A multi-model part quick change device and method for an engine part machining production line

Abstract

The present application relates to the technical field of program-controlled manipulator, and particularly relates to a multi-model part quick changing device and method for engine part machining production line, which integrates photoelectric detection, machine vision, digital twin, double-arm robot flexible operation, PLC logic control and other technologies, realizes detection part taking and placing interlocking detection and completes clamp and system initialization through a changing box with an unbalanced lever swing arm, confirms part ID in a double-verification mode of manual input and visual identification, completes part flexible clamping by relying on a master-slave arm interchangeable double-arm robot combined with force-position hybrid control, realizes robot operation precision checking and generates target coordinates through visual mistake proofing, defect detection and coordinate analysis, plans an optimal path by a digital twin model after PLC full signal checking, drives the robot to complete accurate assembly, and finally verifies the changing result by vision and stores full-process changing data and updates equipment maintenance prediction model.

Description

Technical Field

[0001] This invention relates to the field of program-controlled robotic arms, specifically to a device and method for rapid changeover of multiple models of parts in an engine parts processing production line. Background Technology

[0002] In the field of industrial robot processing, especially in robot workstations for processing engines and precision mechanical parts, it is often necessary to process multiple models of parts with similar structures but different clamping points and dimensional parameters on the same workstation. When switching processing, the inspection parts need to be replaced accordingly, and the existing switching methods still have many technical defects.

[0003] Traditional part changeover inspection relies heavily on manual visual confirmation of the pick-and-place status, which is prone to problems such as incorrect replacement, missing replacements, or misplacement, directly leading to substandard part processing accuracy or even part damage. Part model numbers are only confirmed by manual input without a secondary verification process, making it difficult to detect input errors in a timely manner, further increasing the error rate of changeover operations.

[0004] Meanwhile, robot clamping operations lack precise visual error prevention and accuracy verification mechanisms, resulting in prominent problems such as clamping position deviation and insufficient operation accuracy. Furthermore, existing technologies do not integrate fixture inspection, model matching, robot operation, and accuracy verification into a unified whole, leading to low changeover efficiency due to the independent operation of each module.

[0005] Furthermore, the data such as operation parameters, time consumption, and verification results throughout the entire changeover process have not been systematically stored and analyzed, which cannot provide data support for equipment maintenance and process optimization. Equipment failures are mostly repaired after the fact, making it difficult to meet the requirements of modern production lines for precise, efficient, and intelligent changeover operations for multiple parts.

[0006] To address the aforementioned issues, this invention provides a device and method for rapid changeover of multiple models of parts in an engine parts processing production line. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a device and method for rapid changeover of multiple models of engine parts in a production line. It achieves efficient and accurate changeover by focusing on precise fixture changeover detection, multi-dimensional error prevention and verification, robot collaborative operation, digital twin optimization, and full-process data management. It combines error prevention, accuracy, and intelligence, effectively avoiding errors in each stage of the changeover process.

[0008] To achieve the above objectives, one of the present inventions provides a rapid changeover device for multiple models of engine parts in a production line, comprising a fixture changeover and detection module, a PLC control module, a vision recognition and positioning module, a dual-arm robot operation module, a digital twin path planning module, and a data storage and maintenance module. The fixture change detection module includes a change box with an unbalanced lever arm, a photoelectric sensor, and a fixture coding sensor. The change box has a unique storage groove that matches the detected part. The photoelectric sensor transmits the part in place / out of place signal to the PLC control module. The PLC control module connects to each module and performs interlock logic judgment, ID matching and comparison, signal comprehensive verification and instruction sending. The visual recognition and positioning module completes coordinate system calibration, image acquisition and preprocessing, feature extraction and matching, defect detection and coordinate analysis; The dual-arm robot operation module has an interchangeable master and slave arm structure, and is equipped with a flexible gripper and a force-position hybrid control system to realize parts transfer, flexible clamping and collaborative assembly. The digital twin path planning module constructs a virtual mapping body for the workstation and plans the optimal motion path for the robot; the data storage and maintenance module stores the entire replacement process data, dynamically updates the equipment maintenance prediction model, and realizes preventive maintenance.

[0009] The second aspect of this invention provides a method for rapid changeover of multiple models of parts in an engine parts processing production line, which is used to implement a rapid changeover device for multiple models of parts in an engine parts processing production line, comprising: The part picking and placing status is detected by the changing box containing an unbalanced lever swing arm and photoelectric sensors. The sensor transmits the position / absence signal to the PLC for interlock logic judgment. After the judgment is passed, the fixture coding sensor collects the fixture code and stores it in the PLC. The PLC retrieves the corresponding cylinder parameters to drive the fixture to complete the action adjustment. At the same time, the vision module calibrates the workstation coordinate system and the digital twin model completes the initialization of the virtual mapping body. The operator inputs the part ID on the touch screen. The PLC temporarily stores the ID. The vision module acquires the part image and preprocesses it, extracts multi-dimensional features and matches them with the pre-stored standard model to generate a recognition ID. The PLC compares the manually input ID with the visually recognized ID. If the matching degree reaches the preset threshold, the final ID is confirmed and officially stored. If the standard is not met, the system alarms and re-verifies. The PLC sends instructions to the dual-arm robot with interchangeable master and slave arms. The slave arm holds the semi-finished product in the air with a flexible gripper, while the master arm moves the part to be processed to the corresponding gripper. The two arms work together to calibrate the relative position of the part and the gripper. The contact force is sensed through force-position hybrid control to avoid the part being deformed due to over-clamping, thus completing the precise and flexible clamping of the part to be processed. The vision module acquires images of the clamped parts and matches them with standard clamping images to generate error prevention signals. After the PLC compares the signals and finds they are consistent, it extracts the features of the part image to detect defects. If there are no defects, the coordinate analysis module establishes a multi-polar coordinate system to obtain the polar coordinates of the part and verifies the robot's operation accuracy. Once the accuracy is met, the average polar coordinates are converted into the robot's target coordinates. The PLC completes fixture changeover, ID matching, visual error prevention, and comprehensive verification of robot accuracy signals. After passing the verification, the digital twin model plans the optimal motion path for the robot. The PLC retrieves the corresponding work program and sends instructions and target coordinates to the robot. The dual-arm robot works collaboratively along the path to complete the precise assembly of parts and semi-finished products, realizing the core changeover operation. The vision module once again captures images of the parts, fixtures, and semi-finished products being assembled together. The PLC compares these images with the standard images to complete the final verification. If the images pass the verification, a processing-ready signal is sent. If the images fail the verification, an alarm is triggered and the fault is investigated and readjusted. At the same time, the PLC stores data such as the fixture ID, part ID, operating parameters, and changeover time for this changeover. Based on historical data, the PLC updates the equipment maintenance prediction model to achieve preventative maintenance.

[0010] The third invention provides a computer device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, a code set, or an instruction set. The at least one instruction, the at least one program, the code set, or the instruction set are loaded and executed by the processor to realize a rapid changeover device for multiple models of engine parts in an engine parts processing production line.

[0011] Compared with the prior art, the present invention provides a device and method for rapid changeover of multiple models of engine parts in an engine parts processing production line, which has the following beneficial effects: 1. This solution integrates technologies such as photoelectric detection, machine vision, digital twin, flexible operation of dual-arm robots, and PLC logic control. The change box 1 with unbalanced lever swing arm 4 realizes the interlock detection of picking up and placing of parts 8 and completes the initialization of fixtures and systems. The part ID is confirmed by a dual verification method of manual input and visual recognition. The flexible clamping of parts is completed by a dual-arm robot with interchangeable master and slave arms and force-position hybrid control. The robot's operation accuracy is verified and the target coordinates are generated through visual error prevention, defect detection and coordinate analysis. After full signal verification by PLC, the optimal path is planned by the digital twin model to drive the robot to complete precise assembly. Finally, the change result is visually verified and the full process change data is stored and the equipment maintenance prediction model is updated. 2. This solution achieves multi-level error prevention and collaborative operation of various modules throughout the entire changeover process, effectively avoiding errors in each stage of the changeover and significantly improving the accuracy and efficiency of the changeover. At the same time, it realizes the systematic management of workstation operation data, providing data support for preventive maintenance of equipment and process optimization, and significantly improving the part processing qualification rate and the intelligent operation level of the workstation.

[0012] This solution focuses on precise fixture changeover detection, multi-dimensional error prevention and verification, robot collaborative operation, digital twin optimization, and full-process data management to achieve efficient and accurate fixture changeover. It consists of six core steps that are interconnected and combine error prevention, accuracy, and intelligence. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of the structure of the shape-changing box part of the present invention; Figure 2 This is a schematic diagram of the photoelectric sensor part of the present invention; Figure 3 for Figure 2 A cross-sectional view along the H direction; Figure 4 This is a structural schematic diagram showing the location of the reflector in this invention; Figure 5 Core flowchart for rapid changeover of multiple parts models; Figure 6 Flowchart for fixture changeover inspection and system initial adjustment; Figure 7 Flowchart for dual verification of part ID; Figure 8 Flowchart for flexible clamping of a dual-arm robot; Figure 9 Flowchart for result verification and data management.

[0015] In the diagram: 1. Changing box; 2. Photoelectric sensor; 3. Limiting shaft; 4. Lever arm; 5. Fixed shaft; 6. Copper bushing; 7. Reflector; 8. Detection part. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0017] This addresses the problems of unstable installation of existing electricity meters, limited expansion methods, unintelligent power supply switching, low data processing efficiency, delayed anomaly detection, poor communication reliability, rigid fault handling, and insufficient human-computer interaction and remote management.

[0018] This solution proposes a device and method for rapid changeover of multiple models of parts in an engine parts processing production line.

[0019] This solution focuses on precise inspection of fixture changeover, multi-dimensional error prevention and verification, robot collaborative operation, digital twin optimization, and full-process data management to achieve efficient and accurate changeover, combining error prevention, accuracy, and intelligence. This solution integrates technologies such as photoelectric detection, machine vision, digital twin, flexible operation of dual-arm robots, and PLC logic control. It confirms the part ID through a dual verification method of manual input and visual recognition, and completes the flexible clamping of parts by relying on a dual-arm robot with interchangeable master and slave arms and force-position hybrid control. Visual error prevention, defect detection and coordinate analysis are used to verify the robot's operation accuracy and generate target coordinates. After full signal verification by PLC, the optimal path is planned by the digital twin model to drive the robot to complete precise assembly. Finally, visual verification of the changeover results is performed and the full process changeover data is stored and the equipment maintenance prediction model is updated. This solution implements multi-level error prevention and collaborative operation of various modules throughout the entire changeover process, effectively avoiding errors in each stage of the changeover.

[0020] Example 1, as Figure 1 As shown in Figure 9, an example of a rapid changeover method for multiple engine parts in an engine parts processing production line provided in this application is illustrated.

[0021] S100: The shape change box 1 with unbalanced lever swing arm 4 and photoelectric sensor 2 detects the picking and placing status of part 8. The sensor transmits the position / absence signal to the PLC for interlock logic judgment. After the judgment is passed, the fixture coding sensor collects the fixture code and stores it in the PLC. The PLC retrieves the corresponding cylinder parameters to drive the fixture to complete the action adjustment. At the same time, the vision module completes the workstation coordinate system calibration, and the digital twin model is synchronously constructed and the virtual mapping body matching the physical equipment is initialized. S200: The operator inputs the part ID on the touch screen. The PLC temporarily stores it. The vision module acquires the part image and performs preprocessing. It extracts multi-dimensional features and matches them with the pre-stored standard model to generate a recognition ID. The PLC compares the manually input ID with the visual recognition ID. If the matching degree reaches the preset threshold, the final ID is confirmed and officially stored. If the standard is not met, the system alarms. The operator checks and then repeats this step. S300: The PLC sends clamping instructions to the dual-arm robot with interchangeable master and slave arms. The slave arm clamps the semi-finished product with a flexible fixture and keeps it suspended. The master arm moves the part to be processed to the corresponding fixture. The two arms work together to calibrate the relative position of the part and the fixture. The contact force is sensed in real time through force-position hybrid control to complete the precise and flexible clamping of the part to be processed, avoiding the deformation of the part due to over-clamping. S400: The vision module acquires images of the clamped parts and matches them with standard clamping images to generate error prevention signals. After the PLC compares the signals and finds them to be consistent, it extracts the features of the part image to detect defects. If there are no defects, the coordinate analysis module establishes a multi-polar coordinate system to obtain polar coordinate sets. After verifying that the robot's operation accuracy meets the standard, the average polar coordinates are converted into the robot's operation target coordinates. S500: The PLC completes fixture change, ID matching, visual error prevention, and comprehensive verification of robot accuracy signals. It plans the optimal motion path of the robot through the post-digital twin model. The PLC retrieves the corresponding work program and sends instructions and target coordinates to the robot. The dual-arm robot works collaboratively along the path to complete the precise assembly of parts and semi-finished products, realizing core change. S600: The vision module re-captures images of the parts, fixtures, and semi-finished products being assembled together. The PLC compares these images with the standard images to complete the final verification. If the images pass the verification, a processing-ready signal is sent. If the images fail the verification, an alarm is triggered and the fault is investigated and readjusted. At the same time, the PLC stores core data such as the fixture ID and part ID of this model change. Based on historical data, the PLC updates the equipment maintenance prediction model to achieve preventive maintenance.

[0022] Step S100 is used for fixture changeover detection and system initial adjustment, including: S110: Photoelectric detection of the pick-up and put-down status of part 8, the specific technical details are as follows: S1101: The device is equipped with an integrated unbalanced lever arm 4, photoelectric sensor 2, fixed shaft 5, limit shaft 3, reflector 7, and a unique storage groove for the detection parts 8. The shape and size of the storage groove are uniquely matched with the detection parts 8. According to the model requirements of the part to be processed, the operator takes out all the detection parts 8 of the corresponding model of the part to be processed from the storage groove of the device, and at the same time, accurately puts the detection parts 8 to be replaced back one by one according to the unique matching position of the storage groove.

[0023] S1102: Due to the gravity characteristics of its own structure, the unbalanced lever arm 4 rotates naturally when no detection part 8 is pressed. Its arm end blocks the light beam emitted by the photoelectric sensor 2 to the reflector plate 7, and the light beam cannot be reflected back to the photoelectric sensor 2. When the detection part 8 is accurately placed into the storage groove, the weight of the detection part 8 presses down on the light end of the unbalanced lever arm 4, causing the arm to rotate around the fixed axis 5. The limiting axis 3 limits the maximum rotation angle of the arm. At this time, the light beam emitted by the photoelectric sensor 2 penetrates the through hole of the base of the changing box 1 and reaches the reflector plate 7. After being reflected by the reflector plate 7, it is accurately transmitted back to the photoelectric sensor 2.

[0024] S1103: The photoelectric sensor 2 detects the status of the detection part 8 corresponding to each storage groove in real time, converts the conduction signal when the detection part 8 is in place and the occlusion signal when it is missing into an electrical signal, forming a set of in-place / missing signals for all detection parts 8, and transmits the set of signals to the PLC control system in real time to provide a signal basis for subsequent logic judgment.

[0025] S120: PLC interlock logic judgment, the technical content is as follows: S1201: The PLC control system receives the presence / absence signal set of the detection part 8 transmitted by the photoelectric sensor 2, and performs interlock logic judgment on the signal set. The judgment rule is that all storage grooves of the detection part 8 of the model to be processed output absence signals, and all storage grooves of the detection part 8 of the model to be replaced output presence signals. Only when the above two conditions are met at the same time, the interlock logic judgment result is passed, and the PLC sends a process continue signal to the subsequent process.

[0026] S1202: If any of the detection parts 8 of the corresponding model of the part to be processed has an in-position signal, or if any of the detection parts 8 of the model to be replaced has an out-of-position signal, it is determined that the interlock logic judgment fails. The PLC immediately triggers the system's audible and visual alarm, and at the same time displays the specific location of the abnormal pick-up and put-down of the detection parts 8 on the human-machine interface touch screen, prompting the operator to readjust the pick-up and put-down status of the detection parts 8 until the PLC interlock logic judgment passes.

[0027] S130: Fixture coding acquisition and cylinder motion adjustment, the technical contents are as follows: S1301: The fixture coding sensor collects the unique code of the fixture in the current workstation in real time, converts the collected fixture coding signal into a digital signal and transmits it to the PLC control system. The PLC stores the digital signal in a dedicated fixture ID register to achieve unique identification of the fixture. S1302: The PLC control system has a pre-stored database containing the association between fixture codes and cylinder action parameters. The database has corresponding cylinder extension stroke, action pressure, action sequence, extension number and other cylinder action setting variables for different fixture codes. The PLC accurately retrieves the corresponding cylinder action setting variables from the association database according to the fixture code stored in the fixture ID register and sends a control signal to the output unit of the cylinder drive module. The output unit adjusts the air pressure, flow rate and on / off sequence of the cylinder according to the control signal, and drives each cylinder of the fixture to complete extension, positioning and other actions according to the retrieved parameters, so that the structural state of the fixture is accurately matched with the clamping requirements of the workpiece to be processed, and completes the self-adaptive adjustment of the fixture. S140: Vision module coordinate system calibration, the technical details are as follows: S1401: The visual recognition and positioning module starts the calibration procedure of the image acquisition unit, uses the standard calibration board as the calibration reference, places the standard calibration board in the preset calibration position of the workstation, and the image acquisition unit performs multi-angle, high-resolution image acquisition on the standard calibration board to obtain the standard image of the calibration board. S1402: The vision module extracts features from the standard image, identifies the coordinates of standard feature points on the calibration board, compares the feature point coordinates with the standard parameters of the calibration board pre-stored in the vision module, calculates the deviation between the actual acquired coordinates and the standard parameters, and corrects the parameters of the image acquisition coordinate system of the vision module based on the deviation value. The correction includes the coordinate origin, coordinate axis direction, pixel equivalent, etc., so that the acquisition coordinate system of the vision module is accurately matched with the physical coordinate system of the robot workstation. After calibration, the calibration parameters are saved, and the basic parameters of the image acquisition unit such as exposure, acquisition frame rate, and gain are initialized to ensure the clarity and accuracy of subsequent image acquisition. S150: Initialization of the virtual mapping body of the digital twin model, the technical content of which is as follows: S1501: The digital twin module starts the virtual modeling program of the robot workstation. Based on the actual physical layout of the workstation, it constructs a virtual mapping body containing core equipment such as fixtures, robots, change box 1, material bins, and processing stations. The size ratio and relative position of the equipment in the virtual mapping body are precisely matched with the physical workstation at a 1:1 ratio. S1502: Through the data transmission interface, real-time data such as fixture ID, actual motion parameters of fixture cylinder, and coordinate system parameters calibrated by vision module stored in PLC are synchronized to digital twin module. Parameter values ​​are assigned to the corresponding devices in virtual mapping. At the same time, the actual physical coordinates of core devices such as robot, change box 1, and fixture are collected by position sensors and converted into digital coordinates of virtual mapping, thus completing the coordinate synchronization between virtual mapping and physical workstation. S1504: Perform motion simulation tests on all devices of the virtual mapping body to verify the consistency between the virtual motion trajectory and the physical motion trajectory of the devices, correct the parameters for deviations that occur during the simulation, complete the initialization of the digital twin model virtual mapping body, and provide an accurate virtual foundation for subsequent robot motion path planning.

[0028] S200 is used for dual verification of part IDs and information storage, specifically: S210: Manual ID input and temporary storage, the technical details are as follows: S2101: The operator enters the part model input interface through the human-machine interaction touch screen of the robot workstation, and enters the unique identification ID of the part to be processed in the interface. This ID contains core information such as the part model, specifications, and processing technology. The touch screen converts the input part ID into a digital signal and transmits it to the PLC control system. S2102: After receiving the manually input part ID signal, the PLC stores it in a dedicated temporary part ID register and displays the input part ID information on the touch screen in real time for the operator to check and confirm. If the operator finds an input error, he / she can directly modify it on the touch screen. The modified information is then updated to the temporary part ID register to ensure that the manually input part ID information is accurate.

[0029] S220: Part image acquisition and preprocessing, the technical content of which is as follows: S2201: The image acquisition unit of the visual recognition and positioning module performs all-round image acquisition of the part to be processed according to the initial acquisition parameters. The acquisition range covers the overall shape of the part, key clamping surfaces, positioning holes and other core parts, and obtains multiple original images of the part. S2202: The vision module preprocesses the acquired raw images. The preprocessing process is as follows: image denoising, grayscale conversion, edge enhancement, and image cropping. The median filtering algorithm is used to remove salt-and-pepper noise and Gaussian noise from the raw images while preserving the edge features of the images. Converting color images to grayscale images reduces the amount of image data and improves the efficiency of subsequent processing. The Sobel operator is used to enhance the edges of the grayscale image, highlighting key features such as the outline and positioning holes of the parts. The enhanced image is cropped according to the actual contour of the part, removing background and invalid areas, and retaining only the valid image containing the main body of the part, thus providing high-quality image data for subsequent feature extraction.

[0030] S230: Feature extraction and visual ID matching, the technical content of which is as follows: S2301: The vision module performs multi-dimensional feature extraction on the pre-processed effective image. The extracted features include the shape and size features of the part, the position and size features of the positioning hole, the contour features of the clamping surface, and the texture features of the part surface. The extracted multi-dimensional feature data is converted into a digital feature vector, which serves as the visual recognition feature of the part. S2302: The PLC control system retrieves the pre-stored standard feature model library of all parts from the data storage module. Each part's standard ID in the model library is matched with a corresponding standard feature vector. The vision module matches the extracted feature vector of the part to be processed with each standard feature vector in the standard feature model library, using the Euclidean distance algorithm to calculate the similarity between the feature vector of the part to be processed and each standard feature vector. The calculation formula is as follows: ; Where d(x,y) is the Euclidean distance. The first feature vector of the part to be processed One portion, The first eigenvector of the standard feature vector One portion, The dimension of the feature vector is denoted by Euclidean distance; a smaller Euclidean distance indicates higher feature similarity. The part ID corresponding to the standard feature vector with the highest similarity is used as the visual recognition ID, and this visual recognition ID and similarity value are transmitted to the PLC control system.

[0031] S240: Dual ID consistency comparison and formal storage, the technical content is as follows: S2401: The PLC control system retrieves the manually input part ID from the temporary part ID register and simultaneously receives the visual recognition ID and similarity value transmitted by the vision module. It then compares the manually input ID with the visual recognition ID for consistency. The system has a preset similarity threshold, set according to the precision requirements of the part processing, typically 99% or higher. When the visual recognition similarity value reaches the preset threshold and the manually input ID and the visual recognition ID are completely identical, the dual ID consistency comparison is considered successful. The PLC then transfers the part ID from the temporary register to the formal part ID register, completing the unique identification and formal storage of the part ID, and simultaneously sends a process continue signal to subsequent processes. S2402: If the similarity value does not reach the preset threshold, or if the manually entered ID and the visually recognized ID are inconsistent, it is determined that the dual ID consistency comparison fails. The PLC immediately triggers the system's audible and visual alarm and displays the manually entered ID, visually recognized ID, and similarity value on the touch screen, prompting the operator to verify the actual model of the part to be processed and the input information. After the operator reconfirms, they must either correct the manually entered ID on the touch screen or replace the part to be processed and repeat all operations in this step until the dual ID consistency comparison passes.

[0032] Specifically, S300 is used for flexible clamping of a dual-arm robot, including: S310: Clamping command generation and transmission, the technical content of which is as follows: S3101: The PLC control system retrieves the corresponding clamping operation parameters from the pre-stored robot operation parameter library based on the part ID stored in the part ID formal register. These parameters include the function allocation of the robot master and slave arms, clamping force threshold, part transfer path, clamping positioning coordinates, etc. S3102: The PLC generates clamping operation instructions for the dual-arm robot based on the retrieved clamping operation parameters. These instructions include core components such as motion instructions, position instructions, and force control instructions for the master and slave arms. The clamping operation instructions are transmitted to the control system of the dual-arm robot in real time via industrial Ethernet. After receiving the instructions, the robot control system parses them, confirms the functional allocation of the master and slave arms and the execution parameters of each action, and prepares for subsequent clamping operations.

[0033] S320: Semi-finished product clamping and suspension positioning, technical details are as follows: S3201: The control system of the dual-arm robot activates the function of the driven arm according to the parsed clamping operation instructions. The flexible gripper mounted on the driven arm completes the adaptive adjustment of the gripper opening according to the parameters corresponding to the part ID. The gripper of the flexible gripper is made of elastic wear-resistant material, which has buffering and anti-slip properties to avoid damage to the semi-finished product during the clamping process. S3202: The boom moves to the preset position of the semi-finished product bin according to the path planned by the instruction. The actual position of the semi-finished product is confirmed by visual positioning. After fine-tuning the movement path, the gripper of the flexible fixture slowly closes to stably clamp the assembled semi-finished product. The clamping force is detected in real time by the force sensor to ensure that the clamping force is within the preset clamping force threshold range, which ensures the stability of clamping and avoids deformation of the semi-finished product due to excessive clamping. S3203: After clamping is completed, the boom moves to the preset suspended position according to the instruction. This position avoids other equipment in the workstation and maintains a preset distance from the clamping station. The boom remains stationary, realizing the suspended positioning of the semi-finished product and providing working space for subsequent parts clamping. S330: Transfer of parts to be processed, the technical details are as follows; S3301: The main arm of the dual-arm robot is activated according to the clamping operation command. The gripper of the main arm adjusts its opening according to the shape and size of the part to be processed. The main arm moves to the material bin position of the part to be processed according to the path planned by the command. Through the real-time feedback of the vision positioning module, the actual placement position of the part to be processed is confirmed, and the movement path is finely adjusted in real time to ensure that the gripper is accurately aligned with the part to be processed. S3302: The main arm's grippers slowly close, stably clamping the workpiece to be processed. The clamping force is detected in real time by a force sensor and controlled within a preset range. After clamping, the main arm moves along the planned transfer path, transferring the workpiece from the hopper to the vicinity of the fixture's clamping station. During the transfer, the workpiece's posture is kept stable to avoid collisions with other equipment in the workstation, ensuring the safety and accuracy of the transfer process. S340: Master-slave boom coordinated position calibration, the technical details are as follows: S3401: After the main arm moves the part to be processed to the vicinity of the fixture clamping station, the vision recognition and positioning module performs real-time image acquisition of the fixture clamping reference, the clamping surface of the part to be processed, and the assembly surface of the semi-finished product held by the driven arm. It obtains position images of the three, extracts features from the images, calculates the relative positional deviation between the part to be processed and the fixture clamping reference, and the relative positional deviation between the part to be processed and the assembly surface of the semi-finished product, and transmits the deviation data to the control system of the dual-arm robot in real time. S3402: The robot control system generates coordinated adjustment instructions for the master and slave arms based on the deviation data. The master arm adjusts the position and posture of the workpiece in real time according to the instructions, and the slave arm makes synchronous fine adjustments to the position and posture of the semi-finished product according to the instructions. The adjustment actions of the master and slave arms are coordinated and consistent, gradually reducing the deviation between the workpiece and the clamping reference of the fixture, and ensuring that the assembly surface of the workpiece and the assembly surface of the semi-finished product are precisely aligned. This ensures that the clamping position of the workpiece matches the clamping requirements of the fixture, preparing the position for subsequent clamping operations.

[0034] S350: Precision Flexible Clamping S3501: After the relative positional deviation between the workpiece to be processed and the clamping reference of the fixture is adjusted to the preset deviation range, the main arm slowly pushes the workpiece to be processed, so that the clamping surface of the workpiece to be processed slowly fits into the clamping reference surface of the fixture. During the fitting process, the force-position hybrid control system senses the contact force and contact position between the workpiece and the fixture in real time. The contact force is detected in real time by the force sensor, and the contact position is detected by both the vision module and the position sensor. S3502: After the workpiece to be processed is fully attached to the clamping reference surface of the fixture, the clamping components such as the locating pins and clamping blocks of the fixture complete the positioning and clamping actions according to the preset action sequence, and accurately clamp the workpiece. During the clamping process, the force-position hybrid control system dynamically adjusts the clamping force of the fixture according to the real-time changes in the contact force to ensure that the clamping force is within the preset clamping force range, avoids excessive clamping that may cause deformation of the workpiece, and ensures the stability of the clamping to meet the accuracy requirements of subsequent processing and assembly. S3503: After clamping is completed, the grippers of the main arm slowly open and separate from the workpiece to be processed. The main arm moves to the preset standby position, and the driven arm maintains the suspended positioning of the semi-finished product, completing the precise and flexible clamping operation of the workpiece to be processed. The robot control system sends a clamping completion signal to the PLC.

[0035] Step S400 is used for visual error prevention and accuracy verification, including: S410: Image acquisition of parts after clamping, the technical details are as follows: S4101: The image acquisition unit of the visual recognition and positioning module acquires multi-angle, high-resolution images of the workpiece after clamping according to preset acquisition parameters. The acquisition range covers the overall clamping state of the workpiece, the contact state between the key clamping surfaces and the fixture, the cooperation state of the positioning holes and positioning pins, and other core parts. It acquires multiple original images of the workpiece after clamping to ensure that the images can completely and clearly reflect the actual clamping state of the workpiece, providing accurate image data for subsequent visual error prevention verification.

[0036] S420: Visual error-proofing signal generation and comparison, the technical content of which is as follows: S4201: The vision module preprocesses the acquired raw images of the clamped parts. The preprocessing process is the same as S220, removing noise and invalid areas from the image, enhancing edge features, and retaining valid images. From the valid images, the clamping feature image of the part to be processed is extracted. This image contains core information such as the part's clamping position, orientation, and engagement status with the fixture. The PLC control system retrieves the standard clamping image corresponding to the part ID from the data storage module. This image represents the standard state image when the part is clamped successfully. S4202: The vision module performs pixel-level matching between the extracted actual clamping feature image and the standard clamping image, uses a template matching algorithm to calculate the matching degree between the two images, generates a visual error prevention signal based on the matching degree result, generates an error prevention qualified signal when the matching degree reaches the preset error prevention threshold, and generates an error prevention unqualified signal when the matching degree does not reach the error prevention threshold, and transmits the visual error prevention signal to the PLC control system. S4203: The PLC stores the received visual error prevention signal in the error prevention signal temporary register. At the same time, it retrieves the preset error prevention signal setting variable corresponding to the part ID from the data storage module. It compares the actual error prevention signal in the error prevention signal temporary register with the preset error prevention signal one by one. If the two are completely consistent, the visual error prevention verification is determined to be successful, and a process continue signal is sent to the subsequent process. If the two are inconsistent, the visual error prevention verification is determined to be unsuccessful, and the PLC triggers the system's audible and visual alarm to prompt the operator to re-execute the part clamping operation.

[0037] S430: Part defect detection, the technical content of which is as follows: S4301: After the visual error-proofing verification is passed, the vision module extracts multi-dimensional feature data from the actual clamping feature image of the part without matching errors. The extracted feature data includes the part's external dimensions, the position and size of the positioning holes, the flatness of the clamping surface, and the integrity of the part's surface. Image morphology methods are used to analyze and process the extracted feature data, performing image dilation, erosion, opening, and closing operations to highlight the part's defect features, such as surface scratches, missing material, cracks, dimensional deviations, and positioning hole misalignment. S4302: The processed feature data is compared with the standard feature data corresponding to the part ID in the data storage module. The deviation value between the actual feature data and the standard feature data is calculated. If all deviation values ​​are within the preset defect allowable range, the part is determined to be defect-free, and a coordinate analysis signal is sent to the coordinate analysis module. If any deviation value exceeds the defect allowable range, the part is determined to be defective. The PLC triggers the system's audible and visual alarm, prompting the operator to remove the defective part, replace it with a qualified part to be processed, and then re-execute the clamping and subsequent verification operations.

[0038] S440: Establishment of Multipolar Coordinate System and Acquisition of Polar Coordinates: S4401: After receiving the coordinate analysis signal indicating a defect-free part, the coordinate analysis module establishes a multi-polar coordinate system based on the workstation coordinate system calibrated by the vision module. The origin of the multi-polar coordinate system is the clamping reference point of the fixture, and the main clamping direction of the fixture is the polar axis. Multiple polar angle directions are set according to the structural features and processing requirements of the part, forming a multi-dimensional multi-polar coordinate system. The vision module performs real-time image acquisition on the clamped part, extracting multiple feature points on the part. These feature points include the contour vertices, positioning hole centers, assembly reference points, etc. The rectangular coordinates of these feature points are converted into polar coordinates in the multi-polar coordinate system. Each feature point corresponds to a set of polar coordinate data, including polar radius and polar angle. S4402: Through multiple acquisitions and conversions, multiple sets of polar coordinate data of the part relative to the dual-arm robot are obtained to form a polar coordinate set, ensuring the comprehensiveness and accuracy of the polar coordinate data, and providing a data foundation for subsequent robot accuracy verification and target coordinate generation; S450: Robot Operation Accuracy Verification S4501: The coordinate analysis module compares the acquired polar coordinate set with the robot's preset operational accuracy threshold. The operational accuracy threshold includes the polar diameter deviation threshold and the polar angle deviation threshold, which are set according to the accuracy requirements of part processing and assembly. It calculates the deviation value between each set of polar coordinate data and the standard polar coordinate data. If all deviation values ​​are within the preset operational accuracy threshold range, the robot's operational accuracy is determined to be up to standard, and an accuracy compliance signal is sent to subsequent processes. S4502: If any deviation value exceeds the operation accuracy threshold range, the robot's operation accuracy is determined to be substandard. The coordinate analysis module sends a maintenance signal to the PLC, the PLC triggers the system to stop, and prompts the operator to debug the robot. After debugging, the polar coordinate acquisition and accuracy verification operations in this step are re-executed until the robot's operation accuracy meets the standard. S460: Target coordinate calculation and transmission: S4601: After the robot's operational accuracy meets the standard, the coordinate analysis module processes all polar coordinate data in the polar coordinate group, using the arithmetic mean method to calculate the average value of the polar radius and the average value of the polar angle. The calculation formula is as follows: ,in This is the average value of the polar diameter. The average polar angle. For the first The polar radius of a set of polar coordinates, For the first The polar angle of the polar coordinate system, where m is the number of polar coordinate systems. By calculating the average value, random measurement errors during multiple data acquisition processes are eliminated, improving the accuracy of the coordinate data; S4602: The coordinate analysis module converts the calculated average polar radius and average polar angle into Cartesian coordinates for the robot workstation. The conversion formula is as follows: Where x is the x-coordinate and y is the y-coordinate in Cartesian coordinates, these Cartesian coordinates are the target coordinates for the robot's operation. The coordinate analysis module transmits the generated target coordinates to the PLC control system in real time. The PLC stores the target coordinates in a dedicated target coordinate register, providing precise position commands for subsequent robot operations.

[0039] Step S500 is used for path planning and changeover operation execution, including: S510: Comprehensive full-signal verification, technical details are as follows: S5101: The PLC control system retrieves qualified signals from various registers and modules, including qualified signals for fixture changeover detection, qualified signals for part ID double verification, qualified signals for visual error prevention verification, defect-free detection signals for parts, and compliance signals for robot operation accuracy. It then performs a comprehensive verification on all of these signals. The rule for comprehensive verification is that if all signals are qualified, the comprehensive verification is considered to have passed; if any signal is unqualified, the comprehensive verification is considered to have failed.

[0040] S5102: During the comprehensive verification process, the PLC verifies the validity of each signal to ensure that there is no loss or error in signal transmission. If an invalid signal is found, the system alarm is immediately triggered, prompting the operator to troubleshoot the signal transmission fault. Only when all signals are valid and qualified will the PLC send a work path planning instruction to the digital twin module to start the subsequent path planning and changeover operation execution process. If the comprehensive verification fails, the PLC triggers an audible and visual alarm, suspends the subsequent process, and prompts the operator to troubleshoot the fault. After the fault is troubleshooted, the corresponding verification operation is re-executed until the comprehensive verification of all signals passes.

[0041] S520: Optimal path planning for digital twin models, the technical content of which is as follows: S5201: After receiving the work path planning instruction from the PLC, the digital twin module retrieves the initialized workstation virtual mapping body. Simultaneously, it retrieves the target coordinates of the robot operation from the PLC's target coordinate register, retrieves the part ID from the part ID register, and retrieves the robot operation constraints corresponding to the part ID from the pre-stored operation parameter library. These constraints include the robot's motion speed, acceleration, joint angle limits, obstacle avoidance requirements, etc. Based on the virtual mapping body, and combining the robot's target coordinates and operation constraints, the digital twin module uses Algorithm A to plan the robot's motion path. The evaluation function of Algorithm A is: f(n) = g(n) + h(n), where f(n) is the evaluation value of node n, g(n) is the actual cost from the starting node to node n, and h(n) is the estimated cost from node n to the target node.

[0042] S5201: The algorithm described above searches for multiple feasible paths from the robot's current standby position to the target coordinates. The movement distance, movement time, number of joint rotations, and obstacle avoidance effect of each feasible path are comprehensively evaluated, and the optimal path is selected as the robot's working path. After path planning is completed, the digital twin module performs motion simulation on the optimal path to verify whether there are any interference or stuttering issues in the robot's movement along that path. If problems are found during the simulation, the path is immediately corrected until the simulation verification is successful. The planned optimal motion path data is then transmitted to the PLC control system.

[0043] S530: Robot operation program retrieval, the technical content of which is as follows: S5301: After receiving the optimal motion path data transmitted by the digital twin module, the PLC control system retrieves the corresponding robot operation program from the pre-stored robot operation program library according to the part ID stored in the part ID formal register. This program contains core contents such as the action sequence, motion speed, posture adjustment, force control parameters, etc. of the robot master and slave arms. S5302: The PLC integrates the retrieved robot operation program with the optimal motion path data to generate a suitable robot operation instruction. This instruction includes both the action requirements of the operation program and the position requirements of the optimal path, ensuring that the robot's operation actions match the motion path and providing a complete instruction basis for the robot's precise operation. S540: Collaborative Type Change Operation by Dual-Arm Robots S5401: The PLC control system transmits the fused robot operation instructions and target coordinates to the dual-arm robot control system in real time via industrial Ethernet. The robot control system parses the instructions, clarifying the core requirements such as the operation actions, motion paths, and target coordinates of the master and slave arms. Based on the parsed instructions, the dual-arm robot initiates collaborative operation. The master arm moves from the standby position to the clamping position of the part to be processed according to the optimal motion path and target coordinates, precisely fine-tuning the position and posture of the part to ensure accurate alignment between the part's assembly surface and the semi-finished product's assembly surface. During this fine-tuning process, the vision module acquires images in real time and provides position feedback to the robot control system, achieving closed-loop control. S5402: The boom adjusts the posture of the semi-finished product synchronously with the main boom according to the instructions, while keeping the semi-finished product suspended and clamped. This is coordinated with the adjustment action of the main boom, so that the assembly surface of the semi-finished product and the assembly surface of the part are always accurately aligned. S5403: During the coordinated adjustment of the master and slave arms, the force-position hybrid control system senses the contact force between the part and the semi-finished product in real time. Based on the change of the contact force, it dynamically adjusts the robot's movement speed and posture. When the assembly surface of the part and the semi-finished product begins to contact, it reduces the movement speed and controls the contact force within the preset assembly force range to avoid excessive impact force during the assembly process, which could cause deformation of the part or semi-finished product. S5404: Through the coordinated operation of the master and slave arms, combined with the force-position hybrid control and the buffering effect of the flexible fixture, the precise assembly of the parts to be processed and the semi-finished products is achieved. After the assembly is completed, the master and slave arms move to the preset standby position according to the instructions to complete the core changeover operation of multiple models of parts. The robot control system sends the changeover operation completion signal to the PLC.

[0044] Step S600 is used for result verification and data management, including: S610: Image acquisition of replacement results, technical details are as follows: S6101: The image acquisition unit of the visual recognition and positioning module performs all-round, high-resolution image acquisition of the overall fit status of parts, fixtures, and semi-finished products after the changeover operation is completed, according to the preset acquisition parameters. The acquisition range covers the clamping fit status of parts and fixtures, the assembly fit status of parts and semi-finished products, the fitting status of key assembly surfaces, the fit status of positioning parts, and other core parts. It acquires multiple original images of the changeover results to ensure that the images can completely and clearly reflect the actual effect of the changeover operation and provide accurate image data for subsequent final verification.

[0045] S620: Final verification of the model change results, the technical details are as follows: S6201: The vision module preprocesses the acquired raw images of the changeover results, removing noise and invalid areas, enhancing edge features, and retaining valid images. It extracts feature images of the changeover results from the valid images. These images contain core information such as the mating features between the part and the fixture, and the assembly features between the part and the semi-finished product. The PLC control system retrieves the standard pre-processing image corresponding to the part ID from the data storage module. This image represents the standard state of the part before it enters the processing stage after the changeover operation has passed. S6202: The PLC will accurately match the extracted feature image of the actual transformation result with the standard image before processing. The feature point matching algorithm will be used to calculate the matching degree of the two images. If the matching degree reaches the preset final verification threshold, the transformation result will be deemed to be qualified for final verification. The PLC will send a processing ready signal to the robot workstation, and the workstation will enter the subsequent part processing procedure. S6203: If the matching degree does not reach the preset final verification threshold, the final verification of the changeover result is determined to be unqualified. The PLC immediately triggers the system's audible and visual alarm and displays the specific location and deviation value of the verification deviation on the human-machine interface touch screen, prompting the operator to troubleshoot the fault. Fault types include loose fixtures, assembly deviations, inaccurate part positioning, etc. After the fault is troubleshooted, the operator re-executes the clamping, changeover and verification operations until the final verification of the changeover result is qualified. S630: Transformation data classification and storage, the technical details are as follows: S6301: After the final verification of the changeover result is passed, the PLC control system initiates the data storage process to classify and organize the core data of the entire changeover operation. The data classification includes five categories: fixture-related data, part-related data, operation parameter data, changeover process data, and verification result data. Fixture-related data includes fixture ID, fixture code, cylinder action parameters, etc.; part-related data includes part ID, manual input ID, visual recognition ID, feature vector data, etc.; operation parameter data includes robot operation target coordinates, motion path data, clamping force parameters, assembly force parameters, etc.; changeover process data includes the execution time of each step and the total changeover time, etc.; verification result data includes verification signals, matching degree values, deviation values, etc. of each stage.

[0046] S6302: The PLC transmits the categorized and organized core data to the data storage module in real time through the data transmission interface. The data storage module adopts a structured storage method, storing the data according to the preset database table structure, establishing a unique data file for each changeover operation, ensuring the integrity, accuracy and traceability of the data, and supporting real-time query, export and statistical analysis of the data. S640: Equipment maintenance prediction model update, the technical details are as follows: S6401: After receiving the core data of this changeover operation, the data storage module integrates this data with all the data from historical changeover operations to form a large database for changeover operations on the robot workstation. Big data analytics algorithms are used to comprehensively analyze the data in the database, including trends in the operating parameters of each device, changes in the time consumed in each step of the changeover operation, changes in the matching degree of each verification step, and the frequency and type of faults. S6402: Based on the analysis results, the parameters of the equipment maintenance prediction model are dynamically updated. This model includes maintenance thresholds, maintenance cycles, and fault warning parameters for core equipment such as robots, fixtures, vision modules, sensors, and cylinders. S6403: When the operating parameters of the equipment are close to the maintenance threshold, the model automatically generates an equipment maintenance reminder signal and transmits it to the PLC control system. The PLC displays the maintenance reminder information on the human-machine interface touch screen, prompting the operator to perform preventive maintenance on the corresponding equipment, promptly identify potential equipment failures, reduce the equipment failure rate, extend the equipment service life, and ensure the stable operation of the robot workstation. Meanwhile, through continuous analysis of changeover operation data and model updates, equipment maintenance strategies can be continuously optimized, the accuracy and effectiveness of maintenance can be improved, and data support can be provided for the intelligent operation of robot workstations.

[0047] Experimental example: I. Experimental Procedure Step 1: Pre-experiment preparation The test parts 8 of the engine cylinder head of models A and B are placed into the storage groove of the test part 8 change box 1 according to the unique matching position. The jig cylinder action parameters, standard feature model, standard clamping image and standard pre-processing image corresponding to the parts of models A and B are pre-stored in the PLC control system. The part ID matching similarity threshold is set to 99%, the visual error prevention matching threshold is set to 98%, and the robot operation accuracy threshold is set to ±0.03mm. The basic mapping between the digital twin model and the physical equipment of the robot workstation is completed, and the lens parameters of the vision module are calibrated.

[0048] Step 2: Perform S100 fixture changeover test and system initial adjustment This experiment involved replacing model B parts with model A parts. The operator removed all model A inspection parts (8) from the changeover box 1 and precisely placed model B inspection parts (8) back into their storage slots. The photoelectric sensor 2 in the changeover box 1 detected that all model A inspection parts (8) were missing and all model B inspection parts (8) were in place, transmitting the in-place / out-of-place signal to the PLC. The PLC then processed the signal using interlock logic. The fixture coding sensor collected the current fixture code and transmitted it to the PLC. The PLC retrieved the cylinder action parameters corresponding to the model A parts and drove the fixture cylinder to adjust its extension stroke and pressure. The extension stroke was adjusted to 85mm, and the working pressure to 0.7MPa. Simultaneously, the visual recognition and positioning module calibrated the workstation coordinate system, correcting the origin deviation by 0.01mm. The digital twin model synchronously updated the physical positions of the fixture, robot, and changeover box 1, completing the initialization of the virtual mapping.

[0049] Step 3: Perform dual verification and information storage for S200 part IDs. The operator inputs the part ID (model A) on the human-machine interface touchscreen, and the PLC temporarily stores this ID in the part ID temporary register. The visual recognition and positioning module acquires multi-angle images of the model A part to be processed. The original images undergo median filtering for noise reduction, grayscale conversion, Sobel edge enhancement, and image cropping preprocessing to extract multi-dimensional feature data such as the part's dimensions, positioning hole positions, and clamping surface contours. This data is converted into digital feature vectors and matched against a pre-stored standard feature model of model A parts. A similarity of 99.6% is calculated, generating a visually recognized model A part ID. The PLC compares the manually input ID with the visually recognized ID. If the matching degree reaches a preset threshold, the final model A part ID is confirmed and stored in the part ID permanent register.

[0050] Step 4: Perform flexible clamping on the S300 dual-arm robot. The PLC retrieves the corresponding clamping parameters based on the part ID of model A and sends clamping instructions to the dual-arm robot. The robot automatically assigns the driven arm to clamp the semi-finished engine cylinder head, while the main arm moves the model A part to be processed. After the driven arm clamps the semi-finished product with a flexible fixture, it moves to a preset suspended position, with the clamping force controlled at 50N. The main arm picks up the model A part from the hopper and moves it to the vicinity of the fixture clamping station. The vision module acquires position images in real time and calculates the initial positional deviation of 0.12mm between the part and the fixture clamping reference. The PLC sends a collaborative adjustment instruction to the robot, and the main and driven arms work together to adjust the relative position of the part and the fixture. During the adjustment process, the force-position hybrid control system senses the contact force in real time and controls the clamping force at 80N, ultimately completing the precise flexible clamping of model A part. After clamping, the positional deviation is corrected to 0.02mm.

[0051] Step 5: Perform S400 visual error prevention and accuracy verification The visual recognition and positioning module acquires multi-angle images of the clamped part (model A) and performs pixel-level matching with pre-stored standard clamping images of the part (model A). This generates a visual error-proofing pass signal and transmits it to the PLC. The PLC compares this signal with a preset error-proofing signal. The vision module extracts feature data from the clamped part images and uses image morphology methods for inspection, finding no surface scratches, dimensional deviations, or missing materials. The coordinate analysis module establishes a multi-polar coordinate system with the clamping reference point as the origin, acquiring six sets of polar coordinate data relative to the robot. The maximum polar diameter deviation is calculated to be 0.02mm, and the maximum polar angle deviation is 0.01°, both within the preset robot operation accuracy threshold, thus meeting the accuracy verification standard. The coordinate analysis module calculates the average of the six sets of polar coordinate data, converts the averaged polar coordinates to Cartesian coordinates, generates the robot operation target coordinates, and transmits them to the PLC.

[0052] Step 6: Execute S500 path planning and changeover operation. The PLC performs comprehensive verification of all signals, including fixture changeover inspection pass signal, part ID double verification pass signal, visual error prevention pass signal, and robot accuracy compliance signal. All signals pass the verification. The digital twin model, based on an initialized virtual mapping volume and combined with the robot's target coordinates, uses the A* algorithm to plan the optimal motion path for the master and slave arms. The planned path avoids interference with the material bin and changeover box 1, reducing the total movement distance to 1.2m. The PLC retrieves the robot's operating program corresponding to part A, merges it with the optimal path data, and sends the operating instructions and target coordinates to the robot. The two-armed robots work collaboratively along the path. The master arm performs precise positional fine-tuning of the part, while the slave arm adjusts the semi-finished product's posture. Through force-position hybrid control, the assembly contact force is controlled at 60N, completing the precise assembly of part A and the semi-finished product, achieving the core changeover operation.

[0053] Step 7: Perform S600 result verification and data management The visual recognition and positioning module again acquired images of the mating of part A with the fixture, as well as images of the assembly of the part and the semi-finished product. These images were then matched with pre-stored pre-processing images of part A before standard machining. The matching rate was 99.2%, and the PLC determined that the changeover result was ultimately verified and sent a machining-ready signal to the workstation. Simultaneously, the PLC transmitted core data, including the fixture ID, part A ID, cylinder motion parameters, robot target coordinates, execution time of each step, and matching rate of each stage, to the data storage module. The data storage module integrated this data with historical changeover data and dynamically updated the parameters of the equipment maintenance prediction model. In this experiment, all equipment operating parameters were within the normal threshold range, and the model did not generate a maintenance reminder signal.

[0054] Step 8: Repeat the experiment and record the data To verify the reliability of the experimental results, 10 repeated changeover experiments were conducted following the steps described above, changing from part B to part A. For each experiment, data such as fixture changeover judgment time, part ID matching time, robot clamping time, core changeover operation time, total changeover time, clamping deviation after changeover, assembly deviation, and changeover pass rate were recorded. At the same time, the corresponding data of the traditional changeover method (manual visual confirmation of fixture changeover + single manual input of part ID + robot clamping without visual error prevention) were also recorded for comparative analysis.

[0055] II. Experimental Data This experiment involved 10 replacement machining operations of cylinder head parts for engines of models A and B, and 10 comparative experiments using traditional replacement methods. All experimental data were obtained from actual testing. The core experimental data are as follows: Changeover time data: The average fixture changeover judgment time of the method of this invention is 12s, the average part ID matching time is 18s, the average robot clamping time is 45s, the average core changeover operation time is 30s, and the average total changeover time is 125s; the average manual confirmation time for fixture changeover of the traditional changeover method is 60s, the average part ID input time is 10s, the average robot clamping time is 80s, the average core changeover operation time is 40s, and the average total changeover time is 220s.

[0056] Changeover accuracy data: After 10 changeover experiments using the method of this invention, the clamping deviation of the parts was within the range of 0.01-0.02mm, with an average clamping deviation of 0.015mm. The assembly deviation between the parts and the semi-finished products was within the range of 0.02-0.03mm, with an average assembly deviation of 0.025mm. After 10 changeover experiments using the traditional changeover method, the clamping deviation of the parts was within the range of 0.08-0.15mm, with an average clamping deviation of 0.12mm. The assembly deviation was within the range of 0.10-0.18mm, with an average assembly deviation of 0.14mm.

[0057] Changeover pass rate and error prevention capability data: The method of this invention passed all 10 changeover experiments on the first attempt, with a changeover pass rate of 100%. No problems such as incorrect replacement of part 8, incorrect input of part ID, excessive clamping deviation, or failure to detect part defects occurred during the experiment. In the traditional changeover method, there were 2 instances of missing part 8, 1 instance of incorrect input of part ID, and 3 instances of excessive clamping deviation in 10 changeover experiments. After readjustment, the final changeover pass rate was 70%, and the average time increase due to error adjustment in a single changeover was 85 seconds.

[0058] Equipment operation data: After 10 experiments using the method of this invention to complete the model changeover, all core equipment of the workstation operated stably. The acquisition accuracy of the vision module, the operation accuracy of the robot, and the detection accuracy of the sensor did not show significant attenuation. The data storage module stored a total of 10 sets of complete model changeover process data. After the equipment maintenance prediction model was updated based on the data, the maintenance thresholds of the robot joints, clamp cylinders, and photoelectric sensor 2 were refined and corrected. After the correction, the accuracy of maintenance early warning was improved.

[0059] III. Experimental Conclusions This invention significantly improves the changeover efficiency of multi-model parts in robot workstations. Compared with traditional changeover methods, the total changeover time is reduced by an average of 43.18%, and the execution time of each core step is significantly reduced. Specifically, the fixture changeover detection step is shortened by 80% due to the use of photoelectric sensor 2 for automatic detection and PLC logic judgment, replacing manual visual confirmation. The robot clamping step is shortened by 43.75% due to visual positioning and collaborative adjustment. The core changeover operation step is shortened by 25% due to the use of digital twin model to plan the optimal path, avoiding equipment interference and path duplication. This effectively meets the rapid changeover requirements of modern production lines.

[0060] The method of this invention significantly improves the accuracy of multi-model part changeover. The average clamping deviation in 10 experiments is reduced by 87.5% compared with the traditional changeover method, and the average assembly deviation is reduced by 82.14% compared with the traditional changeover method. Through multiple technical means such as fixture changeover photoelectric interlock detection, part ID double verification, visual error prevention verification, and robot accuracy verification, the accuracy control of the entire process is achieved from the source to the completion of the changeover operation, ensuring that the clamping and assembly status of the parts after changeover meets the processing accuracy requirements.

[0061] This invention possesses extremely strong error prevention capabilities, achieving a 100% changeover pass rate, compared to the 70% pass rate of traditional changeover methods. It solves core problems in traditional changeover processes such as incorrect replacement of part 8, incorrect part ID input, and excessive clamping deviation. The method utilizes photoelectric detection of the unbalanced lever arm 4 to achieve interlocked judgment in the handling of part 8, preventing fixture changeover errors at the source. It employs a dual verification method of manual input and visual recognition for part IDs, avoiding human error from single input. Through visual error prevention verification and part defect detection, it proactively eliminates the influence of clamping deviations and part defects, achieving end-to-end error prevention.

[0062] The method of this invention realizes the systematic storage of data throughout the entire changeover process and the dynamic updating of the equipment maintenance prediction model. By fully storing data such as fixtures, parts, operating parameters, and verification results for each changeover, it provides data support for the process optimization of the workstation. At the same time, by updating the equipment maintenance prediction model based on historical data, it can realize preventive maintenance of equipment, improve the stability and service life of equipment operation, and promote the intelligent operation of robot workstations.

[0063] The rapid changeover method for multiple parts in a robot workstation of the present invention has good practicality and repeatability. Ten repeated experiments have consistently achieved accurate, rapid, and error-proof changeover. All technical links work together smoothly without any obvious system lag or signal transmission problems. It can adapt to the multi-model changeover processing needs of precision parts such as engine cylinder heads in industrial production and can be widely used in robot processing workstations in the automotive, machinery manufacturing and other fields, and has high industrial promotion value.

[0064] Example 2: A rapid changeover device for multiple models of engine parts in an engine parts processing production line, comprising the following modules; It includes a fixture changeover detection module, a PLC control module, a vision recognition and positioning module, a dual-arm robot operation module, a digital twin path planning module, and a data storage and maintenance module; The fixture change detection module includes a change box 1 with an unbalanced lever arm 4, a photoelectric sensor 2, and a fixture coding sensor. The change box 1 is provided with a unique storage groove that matches the detection part 8. The photoelectric sensor 2 transmits the part in place / out of place signal to the PLC control module. The PLC control module connects to each module and performs interlock logic judgment, ID matching and comparison, signal comprehensive verification and instruction sending. The visual recognition and positioning module completes coordinate system calibration, image acquisition and preprocessing, feature extraction and matching, defect detection and coordinate analysis; The dual-arm robot operation module has an interchangeable master and slave arm structure, equipped with flexible fixtures and a force-position hybrid control system to realize parts transfer, flexible clamping and collaborative assembly. The digital twin path planning module constructs a virtual mapping body for the workstation and plans the optimal motion path for the robot; the data storage and maintenance module stores the entire replacement process data, dynamically updates the equipment maintenance prediction model, and realizes preventive maintenance.

[0065] The above-described device is used to implement a method for rapid changeover of multiple models of parts in an engine parts processing production line according to Embodiment 1.

[0066] Example 3: A computer device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to realize a rapid changeover device for multiple models of engine parts in an engine parts processing production line.

[0067] 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 rapid changeover device for multiple models of parts in an engine parts processing production line, characterized in that, Includes the following modules: It includes a fixture changeover detection module, a PLC control module, a vision recognition and positioning module, a dual-arm robot operation module, a digital twin path planning module, and a data storage and maintenance module; The fixture change detection module includes a change box with an unbalanced lever arm, a photoelectric sensor, and a fixture coding sensor. The change box has a unique storage groove that matches the detected part. The photoelectric sensor transmits the part in place / out of place signal to the PLC control module. The PLC control module connects to each module and performs interlock logic judgment, ID matching and comparison, signal comprehensive verification and instruction sending. The visual recognition and positioning module completes coordinate system calibration, image acquisition and preprocessing, feature extraction and matching, defect detection and coordinate analysis; The dual-arm robot operation module has an interchangeable master and slave arm structure, and is equipped with a flexible gripper and a force-position hybrid control system to realize parts transfer, flexible clamping and collaborative assembly. The digital twin path planning module constructs a virtual mapping body for the workstation and plans the optimal motion path for the robot; the data storage and maintenance module stores the entire replacement process data, dynamically updates the equipment maintenance prediction model, and realizes preventive maintenance.

2. The multi-model quick changeover device for engine parts processing production line according to claim 1, characterized in that: The change box also includes a base, a fixed shaft, a limiting shaft, a copper bushing, and a reflector. The unique storage groove is formed in the base. The unbalanced lever arm is sleeved on the fixed shaft through the copper bushing. The limiting shaft restricts the rotation angle of the unbalanced lever arm. The photoelectric sensor and the reflector form a beam transmission path through the through hole in the base. The unbalanced lever arm blocks the beam due to gravity or the detected part is pressed to conduct the beam, thereby generating an in-situ / out-of-situ signal of the detected part.

3. The multi-model quick changeover device for engine parts processing production line according to claim 1, characterized in that: The visual recognition and positioning module includes an image acquisition unit, a feature extraction unit, a defect detection unit, and a coordinate analysis unit. The image acquisition unit completes workstation coordinate system calibration and multi-dimensional image acquisition. The feature extraction unit extracts digital feature vectors from the preprocessed image and matches them with a pre-stored standard model. The defect detection unit detects part defects using image morphology methods. The coordinate analysis unit establishes a multi-polar coordinate system and completes the conversion between polar coordinates and Cartesian coordinates.

4. The multi-model quick changeover device for engine parts processing production line according to claim 1, characterized in that: The flexible gripper of the dual-arm robot operation module is made of elastic and wear-resistant material. The force-position hybrid control system is equipped with force sensors and position sensors to detect the contact force and relative position of the parts in real time during the clamping and assembly process. The contact force is controlled within a preset threshold range to achieve non-damaging and precise clamping and assembly of the parts.

5. The multi-model quick changeover device for an engine parts processing production line according to claim 1, characterized in that: The digital twin path planning module constructs a virtual mapping body that matches the physical workstation 1:1, synchronizes the actual physical position and virtual coordinates of the fixture, robot, and change box, uses a path optimization algorithm to plan the optimal motion path of the robot, and can perform motion simulation verification on the planned path to avoid equipment interference and jamming problems.

6. The multi-model quick changeover device for an engine parts processing production line according to claim 1, characterized in that: The data storage and maintenance module adopts a structured storage method, which classifies and stores fixture IDs, part IDs, cylinder action parameters, robot operation parameters, changeover process data, and verification results of each stage. Based on big data analysis of historical and real-time data, it dynamically updates the maintenance thresholds and early warning parameters of the equipment maintenance prediction model, thereby realizing preventive maintenance reminders for core equipment.

7. A method for rapid changeover of multiple models of parts in an engine parts processing production line, used to implement the apparatus described in any one of claims 1-6, characterized in that, include: S100: The part picking and placing status is detected by the changing box containing an unbalanced lever swing arm and photoelectric sensors. The sensor will transmit the position / absence signal to the PLC for interlock logic judgment. After the judgment is passed, the fixture coding sensor collects the fixture code and stores it in the PLC. The PLC retrieves the corresponding cylinder parameters to drive the fixture to complete the action adjustment. At the same time, the vision module calibrates the workstation coordinate system and the digital twin model completes the initialization of the virtual mapping body. S200: The operator inputs the part ID on the touch screen. The PLC temporarily stores the ID. The vision module acquires the part image and preprocesses it, extracts multi-dimensional features and matches them with the pre-stored standard model to generate a recognition ID. The PLC compares the manually input ID with the visual recognition ID. If the matching degree reaches the preset threshold, the final ID is confirmed and officially stored. If the standard is not met, the system alarms and re-verifies. S300: The PLC sends instructions to the dual-arm robot with interchangeable master and slave arms. The slave arm holds the semi-finished product in the air with a flexible fixture, and the master arm moves the part to be processed to the corresponding fixture. The two arms work together to calibrate the relative position of the part and the fixture. The contact force is sensed through force-position hybrid control to avoid the part being deformed due to excessive clamping, thus completing the precise and flexible clamping of the part to be processed. S400: The vision module acquires images of the clamped parts and matches them with standard clamping images to generate error-proofing signals. After the PLC compares the signals and confirms they match, it extracts the features of the part image to detect defects. If no defects are found, the coordinate analysis module establishes a multi-polar coordinate system to obtain the polar coordinate set of the part, verifies the robot's operating accuracy, and converts the average polar coordinates into the robot's target coordinates after the accuracy is achieved. S500: The PLC completes fixture change, ID matching, visual error prevention, and comprehensive verification of robot accuracy signals. After passing the verification, the digital twin model plans the optimal motion path for the robot. The PLC retrieves the corresponding work program and sends instructions and target coordinates to the robot. The dual-arm robot works collaboratively along the path to complete the precise assembly of parts and semi-finished products, realizing the core change operation. S600: The vision module once again captures the assembly images of the parts, fixtures, and semi-finished products. The PLC compares these images with the standard images to complete the final verification. If the images are qualified, a processing ready signal is sent. If they are not qualified, an alarm is triggered and the fault is investigated and readjusted. At the same time, the PLC stores the fixture ID, part ID, operation parameters, and changeover time of this changeover. Based on historical data, the PLC updates the equipment maintenance prediction model to achieve preventive maintenance.

8. A computer device, characterized in that, The computer device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to realize a multi-model rapid changeover device for an engine parts processing production line as described in any one of claims 1 to 6.

Images