Automatic feeding control system and control method for labeling machine

Abstract

The invention relates to an automatic feeding control system and a control method for a labeling machine, which are characterized in that an intelligent control system and a dynamic self-adaptive PID (Proportion Integration Differentiation) algorithm are additionally arranged, so that a traditional labeling machine is upgraded into intelligent equipment with autonomous sensing, accurate control and data tracing functions. A PLC + vision + servo modular architecture is adopted in hardware, hierarchical design and algorithm integration are achieved in software, and a real-time data interaction network is constructed through PROFINET and an industrial internet. Practical application shows that the labeling precision and efficiency can be remarkably improved, the labor cost and the defective product rate are reduced, and a feasible scheme is provided for intelligent upgrading of the manufacturing industry.

Description

Technical Field

[0001] This application relates to the field of labeling automation technology, and in particular to an automatic feeding control system and control method for a labeling machine. Background Technology

[0002] Traditional labeling machines have the following limitations in practical applications: Low control precision: It relies on manual adjustment of the rolling gap and pressure, with pressure fluctuations exceeding ±3N, resulting in a high label wrinkling rate (not less than 3%), and also a high deviation in the labeling position of cylindrical materials. Low changeover efficiency: Mechanical parts need to be manually changed for different batches of materials, and the changeover time is up to 30 minutes, which cannot meet the needs of small-batch, multi-variety production. Lack of data traceability: The lack of real-time sensing and data storage mechanisms makes it impossible to trace back quality problems, and the optimization of process parameters relies on experience; Low level of intelligence: No predictive maintenance function, equipment downtime averages 8 hours / month; single unit operates independently and cannot be connected to the factory MES system for global scheduling; Traditional equipment is unable to meet the modern labeling requirements of high precision, high flexibility, and intelligence, and urgently needs to be upgraded through hardware modularization, algorithm adaptation, and data interconnection. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of this application is to provide an automatic feeding control system and control method for a labeling machine.

[0004] An automatic feeding control system for a labeling machine includes a support platform 1, a labeling host 10, a conveying device 11, a rolling structure 3, and a limiting device 12, and further includes: Main controller; A visual positioning unit is installed above the output end of the labeling host 10 and is used to acquire material images and calculate the material position and diameter. The sensing and detection unit includes an encoder disposed on the transport device 11 and a pressure sensor embedded in the rolling structure 3; The execution drive unit includes a servo adjustment mechanism that drives the rolling structure 3 to adjust the spacing; and The edge computing unit, which is communicatively connected to the main controller and the vision positioning unit, is configured to execute a visual image processing algorithm and a dynamic adaptive PID control algorithm. The dynamic adaptive PID control algorithm is used to adjust the pressure control parameters of the pressure sensor in real time according to the material diameter obtained by the vision positioning unit.

[0005] Preferably, the main controller is a programmable logic controller (PLC), which communicates with the execution drive unit via a PROFINET bus and with the vision positioning unit and the edge computing unit via Ethernet.

[0006] Preferably, the servo adjustment mechanism includes a servo motor and a planetary gear reducer. The servo motor is connected to the control threaded rod 43 of the rolling structure 3 for electrically adjusting the distance between the roller structure 32 and the extrusion structure 33 in the rolling structure 3.

[0007] Preferably, the dynamic adaptive PID control algorithm is configured to: calculate the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd by interpolation based on the preset fuzzy subset to which the material diameter belongs.

[0008] Preferably, the system further includes a human-machine interaction unit and a communication module. The human-machine interaction unit is used for recipe management, parameter setting and alarm display, and the communication module supports data interaction with a manufacturing execution system (MES) or a cloud platform.

[0009] Preferably, the communication module adopts a layered communication architecture, including the PROFINETIRT protocol for real-time device control, the Modbus TCP / IP protocol for monitoring data interaction, and the MQTT protocol for cloud data upload.

[0010] Preferably, the limiting device 12 is a telescopic limiting component driven by an electric cylinder.

[0011] Preferably, the pressure sensor is embedded inside the extrusion plate 333 of the extrusion structure 33.

[0012] Preferably, the system further includes a database module for storing production data, process parameters, and alarm records.

[0013] An automatic feeding control method for a labeling machine, applied to the aforementioned automatic feeding control system for a labeling machine, includes the following steps: S1: System initialization, loading the process parameters corresponding to the selected material model; S2: Material conveying and visual positioning, the material is conveyed through the conveying device 11 and the visual positioning unit is triggered to obtain the material diameter and position deviation; S3: Dynamic parameter pre-adjustment: Based on the material diameter, the initial pressure control parameters are determined by the dynamic adaptive PID algorithm, and the servo adjustment mechanism is controlled to adjust the initial spacing of the rolling structure 3. S4: Labeling execution and synchronization control, controlling the labeling host 10 and the transport device 11 to synchronize their speeds, and controlling the limit device 12 to complete the labeling; S5: Rolling pressure closed-loop control. When the material enters the rolling structure 3, the extrusion pressure is adjusted in real time according to the feedback from the pressure sensor through the dynamic adaptive PID algorithm. S6: Data collection and traceability. Collect and analyze key data in the production process, and upload the data to the cloud platform for quality assessment and process optimization.

[0014] In summary, this application includes at least one of the following beneficial effects: 1. Intelligent control and precise adjustment Dynamic adaptive PID algorithm: Adjusts PID parameters (Kp, Ki, Kd) in real time through fuzzy logic to adapt to material diameter changes of 20-150mm, achieves pressure control accuracy of ±1N, reduces overshoot by 80% and shortens settling time by 75% compared to traditional PID.

[0015] Visual positioning and multi-axis collaboration: It adopts a 25-megapixel industrial camera and sub-pixel edge detection, with a positioning accuracy of ±0.05mm; it achieves multi-axis synchronization (error ≤1μs) through the PROFINET IRT protocol, and the labeling speed reaches 120 pieces / minute, which is 50% higher than traditional equipment.

[0016] 2. Hardware modularization and flexible integration Modular architecture: The hardware combination of PLC + vision + servo system supports collaborative control of multiple labeling machines; sensors and actuators are connected via aviation connectors, improving maintenance convenience.

[0017] Adaptive mechanical structure: The rolling structure achieves electric adjustment of the spacing (accuracy ±0.01mm) through the first / second control structure, which greatly shortens the changeover time (from 5 minutes to 15 seconds).

[0018] 3. Data-driven approach and quality traceability End-to-end data acquisition: The edge layer processes pressure, displacement, and image data in real time, and the cloud platform uses an LSTM model to predict equipment health (remaining life error < 5%).

[0019] Quality closed-loop optimization: Based on the 3σ principle, abnormal data is marked, the CPK value is ≥1.33, the product qualification rate is improved, and the response time for quality anomalies is shortened from 2 hours to 5 minutes.

[0020] 4. High efficiency and low consumption Reduced energy consumption: Servo motors and electric cylinders replace pneumatic components, reducing energy consumption; adaptive pressure control reduces material damage rate.

[0021] Labor cost savings: Automated feeding and parameter formula management reduce manual intervention, reducing the number of operators required per shift from 3 to 1. Attached Figure Description

[0022] Figure 1 This is a structural diagram of a labeling machine; Figure 2 yes Figure 1 Enlarged view of section A in the middle; Figure 3 yes Figure 2 Enlarged view of section B in the middle; Figure 4 This is a schematic diagram of the system control components of the automatic feeding control system for the labeling machine; Figure 5 This is a schematic diagram of the hardware and software architecture of the automatic feeding control system for a labeling machine. Figure 6 This is a schematic diagram of the workflow of the automatic feeding control system for the labeling machine. Detailed Implementation

[0023] The technical solution of the automatic feeding control system and control method for a labeling machine of this application will be further described in detail below with reference to the accompanying drawings.

[0024] 1. An automatic feeding control system for a labeling machine like Figure 4 As shown, the system includes an automatic feeding and smoothing labeling machine and an automatic feeding control system integrated on the labeling machine. The composition of the labeling machine and the integration method of its control system, system control logic, etc. will be described below.

[0025] In practice, an automatic feeding control system can simultaneously control multiple "automatic feeding and smoothing labeling machines (referred to as labeling machines)" to work, depending on the processing requirements.

[0026] 2. Introduction to the components of a labeling machine Reference Figure 1 , Figure 2 , Figure 3An automatic feeding and smoothing labeling machine includes a support platform 1, a labeling host 10 mounted on the support platform 1, a transport device 11 mounted on the support platform 1 and located on one side of the output end of the labeling host 10, and a limiting device 12 located on the side of the transport device 11 away from the labeling host 10 and with its output end relative to the labeling host 10. The transport device 11 is arranged along the length of the support platform 1 and extends beyond the support platform 1. The transport device 11 is used to transport materials from the left side of the support platform 1 to the right side of the support platform 1. The section of the transport device 11 located on the right side of the labeling host 10 has rolling structures 3 on its front and rear sides to make the labels on the materials adhere more tightly without affecting the efficiency of the transport device 11 in transporting materials. The rolling structure 3 includes two pieces of rollers respectively set with... Fixed platforms 30 and sliding platforms 31 are located on both sides of the conveying device 11 and are arranged along the length of the conveying device 11. A roller structure 32 is provided on the sliding platform 31 to allow the material to rotate during movement, and an extrusion structure 33 is provided on the fixed platform 30 to extrude the material. The upper surfaces of the fixed platform 30 and the sliding platform 31 are provided with grooves 34 arranged along their own width direction. Multiple grooves 34 are provided along the length of the fixed platform 30 and the sliding platform 31. The fixed platform 30 and the sliding platform 31 are provided with an extrusion structure 33 and a roller structure 32 in the same number as the number of grooves 34 provided on them. The roller structure 32 includes a first sliding plate 320 slidably connected to the sliding platform 31 through the grooves 34. The system includes a positioning plate 321 mounted on the sliding platform 31 on the side of the first sliding plate 320 away from the transport device 11, a push spring 322 located between the first sliding plate 320 and the positioning plate 321 for pushing the first sliding plate 320 closer to the transport device 11, a mounting plate 323 disposed on the side of the first sliding plate 320 near the transport device 11, two fixing plates 324 respectively disposed at the upper and lower ends of the mounting plate 323, and a roller 325 disposed between the two fixing plates 324 and vertically rotatably connected to the fixing plates 324. Multiple rollers 325 are arranged along the length of the fixing plates 324. The extrusion structure 33 includes a second sliding plate 330 slidably connected to the fixed platform 30 via a slide groove 34, and a positioning plate 321 mounted on the side of the second sliding plate 320 away from the transport device 11. The conveying device 11 includes a limiting plate 331 mounted on the fixed platform 30, a compression spring 332 located between the second sliding plate 330 and the limiting plate 331 to push the second sliding plate 330 closer to the conveying device 11, and a compression plate 333 disposed on the side of the second sliding plate 330 closer to the conveying device 11. When the limiting device 12 releases its restriction on the material, the material continues to be transported by the conveying device 11. At this time, the material is transported by the conveying device 11 to the space between the roller 325 and the compression plate 333. The material pushes the roller 325 and the compression plate 333 to compress the pushing spring 322 and the compression spring 332. At the same time, during the transportation process, the circumferential surface of the material contacts the circumferential surface of the roller 325, and due to inertia, the material rotates together with the roller 325.During rotation, the extrusion plate 333 is pushed by the extrusion spring 332 to extrude the material on the other side of the circumference, causing the material to adhere tightly to the roller 325. This, in turn, causes the roller 325 to press the label on the material surface firmly, resulting in a tighter label adhesion. Furthermore, the arrangement of the push spring 322 and the extrusion spring 332 provides a certain amount of elastic space for both the roller 325 and the extrusion plate 333, preventing the material from being jammed by the extrusion structure 33 and the roller structure 32 during transport. This achieves the goal of tighter label adhesion without affecting the efficiency of the transport device 11.

[0027] In this embodiment, a first support rod 326 is provided between the first sliding plate 320 and the mounting plate 323. One end of the first support rod 326 is fixedly connected to the mounting plate 323, and the other end is fixedly connected to the first sliding plate 320. The mounting plate 323 is mounted on the first sliding plate 320 through the first support rod 326. The first support rod 326 enables the bottom surface of the mounting plate 323 to extend directly above the transport device 11 without contacting the sliding platform 31 or the transport device 11, thus preventing damage to the bottom surface of the mounting plate 323 when the roller 325 is pushed and squeezed by the material to push the spring 322.

[0028] In this embodiment, a second support rod 334 is provided between the extrusion plate 333 and the second sliding plate 330. One end of the second support rod 334 is fixedly connected to the extrusion plate 333, and the other end is fixedly connected to the second sliding plate 330. The extrusion plate 333 is mounted on the second sliding plate 330 through the second support rod 334. The second support rod 334 enables the bottom surface of the extrusion plate 333 to extend directly above the conveying device 11 without contacting the fixed platform 30 or the conveying device 11, thus preventing damage to the bottom surface of the extrusion plate 333 when the extrusion plate 333 is pushed by the material to compress the extrusion spring 332.

[0029] Although the roller pressing structure 3 allows the label on the material to adhere more tightly without affecting the efficiency of the conveying device 11 in transporting the material, the diameter of the material produced by the user varies from batch to batch. If the distance between the roller structure 32 and the extrusion structure 33 is too large, the roller structure 32 and the extrusion structure 33 will not be able to effectively apply pressure to the material. If the distance between the roller structure 32 and the extrusion structure 33 is too small, the material will not be able to pass through. To solve this technical problem, in this embodiment, the roller structure 32 and the extrusion structure 33 are located far from the conveying device 11. One side of the device 11 is provided with a first control structure 4 and a second control structure 5 for controlling the distance between the roller structure 32 and the extrusion structure 33. Both the fixed platform 30 and the sliding platform 31 are provided with limiting grooves 35 arranged along their own width direction. The limiting grooves 35 are located on the side of the chute 34 away from the transport device 11 and are collinear with the chute 34. The positioning plate 321 is slidably mounted on the sliding platform 31 through the limiting grooves 35, and the limiting plate 331 is slidably mounted on the fixed platform 30 through the limiting grooves 35. The first control structure 4 includes components fixedly mounted on the sliding platform 31. A U-shaped plate 40 on the side away from the transport device 11, a connecting plate 41 located between the U-shaped plate 40 and the sliding platform 31, connecting rods 42 on the connecting plate 41 in the same number as the roller structures 32, and a control threaded rod 43 passing through the U-shaped plate 40 and rotatably connected to the connecting plate 41 from the side of the U-shaped plate 40 away from the sliding platform 31. The control threaded rod 43 is threadedly connected to the U-shaped plate 40. Multiple connecting rods 42 are respectively connected to the side of the positioning plates 321 of multiple roller structures 32 away from the transport structure, and the other end of each is fixedly connected to the connecting plate 41. The control structure 5 is the same as the first control structure 4, but the second control structure 5 is set on the side of the extrusion structure 33 away from the conveying device 11. When the user needs to process the next batch of materials, he only needs to turn the control threaded rod 43 according to the size of the material. The control threaded rod 43 pulls the connecting plate 41 to move back and forth. The connecting plate 41 moves back and forth, and drives the roller structure 32 and the extrusion structure 33 to move through the connecting rod 42, thereby changing the distance between the roller structure 32 and the extrusion structure 33 to adapt to the new size of the material, thus improving the applicability and practicality of the rolling structure 3.

[0030] Although the first control structure 4 and the second control structure 5 improve the applicability and practicality of the rolling structure 3, they also connect multiple roller structures 32 and extrusion structures 33 in series. This means that if a single roller structure 32 is damaged, the entire first control structure 4 and all roller structures 32 connected to it, or the second control structure 5 and all extrusion structures 33 connected to it, must be replaced. This is cumbersome and significantly increases production costs. To solve this technical problem, this embodiment sets the connecting rod 42 to include a fixing rod 420 and a plug-in rod 421. The fixing rod 420 is mounted on the connecting plate 41, and the plug-in rod 421 is mounted on the roller structure 32 or the extrusion structure 33. The end face of the fixing rod 420 away from the connecting plate 41 has a plug-in interface, and the plug-in rod 421 is connected via a plug-in interface. The insertion rod 421 is connected to the fixed rod 420. The insertion point of the insertion rod 421 and the fixed rod 420 is provided with a fixing structure 6 for fixing the insertion rod 421. The fixing structure 6 includes a positioning port 60 opened on the upper surface of the fixed rod 420 and communicating with the insertion port, a limiting port opened on the upper surface of the insertion rod 421 and cooperating with the positioning port 60, and a limiting rod 61 that passes through the positioning port 60 and extends into the limiting port to fix the insertion rod 421. When one of the multiple roller structures 32 and multiple extrusion structures 33 is damaged, the user only needs to pull out the limiting rod 61 on the corresponding insertion rod 421 of the structure, and then slide the structure to make the insertion rod 421 disengage from the fixed rod 420. In this way, the user can disassemble the structure for individual replacement, saving production costs and making it more convenient for the user to replace the roller structure 32 or extrusion structure 33.

[0031] 3. Integration method of control components on labeling machine The integrated components of the intelligent control system on the labeling machine include: a Siemens S7-1214C DC / DC / DC type PLC integrated inside the support platform (1) as the main control unit, which communicates with each actuator through the PROFINET bus; a Basler acA2500-14uc industrial camera and an 850nm bar light source are installed 300mm above the output end of the labeling host (10) to form a visual positioning unit; an Omron E6B2-CWZ6C encoder is configured at the active roller of the transport device (11) to realize displacement detection; a Honeywell FSG15N1A pressure sensor is embedded in the inner side of the extrusion plate (333) of the extrusion structure (33); a Siemens SIMOTICS S-1FL6 servo motor and planetary gear reducer are added to the end of the control thread rod (43) of the first control structure (4) and the second control structure (5) to replace the original manual adjustment mechanism; and the limit device (12) is upgraded to a Festo EMMS-AS-57-10 electric cylinder drive. The above-mentioned hardware equipment forms an organic whole with the structured layout and mechanical structure. The PLC module is integrated into the electrical cabinet inside the support platform (1). The sensors and actuators are connected to the main control line through aviation plugs to realize the real-time transmission of control signals.

[0032] 4. Detailed Introduction to Intelligent Control Systems and Algorithms 4.1 Overall Architecture of Intelligent Control System To achieve intelligent control of the labeling machine, this embodiment adds an intelligent control system based on an industrial internet architecture. This system adopts a closed-loop control model of "perception-decision-execution-feedback," and mainly consists of three parts: a hardware layer, a software layer, and an algorithm layer. The system architecture is as follows: Figure 4 As shown (Note: a system architecture diagram is required in actual applications), the system collects equipment operation data in real time through a distributed sensor network. After data preprocessing by edge computing nodes, the data is uploaded to the industrial control cloud platform to achieve global optimization decisions. Finally, the system drives the actuators to complete precise operations through a real-time control bus.

[0033] 4.2 Hardware system composition (e.g.) Figure 5 (As shown) 4.2.1 Main Control Unit A Siemens S7-1214C DC / DC / DC PLC was selected as the core controller. This controller integrates 14 digital inputs (24V DC), 10 digital outputs (24V DC), and 2 analog input / output channels, supporting high-speed counting up to 100kHz, which can meet the multi-axis collaborative control requirements during the labeling process. The PLC achieves real-time data interaction with expansion modules and peripheral devices via the PROFINET bus, with a communication cycle of ≤1ms, ensuring the real-time performance of control commands.

[0034] 4.2.2 Sensing and Detection System (1) Visual positioning unit: A Basler acA2500-14uc industrial camera (25 megapixels, 14fps frame rate) is used with a Computar M1214-MP2 lens (12mm focal length, F1.4 aperture), and data is transmitted to the PLC through a gigabit Ethernet interface. The camera is installed 300mm above the output terminal of the labeling host 10, using a top-view shooting method, and is equipped with an 850nm wavelength strip light source (30W power, adjustable illuminance) to achieve sub-pixel level recognition of material outlines.

[0035] (2) Displacement detection module: An Omron E6B2-CWZ6C encoder (1000 lines / revolution, NPN output) is installed on the active roller of the conveyor device 11. The displacement data of the conveyor belt is collected in real time through the high-speed counting interface of the PLC, with a resolution of 0.01mm, providing a reference for material positioning.

[0036] (3) Pressure sensing unit: A Honeywell FSG15N1A pressure sensor (range 0-100N, accuracy ±0.5%FS) is embedded inside the extrusion plate 333 of the extrusion structure 33. It outputs real-time pressure value through a 4-20mA analog signal to monitor the contact pressure during the label rolling process.

[0037] (4) Material detection sensor: An Omron E3Z-LS63 photoelectric sensor (detection distance adjustable from 50 to 300 mm) is installed at the inlet of the conveyor device 11 to trigger the labeling process; a Keyence LR-ZB250CP laser displacement sensor (measurement range 0-250 mm, accuracy ±0.03 mm) is installed at the outlet of the rolling structure 3 to detect the shape of the material after labeling.

[0038] 4.2.3 Execution Driver System (1) Transport drive: Siemens V90 servo drive (power 750W) is used in conjunction with 1FL6044-1AF61-2AA1 servo motor. Speed ​​closed-loop control is achieved through PROFINET. The speed range is 0.1-500mm / s and the position control accuracy is ±0.01mm.

[0039] (2) Rolling adjustment mechanism: A Siemens SIMOTICS S-1FL6 servo motor (power 400W) is installed at the end of the control thread rod 43 of the first control structure 4 and the second control structure 5. It works with a planetary gear reducer (reduction ratio 10:1) to realize the electric adjustment of the roller structure 32 and the extrusion structure 33, with a position resolution of 0.001mm.

[0040] (3) Limit device control: The pneumatic drive of the original limit device 12 is upgraded to a Festo EMMS-AS-57-10 electric cylinder with a stroke of 50mm, a thrust of 500N, and a position repeatability of ±0.02mm. Stepless speed regulation is achieved through analog signals.

[0041] 4.2.4 Human-Computer Interaction and Communication Module A Siemens KTP700 Basic PN touchscreen (7-inch TFT display, 800×480 resolution) is configured as the local operation terminal, supporting functions such as recipe management, parameter setting, and alarm display. The system connects to the factory's local area network via a Siemens CP1243-1 communication module, supports Modbus TCP / IP protocol for interface with the MES system, and enables production data uploading and remote monitoring.

[0042] 4.3 Software System Architecture 4.3.1 Layered Design of System Software The architecture adopts a layered design, including: (1) Equipment layer: The PLC control program is developed based on Siemens TIA Portal V16. The implementation process is as follows: ① Function block (FB) development: The initialization module (FB100) is designed as a power-on self-test process, which sequentially detects the sensor status words (such as encoder ready signal 65535H, pressure sensor 4-20mA signal validity) and driver enable status (servo ready bit 16#0008). When there is an abnormality, the OB82 diagnostic interrupt is triggered; the manual / automatic mode switching module (FB101) receives the switching instruction through the touch screen I / O field (address MW100). In automatic mode, the labeling process control module (FB102) is called. In manual mode, each actuator can be controlled individually (such as Q0.0 controlling the start and stop of the transport device). ② Task Scheduling Mechanism: Hierarchical scheduling is achieved through Organization Blocks (OBs). OB1 (cyclic interrupt, default 100ms) executes the main logic scan and calls FB102 to implement timing control of the labeling process; OB35 (timer interrupt, 20ms) performs high-speed data processing, such as encoder pulse counting (IW64) and position conversion (displacement = number of pulses × 0.01mm / pulse); OB121 (programming error interrupt) and OB122 (I / O access error interrupt) handle abnormal faults. Working Principle: An "event-triggered-state machine" control model is adopted. The device status word (such as DB10.DBX0.0#labeling complete, DB10.DBX0.1#fault alarm) is stored through a global data block (DB10). Each functional block achieves data interaction through input and output parameters (such as FB102.IN#MaterialType, FB102.OUT#CycleTime). Structured programming improves code reusability by 40% and shortens fault location time to within 5 minutes.

[0043] (2) Edge layer: Edge computing software deployed on industrial PCs, developed and implemented using C# language, is as follows: ① Visual image preprocessing: The process is streamlined using the OpenCVSharp library. First, the GaussianBlur function (ksize=5×5, sigmaX=1.5) is called for noise reduction. Then, the Canny function (threshold1=50, threshold2=150, apertureSize=3) is used to extract material edges. Finally, the FindContours function is used to obtain contour coordinates (accuracy CV_CHAIN_APPROX_SIMPLE). ② Sensor data fusion: Multi-source data synchronization is achieved based on encoder timestamps (accuracy 1ms). The pressure sensor (sampling period 50ms) and displacement data (sampling period 10ms) are interpolated to the same time axis using the LinearInterpolation function. The fused data is stored as a structure ({TimeStamp:long, Position:double, Pressure:double}). ③ Multi-threaded parallel processing: A ThreadPool is used to manage three core threads. Thread 1 (priority AboveNormal) handles image acquisition and preprocessing (CPU utilization ≤30%), Thread 2 (priority Normal) executes a dynamic adaptive PID algorithm (iterwise every 10ms), and Thread 3 (priority BelowNormal) interacts with the PLC via the Socket protocol (port 502). Working principle: Task scheduling is implemented using Task ParallelLibrary (TPL) based on .NET Framework 4.7. Inter-thread data communication is achieved through ConcurrentQueue (thread-safe queue capacity 1000 lines). MemoryStream is used to optimize image data memory usage (60% reduction in memory consumption compared to Bitmap). The vision processing thread and algorithm thread are synchronized through AutoResetEvent to ensure that the image recognition result (material center coordinate deviation) is transmitted to the control algorithm module within 20ms.

[0044] (3) Cloud Layer: The implementation process using the Siemens MindSphere cloud platform is as follows: ① Data Interaction Process: The edge layer packages production data into JSON format ({DeviceId:"LM-001",Timestamp:1620000000000, OEE:0.92, Pressure:[12.5,13.2,...]}) via the MQTT protocol (client library M2Mqtt), and uploads it to the cloud platform through port 443 after TLS1.3 encryption. The upload cycle is dynamically adjusted according to the data type (key process parameters 1s / time, statistical data 5min / time). ② Production Data Visualization: A custom dashboard is created based on MindSphere Analytics. The OEE indicator is calculated in real time using the formula "availability (runtime / planned time) × performance rate (actual speed / theoretical speed) × quality rate (number of qualified products / total output)". The capacity statistics module generates bar charts by shift (8:00-16:00) and supports drill-down to view the hourly output fluctuation of a single device. ③ Equipment Health Management: Servo motor vibration data is collected (sampling frequency 2kHz), and an LSTM neural network model is trained using the MindSphere Predictive Maintenance module (input features: root mean square vibration acceleration, rate of temperature change; output: remaining life (RUL)). Maintenance work orders are automatically generated when the predicted life is <500 hours. Working Principle: A microservice architecture is used to decouple functions. The Data Ingestion Service processes raw data, the Time Series Service stores historical data (retained for 365 days), the Analytics Service calls Python scripts to execute algorithm calculations, and the front end uses the Angular framework for data visualization. The REST API uses an OAuth 2.0 authorization mechanism to ensure the security of interface calls (token validity period 30 minutes).

[0045] 4.3.2 Core Software Modules (1) Visual positioning module: The development and implementation process based on the OpenCV open source library is as follows: ① Image acquisition trigger: When the encoder count of the transport device reaches the preset value (corresponding to the position 200mm in front of the labeling host output), the PLC triggers the camera to acquire the signal through digital output (Q0.5). The camera response time is ≤10ms, and the bar light source (PWM duty cycle 80%) is turned on simultaneously. ② Contour extraction and feature recognition: After the image is grayscaled (cvtColor function COLOR_BGR2GRAY), it is binarized using the adaptiveThreshold function (blockSize=11, C=2). Then, the material outer contour is extracted by the findContours function (RETR_EXTERNAL mode), and the contour moments (moments function) are calculated to obtain the centroid coordinates (Cx=M10 / M00, Cy= M01 / M00). ③ Position deviation calculation: Compare the actual centroid coordinates (Cx_act, Cy_act) with the theoretical coordinates (Cx_ref=400px, Cy_ref=300px) to obtain the deviation values ​​ΔX=Cx_act-Cx_ref and ΔY=Cy_act-Cy_ref, and convert them into actual physical deviations (ΔX_mm=ΔX×0.02) using pixel equivalents (0.02mm / px). Working principle: A subpixel-level edge detection algorithm is adopted, and the corner points of the contour are optimized by the cornerSubPix function (search window 5×5, zero region (-1,-1), iteration termination condition ε=0.01), with a positioning accuracy of ±0.05mm; the camera eliminates distortion by Zhang Zhengyou calibration method (internal parameter matrix: fx=1200, fy=1200, cx=400, cy=300; distortion coefficient k1=-0.3, k2=0.1) to ensure the consistency of material imaging at different positions; Hough circle transform (HoughCircles function, dp=1, minDist=50, param1=100, param2=30) is used for the diameter measurement of cylindrical materials, and the actual diameter θ is calculated by the number of pixels of circle diameter × pixel equivalent.

[0046] (2) Motion control module: The implementation process of PLCopen motion control specification is as follows: ① Multi-axis collaborative configuration: In TIA Portal, three servo axes are configured through process objects (TO). Axis 1 (transport device) is configured as a speed axis (process object TO_Speed), and axis 2 (rolling adjustment) and axis 3 (limit device) are configured as position axes (TO_Position). Axis synchronization is achieved through PROFINETIRT protocol (synchronization error ≤1μs). ② Speed ​​synchronization control: The speed of the transport device is controlled by calling the MC_MoveVelocity (FB28) function block. The speed reference value is set (e.g., 300mm / s). A speed closed loop is formed through encoder feedback (ID1000) (proportional gain Kp=0.8, integral time Ti=0.1s). When the vision system detects that the material position deviation ΔX>±0.5mm, the compensation speed is superimposed through MC_AddVelocity (FB32) (ΔV=Kv×ΔX, Kv=0.2mm / s / mm). ③ Phase control and position adjustment: The phase control of the labeling host is achieved through electronic gear synchronization. MC_GearIn (FB31) is called to set the gear ratio (drive shaft: labeling host encoder, driven shaft: material follower shaft, transmission ratio 1:1.2) to ensure that the starting edge of the label is aligned with the material seam. The position adjustment of the rolling mechanism is achieved by calling MC_MoveAbsolute (FB21). The target position is calculated based on the material diameter θ using the formula (d=θ+2×δ, where δ is the rolling gap, default 2mm). The closed-loop position control is achieved through a servo motor encoder (131072 lines / revolution), with a positioning accuracy of ±0.01mm. Working principle: Based on the PLCopen state machine model (such as the states of MC_MoveAbsolute: Idle→Ready→Active→Done), the execution status of the function block is determined by the status word (such as 16#047F indicating readiness); each axis exchanges synchronization signals through the shared data block (DB200) (such as when axis 1 is ready, DB200.DBX0.0=1, axis 2 is triggered to move), and the cross-coupling control algorithm is used to eliminate multi-axis synchronization error (synchronization error ≤0.02mm), meeting the dynamic response requirements of high-speed labeling (120 pieces / minute).

[0047] (3) Data Management Module: The implementation process of using SQLite database to store production data is as follows: ① Database Design: Create 3 core data tables: Material Information Table (Material: ID INTEGER PRIMARY KEY, Type TEXT, Diameter REAL, LabelLength REAL), Production Record Table (Production: ID INTEGER PRIMARY KEY AUTOINCREMENT, MaterialID INTEGER, StartTime DATETIME, EndTime DATETIME, Quantity INTEGER, DefectCount INTEGER), and Equipment Parameter Table (Parameters: ID INTEGER PRIMARY KEY, AxisName TEXT, Kp REAL, Ki REAL, Kd REAL, UpdateTime DATETIME). The tables are linked through foreign keys (MaterialID). ② Data read and write operations: Data interaction is achieved through the System.Data.SQLite library. When production starts, the INSERT statement is executed (INSERT INTO Production(MaterialID,StartTime) VALUES(1, '2023-10-01 08:00:00')). During production, the UPDATE statement is executed every 10 minutes to update the production output (UPDATE Production SET Quantity=Quantity+1200 WHERE ID=1). Historical data can be queried using the SELECT statement (SELECT * FROM Production WHERE StartTime BETWEEN '2023-10-01' AND '2023-10-02'). ③ Backup and recovery mechanism: The backup script is triggered daily at 23:59 by the Windows Task Scheduler. The command VACUUM INTO 'D:\Backup\DB_20231001.db' is executed to generate a backup file. The backup file is uploaded to cloud storage (path / mindsphere / backup / ) via FTP protocol (FileZilla client). Data can be restored from the backup file via the RESTORE command.Working principle: The ACID transaction mechanism is used to ensure data consistency (BEGIN TRANSACTION → execute SQL → COMMIT / ROLLBACK). A composite index is created on the production record table (CREATEINDEX idx_Production_Time ON Production(StartTime)) to improve query efficiency (query time is reduced from 500ms to 50ms). Database connections are managed using a connection pool (maximum number of connections 10). Parameterized queries using IDbCommand (AddWithValue) prevent SQL injection attacks. Data files use WAL (Write-Ahead Logging) mode to reduce write lock contention and ensure data integrity in high-concurrency scenarios (100 writes / second).

[0048] (4) Alarm Management Module: The implementation process of the three-level alarm mechanism is as follows: ① Alarm level definition and triggering conditions: Warning level (ALARM_WARNING): The pressure sensor value exceeds the set range by ±10% (e.g., when the setting is 20N, 18N≤P≤22N is normal, 15N≤P<18N or 22N<P≤25N triggers a warning), which is displayed by flashing yellow text on the touch screen; Fault level (ALARM_FAULT): Servo drive fault code ≠0 (e.g., overload code 16#050A), visual positioning timeout (>500ms without return of result), after triggering, partial shutdown is executed (stop labeling of current material, while keeping the transport device running); Emergency stop level (ALARM_EMERGENCY): Emergency stop button signal (I0.0=0), safety door open (I0.1=0), after triggering, all shaft enable (Q1.0=0) is cut off and the air source is shut off. ② Alarm Handling Process: When an alarm occurs, the PLC writes the alarm code (e.g., 1001# Pressure Abnormality), the occurrence time (obtained via RTC real-time clock), and relevant parameters (current pressure value) into the alarm buffer (DB300, capacity 100 entries). Simultaneously, it drives the audible and visual alarms via digital outputs (Q2.0 controls the buzzer, Q2.1 controls the LED) (Warning level: LED flashing frequency 1Hz, buzzer intermittently sounds; Fault level: LED flashing frequency 2Hz, buzzer continuously sounds; Emergency stop level: LED constantly lit, buzzer continuously sounds). ③ Alarm Information Display and Recording: The touchscreen displays the alarm list in real-time (arranged in reverse chronological order) using the WinCC Flexible AlarmControl control. Clicking on an alarm entry displays detailed handling suggestions (e.g., 1001# Pressure Abnormality suggestion: check the compression spring preload, calibrate the sensor zero point). Alarm data is written to the SQLite database hourly (table AlarmLog: ID, Code, Time, Description, Status), and can be exported to CSV format (by triggering ShellExecute via a button to call Excel). Working principle: Real-time alarm response is achieved based on the PLC interrupt mechanism. Warning level is detected by polling through OB35 timer interrupt (100ms cycle), while fault level and emergency stop level are triggered by hardware interrupt (such as OB40 rising edge interrupt). Alarm priority is implemented through priority encoder circuit. Emergency stop level (highest priority) can interrupt the alarm processing flow of other levels. Alarm reset adopts a two-step mechanism of "confirmation-reset". The operator needs to click the "confirm" button on the touch screen (reset MW200.0) and click the "reset" button after the fault is cleared (reset MW200.1) before the system will clear the alarm state, preventing accidental recovery.

[0049] 4.4 Dynamic Adaptive PID Control Algorithm 4.4.1 Algorithm Principle To address the issues of overshoot and oscillation inherent in traditional PID controllers when material diameter changes, this system employs a dynamic adaptive PID control algorithm. This algorithm achieves precise control of the rolling pressure by adjusting PID parameters in real time. The algorithm formula is as follows: u(t) = Kp(θ)·e(t) + Ki(θ)·∫e(τ)dτ + Kd(θ)·de(t) / dt (Formula 1) in: u(t): Controller output (electric cylinder displacement command, unit: mm) e(t): Pressure deviation (set pressure Pset, actual pressure Pact, unit: N) Kp(θ): Dynamic proportionality coefficient, where θ is the material diameter (unit: mm). Ki(θ): Dynamic integral coefficient Kd(θ): Dynamic differential coefficient The dynamic parameter adjustment adopts fuzzy logic rules, dividing the material diameter θ into 5 fuzzy subsets ({minimal, small, medium, large, maximal}), each subset corresponding to a set of PID parameters. The fuzzy rule table is shown in Table 1 below: Table 1. Fuzzy PID Parameter Adjustment Rules θ range (mm) Kp Ki Kd 20-40 (Very Small) 2.8 0.05 0.12 40-60 (small) 2.2 0.08 0.10 60-90 (Medium) 1.6 0.12 0.08 90-120 (Large) 1.2 0.15 0.06 120-150 (Extremely large) 0.8 0.20 0.04 4.4.2 Algorithm Implementation Flow (1) Parameter initialization: When the system starts, it reads the diameter θ0 corresponding to the current material model and calls the initial PID parameters (Kp0, Ki0, Kd0) from the parameter database.

[0050] (2) Real-time detection: The material diameter θ is obtained through the vision system, and the actual pressure Pact is collected by the pressure sensor.

[0051] (3) Fuzzy inference: Based on the fuzzy subset to which θ belongs, calculate the current values ​​of Kp, Ki, and Kd using an interpolation algorithm: Kp = Kp_prev + ΔKp·(θ θ_prev) / Δθ (Formula 2) The symbols are defined as shown in Table 2 below:

[0052] Table 2 This formula is the core interpolation formula of the dynamic adaptive PID algorithm, used to solve the problem of abrupt changes in the Kp value when the material diameter θ is near the boundary of the fuzzy subset. Linear interpolation achieves a smooth transition of Kp between adjacent intervals, avoiding pressure fluctuations caused by the "step" parameter adjustments in traditional fuzzy control.

[0053] Example: When θ = 65mm (located in the "middle" range of 60-90mm): Kp_prev=2.2 (Kp value in the "small" interval), ΔKp=1.6-2.2=-0.6, θ_prev=60mm, Δθ=30mm; Kp=2.2 + (-0.6)·(65-60) / 30 = 2.2 - 0.6×0.167≈2.2-0.1=2.1; Result: By interpolation, Kp=2.1 was obtained when θ=65mm, which is between the "small" interval (2.2) and the "medium" interval (1.6), achieving a smooth transition.

[0054] (4) PID calculation: Calculate the control quantity u(t) according to Formula 1 and output it to the electric cylinder driver.

[0055] Data input: θ is measured in real time by the vision system through Hough circle transformation (accuracy ±0.05mm), and Kp_prev, ΔKp, θ_prev, and Δθ are retrieved from the parameter database (pre-stored in the "Parameters" table of the SQLite database).

[0056] Calculation execution: The formula runs in the C# algorithm module of the edge layer industrial PC, iterating once every 10ms. The calculation result is written to the output register of the PLC via Modbus TCP protocol to control the action of the servo motor of the rolling structure.

[0057] Control objective: By dynamically adjusting Kp, the rolling pressure deviation e(t) = Pset - Pact is stabilized within ±1N to meet the continuous production needs of materials with a diameter of 20-150mm.

[0058] 4.4.3 Validation of Algorithm Advantages The control effects of traditional PID and dynamic adaptive PID were compared through MATLAB / Simulink simulation: when the material diameter changed from 50mm to 100mm, the overshoot of traditional PID reached 25% and the settling time was 3.2s; while the overshoot of dynamic adaptive PID was ≤5% and the settling time was ≤0.8s, and the pressure control accuracy was improved to ±1N (the traditional PID was ±3N).

[0059] 4.5 Data Communication Between Hardware and Software Systems 4.5.1 Communication Architecture The system adopts a three-layer communication network: (1) Control layer: The PROFINET IRT (isochronous real-time) protocol is adopted to realize real-time communication between PLC and servo drives, encoders and other devices, with a cycle of ≤1ms and jitter of ≤1µs.

[0060] (2) Monitoring layer: The Modbus TCP / IP protocol is adopted to realize the data interaction between PLC, touch screen and edge computing PC. The communication rate is 100Mbps and the data update cycle is 100ms.

[0061] (3) Management: Access the Internet through a 4G industrial router, communicate with the cloud platform using the MQTT protocol, with a data upload cycle of 1 second and support for breakpoint resume function.

[0062] 4.5.2 Data Interaction Process (1) Parameter distribution: 1) The operator selects the material model via the touchscreen, triggering a parameter call command; 2) The PLC reads the corresponding process parameters (labeling speed, pressure setting, rolling distance, etc.) for this model from the local database. 3) The PLC sends speed commands to the servo drive and position commands to the electric cylinder via PROFINET; 4) After the driver executes the instruction, it returns a status word, and the PLC confirms that the parameter settings are complete.

[0063] (2) Data upload: 1) The edge computing PC receives image data from the vision system via gigabit Ethernet, processes it, and obtains the material position deviation; 2) The PC writes the deviation value into the PLC's input register via Modbus TCP; 3) After the PLC executes the position compensation algorithm, it stores the actual labeling position, pressure curve, and other data into the local buffer. 4) When the buffer contains 100 records, it is uploaded to the cloud platform via the MQTT protocol.

[0064] 4.5.3 Communication Security Mechanisms (1) Data encryption: The AES-256 algorithm is used to encrypt the uploaded data to ensure that the data is not tampered with during the transmission process; (2) Access control: Restrict device access through MAC address whitelist, and set password protection in PLC program (three-level access control). (3) Anomaly monitoring: Real-time monitoring of communication delay. When no response is received for 3 consecutive cycles, the backup communication link (RS485) is triggered.

[0065] 4.6 Intelligent labeling control methods (such as...) Figure 6 (As shown) 4.4.1 System Initialization Phase (Step S1) Implementation process: 1) When the main power is turned on, the PLC executes a self-test program to sequentially detect the status of each sensor and driver (by reading the device status word). 2) The touch screen displays the initialization interface, prompting the operator to select the material model (supports inputting material codes via barcode scanner). 3) The system loads the corresponding material's process parameter package, including: Labeling host parameters: label length (L=50-200mm adjustable), label dispensing speed (v=100-500mm / s); Transportation device parameters: conveyor belt speed (v=50-300mm / s), material spacing (d=100-500mm); Roller structure parameters: initial pressure (Pset=10-50N), roller speed ratio (i=1:1.2).

[0066] By adopting a parameterized configuration method, the process parameters of different materials are pre-stored as recipes and can be quickly recalled by recipe number, reducing changeover time. The PLC reads the recipe selection instructions from the touchscreen via the Modbus protocol and loads the corresponding parameter file from the SD card. Initialization time is reduced from 30 minutes of traditional manual adjustment to 2 minutes, and parameter setting accuracy reaches 100%, avoiding human input errors.

[0067] 4.4.2 Material conveying and positioning stage (step S2) Implementation process: 1) The operator places the material at the inlet of the conveyor device 11. After the photoelectric sensor E3Z-LS63 detects the material, it sends a trigger signal to the PLC. 2) The PLC starts the encoder counting. When the cumulative number of pulses reaches the set value (corresponding to the material reaching the vision detection position), the industrial camera is triggered to take a picture. 3) The vision system processes images: Image preprocessing: Median filtering (3×3 template) was used to remove noise, and material contours were extracted by adaptive threshold segmentation; Feature extraction: The Hu moment invariant is used to identify the material shape, and the diameter θ is calculated using the minimum circumcircle algorithm; Position calculation: Using the centerline of the conveyor belt as a reference, calculate the X-axis deviation (Δx) and Y-axis deviation (Δy) of the material center.

[0068] 4) The vision system sends θ, Δx, and Δy to the PLC via TCP / IP protocol, with a data transmission delay of ≤20ms.

[0069] This machine vision-based non-contact positioning method achieves precise measurement of material geometry and position through image feature extraction. Employing a sub-pixel-level edge detection algorithm, it improves pixel accuracy from 1 pixel to 0.1 pixels (corresponding to an actual accuracy of 0.02mm). Material positioning accuracy reaches ±0.05mm, an 80% improvement over traditional mechanical positioning methods; it can identify materials with diameters ranging from 20-150mm, improving compatibility by 60%.

[0070] 4.4.3 Dynamic parameter adjustment stage (step S3) Implementation process: 1) The PLC calls the dynamic adaptive PID algorithm module based on the material diameter θ transmitted from the vision system: When θ=65mm, Kp=1.5 is calculated according to Table 1 (between 1.6 for the medium diameter range of 60-90mm and 2.2 for the small diameter range of 40-60mm). Set the initial pressure Pset=25N (recall from the parameter library according to the material, 15-20N for plastic bottles and 25-35N for metal cans).

[0071] 2) The PLC sends position commands to the servo motor of the rolling structure to adjust the initial distance between the roller structure 32 and the extrusion structure 33: D = θ + 2×δ (Formula 4) Where δ is the pre-compression amount (set according to the elastic coefficient of the material, with a value of 0.5-2mm).

[0072] 3) The servo motor drives the screw rod 43 to rotate in absolute position mode, and the encoder provides real-time position feedback to form a position closed-loop control with an adjustment accuracy of ±0.01mm.

[0073] By establishing a correlation model between material diameter and pressure parameters, the pre-adjustment of the rolling mechanism is achieved, laying the foundation for subsequent closed-loop pressure control. A position-pressure composite control strategy is adopted, first using position control for coarse adjustment, and then using pressure control for fine adjustment. During production changeovers, the adjustment time for the rolling mechanism is reduced from 5 minutes of manual operation to 15 seconds, and the pressure pre-adjustment error is ≤2N, ensuring the quality of label adhesion.

[0074] 4.4.4 Labeling Execution Phase (Step S4) Implementation process: 1) When the material reaches the output of the labeling host 10, the PLC calculates the real-time position of the material based on the encoder data and synchronously controls the rotation speed of the labeling host through electronic gears: n_label = n_conveyor × D / L_label (Formula 5) Where n_label is the label roller speed (r / min), n_conveyor is the conveyor belt speed (m / min), and L_label is the label length (m).

[0075] 2) The electric cylinder of the limit device 12 extends to push the material to stick to the output end of the labeling host. The pressure sensor monitors the contact pressure in real time and maintains the pressure stable within the set value ±1N through PID control. 3) The labeling host rotates, causing the material to rotate. At the same time, the label dispensing mechanism delivers the label. The vision system monitors the label edge position in real time. When the starting edge of the label is detected, the labeling completion signal is triggered. 4) The electric cylinder retracts, and the conveyor belt transports the material to the rolling structure 3.

[0076] Electronic gear synchronization technology is used to match the speed of materials and labels, and pressure closed-loop control ensures uniform initial label bonding pressure. Visual edge detection uses a dynamic threshold method to adapt to the detection requirements of different color labels. The labeling position deviation is ≤±0.1mm, the label wrinkling rate is reduced from 3% in traditional equipment to below 0.5%, and the labeling speed reaches 120 pieces / minute (50% improvement over traditional equipment).

[0077] 4.4.5 Rolling quality control stage (step S5) Implementation process: 1) The material enters between the roller 325 and the extrusion plate 333, and the pressure sensor on the extrusion plate 333 collects the pressure signal Pact in real time; 2) The PLC compares Pact with the setpoint Pset, calculates the control quantity u(t) using a dynamic adaptive PID algorithm (Formula 1), and outputs it to the electric cylinder to adjust the position of the extrusion plate. When Pact < Pset, the electric cylinder extends (u(t) is positive), increasing the extrusion force; When Pact > Pset, the electric cylinder retracts (u(t) is negative), reducing the extrusion force.

[0078] 3) The encoder of the roller structure 32 monitors the material rotation speed in real time and maintains the linear speed synchronization between the roller and the material through speed closed-loop control to avoid the label from being stretched or wrinkled. 4) After the rolling is completed, the laser displacement sensor at the exit detects the flatness of the material surface and the data is uploaded to the edge computing PC for quality assessment.

[0079] Dynamic adjustment of label bonding pressure is achieved through closed-loop pressure control, combined with synchronous speed control to ensure uniform force on the label during rolling. Laser triangulation is used to detect label surface flatness with a resolution of 0.01mm. Label bonding strength is improved by 40% (verified by a 90° peel test), bubble formation rate is ≤0.3%, and surface flatness error is ≤0.05mm.

[0080] 4.4.6 Data Traceability and Optimization Phase (Step S6) Implementation process: 1) Edge computing PCs analyze the collected production data (labeling location, pressure curves, material parameters, etc.) and generate quality reports: Process capability index CPK calculation (target value ≥ 1.33); Pressure fluctuation trend analysis (using moving average filtering algorithm); Abnormal data labeling (based on the 3σ principle).

[0081] 2) Data is uploaded to the cloud platform via the MQTT protocol to achieve: Production reports are automatically generated (daily / weekly / monthly reports); Equipment health status assessment (predictive maintenance based on vibration and temperature data); Recommendations for optimizing process parameters (using machine learning algorithms to discover the optimal combination of parameters).

[0082] 3) Operators can view real-time production data and historical trends via touch screen or remote terminal, and support the backtracking of abnormal data.

[0083] Quality traceability and continuous optimization based on industrial big data analysis: By using machine learning algorithms (such as random forests) to establish a correlation model between process parameters and product quality, intelligent parameter recommendations can be achieved.

[0084] Production data traceability reached 100%, the response time for quality anomalies was shortened from 2 hours to 5 minutes, and the product qualification rate was increased to 99% through parameter optimization.

[0085] This embodiment upgrades a traditional labeling machine into an intelligent device with autonomous sensing, precise control, and data traceability by adding an intelligent control system and a dynamic adaptive PID algorithm. The hardware adopts a modular architecture of PLC + vision + servo, while the software features a layered design and algorithm integration, building a real-time data interaction network with the Industrial Internet via PROFINET. Practical applications show that this system can significantly improve labeling accuracy and efficiency, reduce labor costs and defect rates, and provide a feasible solution for the intelligent upgrading of the manufacturing industry.

[0086] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An automatic feeding control system for a labeling machine, comprising a support platform (1), a labeling host (10), a transport device (11), a rolling structure (3), and a limiting device (12), characterized in that, Also includes: Main controller; A visual positioning unit is installed above the output end of the labeling host (10) to acquire material images and calculate the material position and diameter; The sensing and detection unit includes an encoder disposed on the transport device (11) and a pressure sensor embedded in the rolling structure (3); The execution drive unit includes a servo adjustment mechanism that drives the rolling structure (3) to adjust the spacing; as well as The edge computing unit, which is communicatively connected to the main controller and the vision positioning unit, is configured to execute a visual image processing algorithm and a dynamic adaptive PID control algorithm. The dynamic adaptive PID control algorithm is used to adjust the pressure control parameters of the pressure sensor in real time according to the material diameter obtained by the vision positioning unit.

2. The automatic feeding control system for the labeling machine according to claim 1, characterized in that, The main controller is a programmable logic controller (PLC), which communicates with the execution drive unit via the PROFINET bus and with the vision positioning unit and the edge computing unit via Ethernet.

3. The automatic feeding control system for the labeling machine according to claim 1, characterized in that, The servo adjustment mechanism includes a servo motor and a planetary gear reducer. The servo motor is connected to the control thread rod (43) of the rolling structure (3) for electrically adjusting the distance between the roller structure (32) and the extrusion structure (33) in the rolling structure (3).

4. The automatic feeding control system for the labeling machine according to claim 1, characterized in that, The dynamic adaptive PID control algorithm is configured to calculate the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd by interpolation based on the preset fuzzy subset to which the material diameter belongs.

5. The automatic feeding control system for the labeling machine according to claim 1, characterized in that, The system also includes a human-machine interaction unit and a communication module. The human-machine interaction unit is used for recipe management, parameter setting and alarm display, and the communication module supports data interaction with the Manufacturing Execution System (MES) or cloud platform.

6. The automatic feeding control system for the labeling machine according to claim 5, characterized in that, The communication module adopts a layered communication architecture, including the PROFINET IRT protocol for real-time device control, the Modbus TCP / IP protocol for monitoring data interaction, and the MQTT protocol for cloud data upload.

7. The automatic feeding control system for the labeling machine according to claim 1, characterized in that, The limiting device (12) is a telescopic limiting component driven by an electric cylinder.

8. The automatic feeding control system for the labeling machine according to claim 1, characterized in that, The pressure sensor is embedded inside the extrusion plate (333) of the extrusion structure (33).

9. The automatic feeding control system for a labeling machine according to any one of claims 1-8, characterized in that, The system also includes a database module for storing production data, process parameters, and alarm records.

10. An automatic feeding control method for a labeling machine, applied to the automatic feeding control system for a labeling machine as described in any one of claims 1-9, characterized in that, Includes the following steps: S1: System initialization, loading the process parameters corresponding to the selected material model; S2: Material conveying and visual positioning, the material is conveyed through the conveying device (11) and the visual positioning unit is triggered to obtain the material diameter and position deviation; S3: Dynamic parameter pre-adjustment, based on the material diameter, the initial pressure control parameters are determined by the dynamic adaptive PID algorithm, and the servo adjustment mechanism is controlled to adjust the initial spacing of the rolling structure (3); S4: Labeling execution and synchronization control, controlling the labeling host (10) and the transport device (11) to synchronize their speeds, and controlling the limit device (12) to complete the labeling; S5: Rolling pressure closed-loop control, the material enters the rolling structure (3), and the extrusion pressure is adjusted in real time through the dynamic adaptive PID algorithm according to the feedback of the pressure sensor; S6: Data collection and traceability. Collect and analyze key data in the production process, and upload the data to the cloud platform for quality assessment and process optimization.

Images