 23-connector; 24-rotary clamping actuator; 25-pressing drive component; 26-pressing actuator; 27-insulation layer treatment drive component; 28-cutting machine; 29-cable clamp; 30-vibratory feeder. Detailed Implementation

[0020] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0021] like Figures 1-4 As shown, the present invention provides a jumper laser welding device, comprising: Linear transmission mechanism 1 is used to realize the sequential flow of each workstation; The cable clamp directional conveying mechanism 2 is used to automatically directionally convey the cable clamp 29 to the clamping station; The rotary clamping mechanism 3 is disposed on the linear transmission mechanism 1 and forms a spatial cooperation relationship with the cable clamp directional conveying mechanism 2. It is used to complete the spin-forming assembly of the cable clamp 29 and the jumper 12 to form an assembly. The pressing mechanism 4 includes a pressing component and a pressing force detector electrically connected thereto, for force-controlled pressing of the assembly; The insulation layer processing mechanism 5 is located downstream of the pressing mechanism 4 and is used to peel off the insulation layer of the crimped part of the assembly. The intelligent welding mechanism 6 is located at the downstream station of the insulation layer processing mechanism 5 and is used to generate a welding path through image recognition and perform laser welding.

[0022] This invention proposes a jumper laser welding device and method, which realizes unmanned operation of the entire process from assembly to welding, improves overall efficiency, controls the defect rate, and significantly reduces production costs.

[0023] Preferably, such as Figures 1-3 As shown, the intelligent welding mechanism 6 includes: Visual positioning unit 7 is used to acquire feature images of the pressing end face; The deep learning processing unit uses deep learning to extract contours. The path planning unit automatically generates laser scanning trajectories based on contour features; The program generation module generates a processing program based on laser scanning trajectory data and welding parameters; The galvanometer control unit converts the processing program into laser control commands to complete laser welding. The visual positioning unit captures the geometric features of the assembly's press-fit end face (such as cable clamp contours and weld point positions) in real time through high-definition imaging, providing raw image data input for contour extraction; The deep learning processing unit automatically identifies and segments key welding areas in the image (such as cable clamp edges and insulation peeling areas), eliminates deviations from manually preset paths, and adapts to contour features under different assembly angles. The path planning unit automatically calculates the scanning trajectory of the laser beam based on the extracted contour features: A three-dimensional welding path is planned based on the curved shape of the cable clamp to generate a dynamic trajectory, ensuring full coverage of the weld. The program generation module integrates welding parameters (power, speed, spot diameter) and geometric path data to generate executable machining program code for the galvanometer system, achieving the matching of welding process and motion trajectory; The galvanometer control unit converts the processing program into electrical signals that drive the galvanometer motor, controlling the lens deflection angle; relying on millisecond-level dynamic scanning capability (X / Y axis galvanometer coordination), it enables the laser focus to move precisely along the planned path; in conjunction with the dynamic focusing system, it maintains a constant spot diameter in the welding area, improving weld consistency; This mechanism overcomes three major limitations of traditional welding—adaptability defects, precision bottlenecks, and efficiency bottlenecks—through the synergistic effect of a visual positioning unit, a deep learning processing unit, a path planning unit, a program generation module unit, and a galvanometer control unit. Adaptive defects: Deep learning processing units automatically compensate for assembly position deviations, replacing the rigid constraints of preset paths; Precision bottleneck: The galvanometer control unit achieves micro-motion at the 0.5-10μm level to ensure the quality of welds on complex curved surfaces; Efficiency bottleneck: The millisecond-level scanning speed of the galvanometer (up to 7m / s) significantly shortens the welding cycle.

[0024] Preferably, the deep learning processing unit includes: The image acquisition module is used to acquire the original image of the crimped area after the insulation layer has been removed; The annotation processing module is configured to annotate the indented edges in the original image and generate an annotated training dataset. The model training module adopts the FCN algorithm architecture, which trains the labeled dataset by configuring a deep learning environment to generate a deep learning model; The model conversion module converts deep learning models into ONNX universal format files; The edge extraction module loads the ONNX model to perform semantic segmentation on the new input image and outputs the outer contour data of the pressing area. The FCN algorithm adopts an end-to-end fully convolutional network structure; The training dataset contains at least 300 labeled samples; The image acquisition module uses a high-resolution camera to acquire original images of the crimped area, ensuring the clarity and detail integrity of the input data and providing high-quality visual input for subsequent processing. The annotation processing module performs pixel-level annotations (such as semantic segmentation annotations) on the indented edges in the original image to generate a structured training dataset; The labeled data covers multiple samples (≥300 images) to enhance the model's robustness to assembly deviations and lighting changes; The model training module uses the FCN algorithm's end-to-end fully convolutional structure to support inputs of arbitrary sizes. It restores spatial resolution through deconvolutional layers to achieve pixel-level contour prediction. The deep learning environment is configured with automatic gradient adjustment to enhance feature extraction capabilities. The model conversion module converts the trained model into ONNX universal format files, enabling cross-platform deployment (such as embedded devices or industrial PCs) and ensuring algorithm compatibility and execution efficiency. The edge extraction module loads the ONNX model to perform real-time semantic segmentation of new images and outputs the outer contour data of the crimped area with an error controlled within ±0.1mm. Through multi-level feature fusion of FCN, it accurately identifies the crimped edges under different angles and lighting conditions.

[0025] The collaborative efforts of the image acquisition module, annotation processing module, model training module, model conversion module, and edge extraction module have solved the two major bottlenecks of manual reliance and accuracy limitations in traditional welding: Manual dependency: Automatic labeling and training replace manually preset rules, adapting to complex industrial scenarios; Accuracy limitations: The end-to-end structure of the FCN algorithm enables sub-pixel level contour extraction, providing reliable input for laser welding path planning.

[0026] Preferably, such as Figure 1-4 As shown, the linear transmission mechanism 1 includes: The support frame consists of a base 8, a vertical plate 9, and layered fixed and / or movable load-bearing plates 10. A clamping array, with wire clamps 11 evenly spaced on each carrier plate 10, wherein the wire clamps 11 clamp jumpers 12. The driving component includes: a group of parallel guide rails 13, a stepper motor 14 disposed between the guide rails 13, and a slider 15 connected to the movable support plate 10 and the guide rails 13. The stepper motor 14 is used to realize the dynamic alignment of the wire clamp 11 with each preset station. Stepper motor 14 drives slider 15 to move linearly along guide rail 13. In conjunction with slider 15, movable support plate 10 is moved to ensure that wire clamp 11 is dynamically aligned with preset workstations (such as assembly, pressing, insulation layer treatment, welding), replacing manual intervention and realizing unmanned operation of the entire process.

[0027] Preferably, such as Figures 1-4 As shown, the cable clamp directional conveying mechanism 2 includes: Vibratory feeder 16 is used to correct the orientation of cable clamp 29 and serialize its output. The directional transmission component 17 has its input end 18 connected to the output end 19 of the vibratory feeder, and is used to transport the sorted cable clamps 29 to the clamping station. The vibratory feeder 16 includes an eccentric rotator and a vibratory feeder 30. The eccentric rotator drives the vibratory feeder 30 to vibrate horizontally, so that the randomly input cable clamps are automatically adjusted to a preset orientation (such as terminals facing forward / interfaces facing upward) during screening. Based on amplitude and frequency control, the corrected cable clamps are continuously output at fixed intervals to provide a stable feed flow for directional transmission. The directional transmission component 17 is preferably a belt conveyor. The input end of the directional transmission component 17 is seamlessly connected to the output end of the vibratory feeder 16. The serialized cable clamps 29 are carried by the belt conveyor to avoid secondary positional deviation.

[0028] The vibratory feeder 16 and the directional conveyor 17 work together to replace manual sorting and improve production efficiency.

[0029] Preferably, such as Figures 1-4 As shown, the rotating clamping mechanism 3 includes: 20 brackets Linear guide 21 is mounted on bracket 20 and connected to bracket 20; The rotary drive component 22 is slidably disposed on the linear guide component 21; A rotary clamping actuator 24 is connected to the rotary drive 22, wherein the linear guide 21 guides the rotary drive 22 to move the rotary clamping actuator 24 between the clamping station and the assembly station, and the rotary drive 22 drives the rotary clamping actuator 24 to perform clamping and rotary assembly actions. The vibratory feeder 16 of the cable clamp directional conveying mechanism 2 sorts the messy cable clamps 29 into a specific orientation and conveys them to the clamping station via the directional conveyor 17; the rotary drive 22 of the rotary clamping mechanism 3 slides along the linear guide 21 to the clamping station, and after the rotary drive 22 drives the rotary clamping actuator 24 to complete the clamping of the cable clamp 29, the rotary drive 22 slides along the linear guide 21 to the assembly station, and the rotary drive 22 drives the rotary clamping actuator 24 to rotate and screw the cable clamp 29 to the end of the jumper 12 to form an assembly; The rotary drive component 22 (such as a servo motor) drives the rotary clamp actuator 24 to rotate, thereby achieving the clamping action (opening and closing) and rotary assembly action of the cable clamp 29; the linear guide component 21 is preferably a Y-axis guide component, and the linear sliding of the linear guide component 21 cooperates with the rotary drive component 22 to achieve continuous operation of the entire process of "grabbing-translation-rotation-assembly", which improves production efficiency.

[0030] Preferably, such as Figures 1-3 As shown, the pressing assembly includes: Pressing drive component 25, Pressing actuator 26 is connected to the pressing drive 25 in a transmission manner; The controller, wherein the clamping force detector is electrically connected to the controller, and the controller is electrically connected to the pressing drive 25, is used to control the working state of the pressing drive 25 according to the detected clamping force; The pressing drive 25 drives the pressing actuator 26 to move downward from the initial position to press the assembly. During the pressing process, the pressing force detector detects the pressing pressure in real time. When the detected pressing pressure reaches the preset pressure threshold, the controller controls the pressing drive 25 to lift the pressing actuator 26 back to the initial position.

[0031] Preferably, such as Figures 1-3 As shown, the insulation layer processing mechanism 5 includes: Insulation layer treatment drive component 27, A cutting machine 28 is connected to the insulation layer processing drive 27. The cutting machine 28 has circumferentially arranged cutting blades (not shown). The insulation layer processing drive 27 drives the cutting machine 28 to rotate so that the cutting blades can cut and peel off the insulation layer of the assembly crimping part. The insulation layer processing drive 27 (such as a servo motor) drives the cutting machine 28 to rotate, so that the circumferentially arranged cutting blades (not shown) can perform circumferential cutting on the insulation layer of the cable crimping part, with a peeling depth accurate to ±0.02mm, so as to avoid damage to the internal conductor.

[0032] like Figures 1-4As shown, this invention provides a jumper laser welding device. A linear transmission mechanism 1 drives a carrier plate 10 to precisely displace along a guide rail via a stepper motor 14, causing the jumper 12 carried by the clamp 11 to flow to the assembly station according to a preset rhythm. A vibratory feeder 16 of the cable clamp directional conveying mechanism 2 sorts the disordered cable clamps 29 according to a specific orientation and conveys them to the clamping station via a directional transmission component 17. A rotary clamping mechanism 3's rotary drive component 22 slides along the linear guide 21 to the clamping station, and the rotary drive component 22 drives the rotary clamp to perform the operation. After component 24 completes the clamping of cable clamp 29, rotary drive component 22 slides along linear guide component 21 to the assembly station. Rotary drive component 22 drives rotary clamping actuator 24 to rotate and screw cable clamp 29 onto the end of jumper wire 12, forming an assembly. Linear transmission mechanism 1 drives bearing plate 10 to precisely move along guide rail 13 via stepper motor 14, so that the assembly carried by clamp 11 flows to the pressing station according to a preset rhythm. Pressing drive component 25 drives pressing actuator 26 to move downward from the initial position to press the assembly. The pressing process... In the process, the clamping force detector monitors the clamping pressure in real time. When the detected clamping pressure reaches the preset pressure threshold, the controller controls the clamping drive 25 to lift the clamping actuator 26 to the initial position, completing the clamping. The linear transmission mechanism 1 drives the carrier plate 10 to precisely move along the guide rail 13 via the stepper motor 14, so that the clamped assembly carried by the wire clamp 11 flows to the insulation treatment station according to the preset rhythm. The insulation layer treatment drive 27 drives the cutting machine 28 to rotate, driving the circumferentially arranged cutter (not shown) to perform circumferential cutting on the insulation layer of the jumper wire clamping part, cutting and peeling off the insulation layer of the clamping part of the assembly. The linear transmission mechanism 1 drives the carrier plate 10 to precisely move along the guide rail 13 via the stepper motor 14, so that the assembly carried by the wire clamp 11 after the insulation layer has been peeled off flows to the welding station according to the preset rhythm. The visual positioning unit 7 acquires the image of the clamping end face, extracts the contour features through the deep learning model implemented by the FCN algorithm, generates a spiral progressive laser scanning trajectory based on the contour data, and the galvanometer system executes the processing instructions to complete the laser welding.

[0033] On the other hand, such as Figure 5 As shown, the present invention also provides a jumper laser welding method, which is produced using the jumper laser welding apparatus as described in any of the preceding claims, and includes the following steps: S1 conveys the jumper 12 to the cable clamp 29 installation station via the linear transmission mechanism 1; S2 forms a spatial coordination relationship through the cable clamp directional conveying mechanism 2, the rotary clamping mechanism 3, and the linear transmission mechanism 1, so as to realize the spin forming of the cable clamp 29 and the jumper 12 into an assembly; S3 performs pressure-controlled pressing operations on the assembly at the pressing station; S4 is the process of peeling off the insulation layer of the assembly after pressing at the insulation layer peeling station. At the welding station, the S5 generates a welding path for the assembly with the insulation layer stripped through image recognition and performs laser welding.

[0034] This invention proposes a jumper laser welding method that enables unmanned operation of the entire process from assembly to welding, improving overall efficiency, controlling the defect rate, and significantly reducing production costs.

[0035] Preferably, step S5 specifically includes the following steps: Images of the pressed cross-section are acquired using the visual positioning unit 7; The outer contour data was obtained by segmenting the model using the FCN algorithm. Automatically generate laser galvanometer control commands based on contour features; Complete the laser welding operation; The establishment of the FCN algorithm model includes: Collect and label no fewer than 300 sets of sample images after insulation layer removal; Configure a deep learning training environment for model training; Output a generic model file in ONNX format; The model is deployed to the welding control system; compared with traditional processes, this application improves positioning accuracy, adaptability and yield.

[0036] It is understood that this invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of this invention. Furthermore, under the teachings of this invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of this invention. Therefore, this invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are protected by this invention. Furthermore, under the teachings of this invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of this invention. Therefore, this invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of this invention.

Claims

1. A jumper laser welding device, characterized in that, include: Linear transmission mechanism, used to realize sequential flow of each workstation; A cable clamp directional conveying mechanism is used to automatically and directionally convey cable clamps to the clamping station; A rotary clamping mechanism is configured on the linear transmission mechanism and forms a spatial cooperation relationship with the cable clamp directional conveying mechanism. It is used to complete the spin-forming assembly of the cable clamp and the jumper to form an assembly. A pressing mechanism, comprising a pressing assembly and a pressing force detector electrically connected thereto, is used for force-controlled pressing of an assembly; An insulation layer processing mechanism is located downstream of the pressing mechanism and is used to peel off the insulation layer of the crimped part of the assembly. An intelligent welding mechanism, located downstream of the insulation layer processing mechanism, is used to generate a welding path through image recognition and perform laser welding.

2. The jumper laser welding device according to claim 1, characterized in that, The intelligent welding mechanism includes: A visual positioning unit is used to acquire feature images of the pressing end face; The deep learning processing unit uses deep learning to extract contours. The path planning unit automatically generates laser scanning trajectories based on contour features; The program generation module generates a processing program based on laser scanning trajectory data and welding parameters; The galvanometer control unit converts the processing program into laser control commands to complete the laser welding.

3. The jumper laser welding apparatus according to claim 2, characterized in that, The deep learning processing unit includes: The image acquisition module is used to acquire the original image of the crimped area after the insulation layer has been removed; The annotation processing module is configured to annotate the indented edges in the original image and generate an annotated training dataset. The model training module adopts the FCN algorithm architecture, which trains the labeled dataset by configuring a deep learning environment to generate a deep learning model; The model conversion module converts deep learning models into ONNX universal format files; The edge extraction module loads the ONNX model to perform semantic segmentation on the new input image and outputs the outer contour data of the pressing area. The FCN algorithm adopts an end-to-end fully convolutional network structure; The training dataset contains at least 300 labeled samples.

4. The jumper laser welding device according to claim 1, characterized in that, The linear transmission mechanism includes: The support frame consists of a base, vertical plates, and layered fixed and / or movable load-bearing plates; A clamping array, consisting of wire clamps evenly spaced on each support plate, wherein the wire clamps hold jumper wires. The driving component includes: a group of parallel guide rails, a stepper motor disposed between the guide rails, and a slider connected to the movable support plate and the guide rails. The stepper motor is used to realize the dynamic alignment of the wire clamp with each preset station.

5. The jumper laser welding apparatus according to claim 1, characterized in that, The cable clamp directional conveying mechanism includes: Vibratory feeder is used for orientation correction and serialized output of cable clamps; The directional conveyor has its input end connected to the output end of the vibratory feeder, and is used to convey the sorted cable clamps to the clamping station.

6. The jumper laser welding apparatus according to claim 1 or 5, characterized in that, The rotary clamping mechanism includes: support A linear guide component, which is mounted on and connected to the bracket; A rotary drive component is slidably mounted on the linear guide component; A rotary clamping actuator is connected to the rotary drive, wherein the linear guide guides the rotary drive to move the rotary clamping actuator between the clamping station and the assembly station, and the rotary drive drives the rotary clamping actuator to perform clamping and rotary assembly actions.

7. The jumper laser welding apparatus according to claim 1, characterized in that, The pressing assembly includes: Pressing drive component, A pressing actuator that is connected to the pressing drive; The controller, wherein the clamping force detector is electrically connected to the controller, and the controller is electrically connected to the clamping drive, is used to control the working state of the clamping drive based on the detected clamping force.

8. The jumper laser welding apparatus according to claim 1, characterized in that, The insulating layer processing mechanism includes: Insulation layer treatment of drive components, A cutting machine is connected to the insulation layer treatment drive unit. The cutting machine has circumferentially arranged cutting blades. The insulation layer treatment drive unit drives the cutting machine to rotate so that the cutting blades can cut and peel off the insulation layer of the assembly crimping part.

9. A method for laser welding jumpers, characterized in that, The production process using the jumper laser welding apparatus according to any one of claims 1-7 includes the following steps: S1 delivers the jumper wire to the cable clamp installation station via a linear transmission mechanism; S2 achieves the coiling of cable clamps and jumpers by forming a spatial coordination relationship between the cable clamp directional conveying mechanism, the rotary clamping mechanism, and the linear transmission mechanism. S3 performs pressure-controlled pressing operations on the assembly at the pressing station; S4 is the process of peeling off the insulation layer of the assembly after pressing at the insulation layer peeling station. At the welding station, the S5 generates a welding path for the assembly with the insulation layer stripped through image recognition and performs laser welding.

10. The jumper laser welding method according to claim 9, characterized in that, Step S5 specifically includes the following steps: Images of the compression cross-section are acquired using a visual positioning unit; The outer contour data was obtained by segmenting the model using the FCN algorithm. Automatically generate laser galvanometer control commands based on contour features; Complete the laser welding operation; The establishment of the FCN algorithm model includes: Collect and label no fewer than 300 sets of sample images after insulation layer removal; Configure a deep learning training environment for model training; Output a generic model file in ONNX format; Deploy the model into the welding control system.

Images