PCB intelligent sorting process and system based on unmanned factory

Abstract

The invention discloses an intelligent PCB sorting process and system based on an unmanned factory, and the process comprises the steps: obtaining surface images and internal structure parameters of PCBs in real time through an image collection module and a scanning module on sorting equipment, and generating a standardized feature data set; the lightweight classification model performs classification and quality grade division on the PCBs in combination with a process grade judgment rule, and dynamically allocates sorting priorities based on MES system order demands to generate a task queue; the path planning algorithm calculates the optimal grabbing path of the mechanical arm and the AGV conveying path according to the priority queue, and a sorting instruction set is formed and issued to the execution unit through the industrial communication network; the mechanical arm completes PCB grabbing and placing operation according to the instruction, and meanwhile, the visual verification module detects a sorting result in real time and outputs a sorting state report and quality data; according to the process, high efficiency, accuracy and unmanned operation of PCB sorting are realized through automatic data acquisition, intelligent classification decision and a closed-loop optimization mechanism.

Description

Technical Field

[0001] The present invention relates to the field of PCB sorting, and in particular to a PCB intelligent sorting process and system based on an unmanned factory. Background Art

[0002] As core components of electronic devices, printed circuit boards (PCBs) are rapidly evolving in manufacturing processes toward high density and miniaturization. With the surge in demand for PCBs in sectors such as consumer electronics and automotive electronics, traditional production models face significant challenges. In particular, during sorting, PCBs must be precisely classified based on model, process parameters (such as plating thickness and pad layout), and quality grade to meet subsequent packaging, warehousing, and order delivery requirements. Currently, under the trend of intelligent manufacturing, unmanned factories (dark factories) require full unmanned operation throughout the entire process. However, PCB sorting still relies on manual visual inspection or semi-automated equipment, which presents challenges such as low efficiency, high error rates, and insufficient processing capacity for mixed-spec lines. These factors severely restrict companies' ability to achieve large-scale flexible production.

[0003] Existing PCB sorting technology primarily relies on mechanical positioning combined with basic visual recognition. For example, after capturing PCB surface images with a fixed industrial camera, simple classification is performed based on a threshold segmentation algorithm. This technology can only recognize a single model or a limited number of features (such as size and color) and requires a fixed, pre-set grasping path. This leads to the following core flaws: insufficient sorting accuracy and compatibility. For example, in scenarios with mixed PCB types (such as rigid boards, flexible boards, and high-frequency boards), traditional systems lack the ability to detect internal structural parameters (such as copper layer thickness and pore size distribution), making it impossible to distinguish between boards with similar process complexity. This results in a sorting error rate exceeding 5%, requiring manual re-inspection and intervention. Furthermore, fixed path planning algorithms struggle to adapt to dynamic order demands, resulting in sorting efficiency stagnating below an average of 5,000 pieces per day for a long time, failing to meet the high throughput (≥20,000 pieces / day) and zero-defect sorting requirements of smart manufacturing.

[0004] In view of this, it is necessary to improve the PCB sorting technology in the existing technology to solve the technical problems of high sorting error rate and insufficient processing capacity for mixed lines of multi-specification boards. Summary of the Invention

[0005] The purpose of the present invention is to provide a PCB intelligent sorting process and system based on an unmanned factory to solve the above technical problems.

[0006] To achieve this object, the present invention adopts the following technical solutions: A PCB intelligent sorting process based on an unmanned factory includes the following steps: Through the image acquisition module and scanning module deployed on the sorting equipment, the surface image data and internal structure parameters of the PCB board are acquired in real time, and the model identification, pad distribution characteristics and process parameters of the PCB board are extracted to generate a standardized feature data set; The feature data set is input into the lightweight classification model, and the PCB panels are classified by model and quality grade according to the preset process grade judgment rules. At the same time, the sorting priority is dynamically allocated based on the order requirements of the MES system to generate a sorting task priority queue; According to the sorting task priority queue, the optimal grasping path of the robot arm and the AGV delivery route are calculated through the path planning algorithm, and a sorting instruction set is generated and sent to the execution unit through the industrial communication network; Based on the sorting instruction set, the robotic arm is controlled to complete the grabbing and placement operations of the PCB panels, and the sorting results are simultaneously inspected in real time through the visual verification module. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality verification data are output.

[0007] Optionally, the image acquisition module and scanning module deployed on the sorting equipment can acquire the surface image data and internal structure parameters of the PCB board in real time, extract the model identification, pad distribution characteristics and process parameters of the PCB board, and generate a standardized feature data set, specifically including: The industrial cameras and scanning modules deployed on the sorting equipment capture multi-angle images and scan the internal structures of the PCBs on the conveyor belt, synchronously acquiring surface image data and internal structure parameters, and generating raw data packets based on a timestamp alignment mechanism; the internal structure parameters include internal copper layer thickness distribution data and aperture characteristic parameters; Preprocessing the raw data packet includes: performing adaptive denoising and edge enhancement operations on the surface image data to eliminate image distortion caused by uneven lighting and mechanical vibration; applying a dynamic filtering algorithm to the internal scan data to remove environmental interference signals, and matching benchmark values ​​with a preset PCB process parameter library to generate denoised standard data units; Feature extraction is performed on the standard data unit, specifically including: identifying model identification characters, positioning pad coordinates and spacing distribution from the surface image data, extracting copper layer thickness mean, pore size distribution density and microcrack defect parameters from the internal scanning data, and forming a multidimensional feature vector.

[0008] Optionally, feature extraction is performed on the standard data unit, followed by: Input the multidimensional feature vector into the dynamic template generation module, convert the heterogeneous feature parameters into structured data of unified dimensions according to preset standardization rules, and add unique identifiers and timestamp tags; The data verification engine performs integrity detection and logical verification on the standardized structured data, eliminates outliers, generates a standardized feature data set containing complete metadata, and stores it in the database of the edge computing node.

[0009] Optionally, the feature data set is input into a lightweight classification model, and the PCB panels are classified by model and quality grade in combination with a preset process grade judgment rule. At the same time, sorting priorities are dynamically allocated based on the order requirements of the MES system, and a sorting task priority queue is generated. Specifically, the following steps are included: Inputting the feature dataset into a lightweight classification model, performing semantic parsing on the model identification characters of the PCB board using a pre-trained PCB model feature matching network, generating an initial model classification label set, and associating the corresponding process parameter threshold ranges; Based on a preset process grade determination rule library, each PCB board in the initial model classification label set is classified into a quality grade, including: calculating a quality score based on pad distribution uniformity, copper layer thickness deviation value, and microcrack defect density, and mapping it to a preset process grade interval to generate an enhanced classification result with a quality grade identification; Acquire order demand data from the MES system in real time, extract order urgency coefficient, storage location status, and customer customization parameters, build a multi-objective optimization function, dynamically calculate the sorting priority weight of each PCB board, and generate a priority weight matrix; The enhanced classification results are associated and matched with the priority weight matrix, and a sorting task priority queue is generated through a dynamic weighted sorting algorithm, and is aggregated into a task instruction package and transmitted to the path planning module.

[0010] Optionally, according to the sorting task priority queue, the optimal grasping path of the robot arm and the AGV delivery route are calculated by a path planning algorithm, a sorting instruction set is generated, and the set is sent to the execution unit through the industrial communication network, specifically including the following steps: Based on the sorting task priority queue, the task parsing module extracts the target storage location, process level constraints, and time window requirements of the PCB panels, and generates a sorting task parameter matrix containing time and space constraints; A hybrid path planning algorithm is used, combining the robot arm's kinematic model with the AGV's dynamic obstacle avoidance rules. With the optimization goal of minimizing the total sorting time and energy consumption, the robot arm's grasping trajectory sequence and the AGV's conveying path map are calculated to generate an initial path planning solution. The initial path planning scheme is input into the instruction compilation engine, and the robot arm joint angle, AGV speed curve and collaborative timing are encoded based on the preset industrial protocol standard to generate an executable sorting instruction set and embed the equipment status self-check code; The sorting instruction set is transmitted to the execution unit through the industrial communication network, the instruction confirmation signal of the robot arm and AGV is received in real time, and the instruction timing offset is dynamically adjusted according to the network delay to complete the issuance of the sorting instruction.

[0011] Optionally, based on the sorting instruction set, the robot arm is controlled to complete the grabbing and placing operations of the PCB panels, and the sorting results are simultaneously inspected in real time by the visual verification module. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality verification data are output. Specifically, the following steps are included: Parse the robotic arm grabbing coordinate sequence and AGV target placement coordinates in the sorting instruction set, drive the robotic arm end effector to the specified position through the motion control module, and complete non-destructive grabbing of PCB panels based on the composite grabbing mechanism of electromagnetic suction cup and flexible gripper, and generate a grabbing action log; The robot arm joint torque data and the pressure feedback signal of the end effector are collected in real time, and the grasping force is dynamically adjusted based on the preset mechanical tolerance threshold. If the grasping offset or abnormal pressure fluctuation is detected, the grasping retry mechanism is triggered and the grasping action log is updated.

[0012] Optionally, if abnormal fluctuations in the grabbing offset or pressure are detected, a grabbing retry mechanism is triggered and the grabbing action log is updated, and the following further includes: The AGV collaborative scheduling module delivers the grabbed PCB panels to the target storage location, and simultaneously activates the visual verification module deployed on the AGV carrier to perform real-time scanning of the surface integrity and placement posture of the PCB panels, generating an original verification image data stream. Performing anomaly analysis on the original verification image data stream based on a dynamic adaptive detection algorithm, including: identifying scratches, warpage, and pad oxidation defects using a defect detection model optimized by transfer learning, comparing process parameter thresholds in a standardized feature data set, and outputting anomaly determination results and defect type labels; If the abnormal judgment result is marked as a defective panel, a rework instruction is sent to the MES system through the rework signal generation module, triggering the AGV to automatically return to the sorting starting station.

[0013] Optionally, the synchronous visual inspection module performs real-time quality inspection on the sorting results. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality inspection data are output. The following also includes: The sorting completion status report and quality verification data are sent back to the MES system. The sorting efficiency and error types are statistically analyzed through the data analysis module, and the parameter library of the lightweight classification model and the optimization weight of the path planning algorithm are dynamically updated to form a closed-loop iterative mechanism.

[0014] The present invention also provides a PCB intelligent sorting system based on an unmanned factory, which implements sorting by adopting the above-mentioned PCB intelligent sorting process based on an unmanned factory. The PCB intelligent sorting system specifically includes: Sorting equipment, wherein the sorting equipment is provided with an industrial camera and a scanning module for obtaining surface image data and internal structural parameters of PCB boards in real time; A robotic arm equipped with an end effector comprising an electromagnetic suction cup and a flexible gripper, for supporting non-destructive grasping of PCBs of varying thicknesses; wherein the robotic arm is integrated with a joint torque sensor and an end pressure sensor; AGV conveying components, equipped with visual verification modules and positioning devices, enable directional conveying of panels; Industrial communication network for transmitting sorting instruction sets; The control center integrates lightweight classification models, dynamic path planning algorithms, and multi-objective optimization engines to drive the coordinated operation of the sorting system; The closed-loop management module realizes the feedback of sorting data and iterative update of process parameters through the MES system interface.

[0015] Compared with the existing technology, the present invention has the following beneficial effects: first, the image acquisition module and scanning module on the sorting equipment acquire the surface image and internal structure parameters of the PCB board in real time, extract the model identification, pad distribution characteristics and process parameters, and generate a standardized feature data set; then, the lightweight classification model combines the process grade judgment rules to classify and divide the PCB boards into quality grades, and dynamically allocates sorting priorities based on the order requirements of the MES system to generate a task queue; the path planning algorithm calculates the optimal grasping path of the robot arm and the AGV conveying route according to the priority queue, forms a sorting instruction set and sends it to the execution unit through the industrial communication network; the robot arm completes the PCB grasping and placement operations according to the instructions, and at the same time, the visual verification module detects the sorting results in real time, abnormal boards trigger rework signals, and outputs a sorting status report and quality data; this process achieves a deep integration of high efficiency, accuracy and unmanned adaptation of PCB sorting through automated data acquisition, intelligent classification decision-making and closed-loop optimization mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] The structures, proportions, sizes, etc. depicted in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not intended to limit the conditions under which the present invention can be implemented and therefore have no substantive technical significance. Any structural modifications, changes in proportional relationships, or adjustments in size should still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and objectives that can be achieved by the present invention.

[0018] Figure 1 This is one of the flow charts of the PCB intelligent sorting process based on the unmanned factory of the first embodiment; Figure 2 This is the second flow chart of the PCB intelligent sorting process based on the unmanned factory of the first embodiment; Figure 3 This is a schematic diagram of the layout of the PCB intelligent sorting system of the second embodiment. DETAILED DESCRIPTION

[0019] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] In the description of the present invention, it should be understood that the terms "upper," "lower," "top," "bottom," "inner," "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate the description of the present invention and simplify the description. They are not intended to indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. It should be noted that when a component is considered to be "connected" to another component, it may be directly connected to the other component or there may be a centrally located component.

[0021] The technical solution of the present invention will be further described below with reference to the accompanying drawings and through specific implementation methods.

[0022] Example 1: Combine Figure 1 and Figure 2 As shown, an embodiment of the present invention provides a PCB intelligent sorting process based on an unmanned factory, comprising the following steps: S1, through the image acquisition module and scanning module deployed on the sorting equipment, acquires the surface image data and internal structure parameters of the PCB board in real time, and extracts the model identification, pad distribution characteristics and process parameters of the PCB board to generate a standardized feature data set; By deploying a multispectral industrial camera array and infrared scanning modules, multimodal data acquisition of PCB surface images and internal structural parameters is achieved. The image acquisition module captures surface features (such as pad layout and character markings), while the infrared scanning module penetrates the surface to obtain internal parameters such as copper layer thickness and pore size distribution. A timestamp alignment mechanism is used to ensure data consistency in time and space.

[0023] S2 inputs the feature data set into the lightweight classification model. Combined with the preset process grade judgment rules, the PCB panels are classified by model and quality grade. At the same time, the sorting priority is dynamically assigned based on the order requirements of the MES system, and a sorting task priority queue is generated. Using lightweight classification models (such as the improved YOLOv7-PCB network), the feature dataset is classified by model and quality grade. Quality grade is determined by combining a pre-set process rule library (such as pad uniformity thresholds and copper layer thickness tolerances). Furthermore, the MES system collects real-time information such as order urgency and warehouse location status, and constructs a multi-objective optimization function to dynamically assign sorting priorities.

[0024] S3, based on the sorting task priority queue, calculates the optimal grasping path of the robot arm and the AGV delivery route through the path planning algorithm, generates a sorting instruction set, and sends it to the execution unit through the industrial communication network; Hybrid path planning algorithms (such as a fusion of RRT and Dijkstra algorithms) are used to calculate the robotic arm's grasping trajectory and the AGV's transport path. This minimizes the total sorting time and energy consumption, generating a time- and space-coordinated sorting instruction set. The industrial communication network (TSN protocol) ensures low-latency transmission of instructions, and CRC checksums safeguard data integrity.

[0025] S4, based on the sorting instruction set, controls the robotic arm to complete the grabbing and placement operations of PCB panels, and simultaneously performs real-time quality inspection on the sorting results through the visual verification module. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality verification data are output.

[0026] The robotic arm performs grasping operations based on sorting instructions. The end effector utilizes a composite mechanism of electromagnetic suction cups and flexible grippers to adapt to PCBs of varying thicknesses. Simultaneously, the multispectral vision module on the AGV is activated, using a defect detection model optimized through transfer learning to identify anomalies such as scratches and warpage in real time. This triggers a rework signal and associates a unique identifier to trace the defect source.

[0027] S5, sends the sorting completion status report and quality verification data back to the MES system, uses the data analysis module to count the sorting efficiency and error types, and dynamically updates the parameter library of the lightweight classification model and the optimization weight of the path planning algorithm to form a closed-loop iterative mechanism.

[0028] Sorting completion status reports and quality verification data are transmitted back to the MES system. The data analysis module calculates sorting efficiency and error type distribution, dynamically updating the classification model parameter library (such as defect feature weights) and path planning algorithm optimization rules (such as obstacle avoidance priority adjustment). This includes a data-driven parameter iteration mechanism (optimizing the model based on historical sorting data) and edge-cloud collaborative computing (achieving low-latency feedback). This creates a closed process loop, continuously improving the system's adaptability and addressing the inability of traditional sorting processes to dynamically optimize.

[0029] The working principle of the present invention is as follows: first, the image acquisition module and scanning module on the sorting equipment acquire the surface image and internal structure parameters of the PCB board in real time, extract the model identification, pad distribution characteristics and process parameters, and generate a standardized feature data set; then, the lightweight classification model combines the process grade judgment rules to classify and divide the PCB boards into quality grades, and dynamically allocates sorting priorities based on the order requirements of the MES system to generate a task queue; the path planning algorithm calculates the optimal grasping path of the robot arm and the AGV conveying route according to the priority queue, forms a sorting instruction set, and sends it to the execution unit through the industrial communication network; the robot arm completes the PCB grasping and placement operations according to the instructions, while the visual verification module detects the sorting results in real time, and abnormal panels trigger rework signals, and outputs a sorting status report and quality data; this process achieves a deep integration of high efficiency, accuracy and unmanned adaptation of PCB sorting through automated data acquisition, intelligent classification decision-making and closed-loop optimization mechanism.

[0030] In this embodiment, it is specifically explained that step S1 specifically includes: S11 uses industrial cameras and scanning modules deployed on the sorting equipment to capture multi-angle images and scan the internal structures of PCBs on the conveyor belt, synchronously acquiring surface image data and internal structure parameters, and generating raw data packets based on a timestamp alignment mechanism; internal structure parameters include internal copper layer thickness distribution data and aperture characteristic parameters; The industrial cameras and scanning modules on the sorting equipment capture multi-angle images of PCBs on the conveyor belt (including surface texture, character markings, etc.) and scan internal structures (such as copper layer thickness and pore size distribution). The industrial cameras use multispectral imaging technology to capture data in the visible and near-infrared bands, while the infrared scanning module uses penetrating detection principles to obtain internal parameters. A timestamp alignment mechanism (ensuring spatiotemporal synchronization between surface images and internal scan data) generates a standardized data package containing the original surface image and internal parameters, providing a complete data source for subsequent processing and addressing the insufficient feature dimensionality associated with traditional single-source data collection.

[0031] S12, preprocessing the original data packet, specifically including: performing adaptive denoising and edge enhancement operations on the surface image data to eliminate image distortion caused by uneven lighting and mechanical vibration; using a dynamic filtering algorithm to remove environmental interference signals from the internal scan data; and matching the reference values ​​with a preset PCB process parameter library to generate denoised standard data units; The original data packets are processed modally: the surface image uses an adaptive denoising algorithm (such as wavelet transform) to eliminate blurring caused by uneven lighting or mechanical vibration, and the pad outline is enhanced through edge enhancement technology; the internal scanning data uses a dynamic filtering algorithm (such as Kalman filtering) to remove environmental electromagnetic interference signals, and at the same time, the reference value is matched based on the preset PCB process parameter library (such as the copper layer thickness tolerance of ±0.05mm) to correct data offset.

[0032] S13, based on the improved YOLOv7-PCB model, features are extracted from standard data units. Specifically, this includes: identifying model identification characters, positioning pad coordinates and spacing distribution from surface image data, extracting the mean copper layer thickness, pore size distribution density and microcrack defect parameters from internal scanning data, and forming a multi-dimensional feature vector.

[0033] Based on the improved YOLOv7-PCB model (which adds a PCB process feature layer to the original YOLOv7), feature extraction is performed on standard data units. The surface image is used by the OCR module to identify model identification characters and locate the pad coordinates and spacing distribution. Internal scanning data is used through regression analysis to extract the mean copper layer thickness and pore size distribution density, and detect microcrack defects.

[0034] S14, inputting the multi-dimensional feature vector into the dynamic template generation module, converting the heterogeneous feature parameters into structured data of uniform dimensions according to preset standardization rules (including data format, unit unification, and normalization processing), and adding a unique identifier and timestamp tag; The multidimensional feature vector is fed into the dynamic template generation module, which then standardizes the heterogeneous data according to pre-set rules (e.g., normalization). For example, the copper layer thickness is standardized to microns, the pad spacing is normalized to a percentage, and unique identifiers (e.g., UUIDs) and timestamps are added.

[0035] S15, the data verification engine performs integrity detection and logic verification on the standardized structured data, eliminates outliers, generates a standardized feature data set containing complete metadata, and stores it in the database of the edge computing node.

[0036] The data verification engine performs integrity checks (such as missing fields) and logic verification (such as whether the copper layer thickness exceeds the process library threshold) on the standardized structured data, eliminating outliers and triggering a re-sampling process. Metadata tags (such as sorting batch numbers) are added to the verified data and stored in the distributed database of the edge computing node.

[0037] In this embodiment, it is specifically explained that step S2 specifically includes the following steps: S21, inputting the feature dataset into the lightweight classification model, performing semantic parsing on the model identification characters of the PCB board through a pre-trained PCB model feature matching network, generating an initial model classification label set, and associating the corresponding process parameter threshold range; A pre-trained PCB model feature matching network performs semantic parsing (e.g., optical character recognition) on the model identification characters in the feature dataset to generate an initial set of model classification labels. The classification results are then dynamically associated with preset process parameter thresholds (e.g., copper layer thickness tolerance, aperture tolerance). This includes a pre-trained feature matching network (based on a Transformer architecture to optimize character recognition) and a dynamic process parameter binding mechanism (associating model numbers with process library parameters).

[0038] S22, based on a preset process grade determination rule library, classifying each PCB board in the initial model classification label set into a quality grade, including: calculating a quality score based on pad distribution uniformity, copper layer thickness deviation value, and microcrack defect density, and mapping it to a preset process grade interval, thereby generating an enhanced classification result with a quality grade identifier; Based on a process grade determination rule library, each PCB component is scored multi-dimensionally for pad distribution uniformity (calculated via coefficient of variation), copper layer thickness deviation (compared to process library benchmark values), and microcrack defect density (number of defects per unit area), and then mapped to preset grade intervals (e.g., A / B / C). This system utilizes a multi-dimensional scoring algorithm (weighted fusion of different parameters) and grade mapping rules (preset threshold intervals) to generate enhanced classification results with quality grade identification, providing quality dimension input for priority allocation and overcoming the limitations of traditional single-threshold determination.

[0039] S23, real-time acquisition of order demand data from the MES system, extraction of order urgency coefficient, storage location status, and customer customization parameters, construction of a multi-objective optimization function, dynamic calculation of the sorting priority weight of each PCB board, and generation of a priority weight matrix; By acquiring real-time information from the MES system regarding order urgency (e.g., remaining delivery time), warehouse location status (e.g., target location capacity), and customized parameters (e.g., special process requirements), the system constructs a multi-objective optimization function (including minimizing delivery delays and maximizing location utilization) and dynamically calculates sorting priority weights. This includes a multi-objective optimization algorithm (e.g., the NSGA-II genetic algorithm) and a dynamic weight adjustment mechanism (which responds to order changes in real time). This generates a priority weight matrix, enabling flexible scheduling of sorting tasks and addressing the rigid resource allocation issues inherent in traditional static rules.

[0040] S24, associate and match the enhanced classification results with the priority weight matrix, generate a sorting task priority queue through a dynamic weighted sorting algorithm, and aggregate them into a task instruction package to be transmitted to the path planning module.

[0041] The enhanced classification results are matched against the priority weight matrix in spatiotemporal order (e.g., aligning sorting station numbers with time windows). A dynamic weighted sorting algorithm (based on Pareto front optimization) is then used to generate a sorting task priority queue. This queue is then packaged into a task instruction package (including target coordinates and process level constraints). This includes a dynamic weighted sorting algorithm (balancing quality and timeliness) and an instruction encapsulation protocol (e.g., JSON structured data). The output of this task instruction package directly drives the path planning module, ensuring accurate delivery of sorting instructions.

[0042] In this embodiment, it is specifically explained that step S3 specifically includes the following steps: S31, based on the sorting task priority queue, extract the target storage location, process level constraints and time window requirements of the PCB panels through the task analysis module, and generate a sorting task parameter matrix including time and space constraints; In this step, the task parsing module extracts the target storage location (e.g., storage location coordinates), process level constraints (e.g., anti-static level requirements), and time window requirements (e.g., urgent orders must be sorted within a specified time period) of the PCB panels from the sorting task priority queue. This step then constructs a sorting task parameter matrix that incorporates temporal and spatial constraints. This multidimensional task parameter matrix provides precise input for path planning, improving its adaptability and efficiency.

[0043] S32 uses a hybrid path planning algorithm, combining the robot arm's kinematic model with the AGV's dynamic obstacle avoidance rules, with the optimization goal of minimizing the total sorting time and energy consumption. It calculates the robot arm's grasping trajectory sequence and the AGV's conveying path diagram to generate an initial path planning solution. A hybrid path planning algorithm is used to calculate the robotic arm grasping trajectory sequence and the AGV conveying path diagram. The robotic arm kinematic model (joint angular velocity limit) and the AGV dynamic obstacle avoidance rules (such as real-time detection by lidar) are combined to minimize the total sorting time and energy consumption as the optimization goal, achieve spatiotemporal coordination of equipment movements, avoid the risk of equipment collision caused by fixed paths, and reduce energy costs.

[0044] S33: Input the initial path planning solution into the instruction compilation engine, encode the robot arm joint angle, AGV speed curve and coordination timing based on the preset industrial protocol standard, generate an executable sorting instruction set, and embed the equipment status self-check code; The initial path planning solution is input into the instruction compilation engine, and the robot arm joint angle (unit: radian), AGV speed curve (such as Bezier curve control points) and collaborative timing are encoded into an executable instruction set based on industrial protocol standards (such as OPC UA), and the device status self-check code (such as CRC check code) is embedded.

[0045] S34 transmits the sorting instruction set to the execution unit through the industrial communication network, receives the instruction confirmation signal from the robot arm and AGV in real time, and dynamically adjusts the instruction timing offset according to the network delay to complete the issuance of the sorting instruction.

[0046] The sorting instruction set is broadcast to the execution unit via the industrial communication network, and the instruction confirmation signal from the robot arm and AGV is received in real time. If network delay is detected (e.g., >5ms), the instruction timing offset is dynamically adjusted to ensure precise synchronization of multi-device actions.

[0047] In this embodiment, it is specifically explained that step S4 specifically includes the following steps: S41: Parse the robotic arm's grabbing coordinate sequence and the AGV's target placement coordinates in the sorting instruction set, drive the robotic arm's end effector to the specified position through the motion control module, and complete non-destructive grabbing of the PCB using a composite grabbing mechanism consisting of an electromagnetic chuck and a flexible gripper, generating a grabbing action log. By parsing the robotic arm's grasping coordinate sequence and the AGV's target placement coordinates within the sorting instruction set, the motion control module drives the robotic arm's end effector to the target location. A composite grasping mechanism, comprised of an electromagnetic chuck and flexible gripper, performs non-destructive grasping, generating a grasping action log containing timestamps, grasping coordinates, and force feedback data. This achieves precise positioning and flexible grasping, resolving the issue of panel damage caused by traditional rigid grippers and providing a reliable grasping foundation for subsequent processes.

[0048] S42 collects the joint torque data of the robotic arm and the pressure feedback signal of the end effector in real time, and dynamically adjusts the grasping force based on the preset mechanical tolerance threshold. If the grasping offset or abnormal pressure fluctuation is detected, the grasping retry mechanism is triggered and the grasping action log is updated.

[0049] Real-time acquisition of arm joint torque data (reflecting load status) and end-effector pressure signals (monitoring adsorption stability) dynamically adjusts grasping force based on preset mechanical tolerance thresholds (e.g., pressure fluctuation range of ±5N). If grasp offset or pressure anomalies are detected, a grasp retry mechanism is triggered and a log is updated.

[0050] S43: The grabbed PCB is delivered to the target storage location through the AGV collaborative scheduling module. The visual verification module deployed on the AGV is simultaneously activated to perform real-time scanning of the surface integrity and placement posture of the PCB, generating an original verification image data stream. The AGV collaborative scheduling module delivers PCBs according to the coordinates of the target storage location and simultaneously activates the multispectral vision verification module (visible light + near-infrared bands) on the AGV carrier to scan the PCB surface integrity (scratches, stains) and placement posture in real time, generating a high-frame-rate verification image data stream. This enables simultaneous delivery and quality inspection, eliminating rework delays caused by post-sorting quality inspection.

[0051] S44, performing anomaly analysis on the original verification image data stream based on a dynamic adaptive detection algorithm, including: identifying scratches, warpage, and pad oxidation defects through a defect detection model optimized by transfer learning, comparing process parameter thresholds in a standardized feature data set, and outputting anomaly determination results and defect type labels; The defect detection model optimized by transfer learning (pre-training data set contains multiple PCB defect samples) analyzes the verification image data stream to identify scratches, warpage and pad oxidation defects, and compares the process parameter thresholds in the standardized feature data set (such as pad oxidation area ≤ 0.1mm 2 Output abnormality determination results and defect type labels to provide a quantitative basis for rework decisions and improve defect detection rates.

[0052] S45, if the abnormal judgment result is marked as a defective panel, a rework instruction is sent to the MES system through the rework signal generation module, triggering the AGV to automatically return to the sorting starting station.

[0053] If the abnormality is determined to be a defective panel, the rework signal generation module sends a rework instruction (including a unique identifier, defect type, and grab log) to the MES system, triggering the AGV to automatically plan a return path to the sorting starting station. The rework path is dynamically planned to avoid the current sorting path. This achieves closed-loop processing of defective panels, avoids efficiency losses caused by manual intervention, and ensures the integrity of the unmanned sorting process.

[0054] Example 2: Combine Figure 3 As shown, the present invention also provides a PCB intelligent sorting system based on an unmanned factory, which adopts the PCB intelligent sorting process based on an unmanned factory as in Example 1 to achieve sorting. The PCB intelligent sorting system specifically includes: The sorting device 10 is provided with an industrial camera 11 and a scanning module 12 for acquiring surface image data and internal structural parameters of PCB boards in real time.

[0055] The robotic arm 20 is equipped with an end effector consisting of an electromagnetic suction cup and a flexible gripper 21, and is used to support non-destructive grasping of PCBs of various thicknesses. The robotic arm is integrated with a joint torque sensor and an end pressure sensor.

[0056] The AGV conveying component 30 is equipped with a visual verification module and a positioning device 31 to achieve directional conveying of panels; the industrial communication network 40 is used to transmit the sorting instruction set.

[0057] The control center 50 integrates a lightweight classification model, a dynamic path planning algorithm, and a multi-objective optimization engine to drive the coordinated operation of the sorting system.

[0058] The closed-loop management module realizes the feedback of sorting data and iterative update of process parameters through the MES system interface.

[0059] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A PCB intelligent sorting process based on an unmanned factory, characterized in that: The following steps are involved: Through the image acquisition module and scanning module deployed on the sorting equipment, the surface image data and internal structure parameters of the PCB board are acquired in real time, and the model identification, pad distribution characteristics and process parameters of the PCB board are extracted to generate a standardized feature data set; The feature data set is input into the lightweight classification model, and the PCB panels are classified by model and quality grade according to the preset process grade judgment rules. At the same time, the sorting priority is dynamically allocated based on the order requirements of the MES system to generate a sorting task priority queue; According to the sorting task priority queue, the optimal grasping path of the robot arm and the AGV delivery route are calculated through the path planning algorithm, a sorting instruction set is generated, and it is sent to the execution unit through the industrial communication network; Based on the sorting instruction set, the robotic arm is controlled to complete the grabbing and placement operations of the PCB panels, and the sorting results are simultaneously inspected in real time through the visual verification module. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality verification data are output.

2. The PCB intelligent sorting process based on an unmanned factory according to claim 1 is characterized in that: The image acquisition module and scanning module deployed on the sorting equipment can obtain the surface image data and internal structure parameters of the PCB board in real time, extract the model identification, pad distribution characteristics and process parameters of the PCB board, and generate a standardized feature data set, which specifically includes: The industrial cameras and scanning modules deployed on the sorting equipment capture multi-angle images and scan the internal structures of the PCBs on the conveyor belt, synchronously acquiring surface image data and internal structure parameters, and generating raw data packets based on a timestamp alignment mechanism; the internal structure parameters include internal copper layer thickness distribution data and aperture characteristic parameters; Preprocessing the raw data packet includes: performing adaptive denoising and edge enhancement operations on the surface image data to eliminate image distortion caused by uneven illumination and mechanical vibration; applying a dynamic filtering algorithm to the internal scan data to remove environmental interference signals, and matching benchmark values ​​with a preset PCB process parameter library to generate denoised standard data units; The standard data unit is subjected to feature extraction, specifically comprising: identifying model identification characters, positioning pad coordinates and spacing distribution from the surface image data, extracting copper layer thickness mean, pore size distribution density and microcrack defect parameters from the internal scanning data, and forming a multidimensional feature vector.

3. The PCB intelligent sorting process based on an unmanned factory according to claim 2 is characterized in that: Extracting features from the standard data unit, and then further comprising: Input the multidimensional feature vector into the dynamic template generation module, convert the heterogeneous feature parameters into structured data of unified dimensions according to preset standardization rules, and add unique identifiers and timestamp tags; The data verification engine performs integrity detection and logical verification on the standardized structured data, eliminates outliers, generates a standardized feature data set containing complete metadata, and stores it in the database of the edge computing node.

4. The PCB intelligent sorting process based on an unmanned factory according to claim 3 is characterized in that: The feature data set is input into the lightweight classification model. Combined with the preset process grade judgment rules, the PCB panels are classified by model and quality grade. At the same time, the sorting priority is dynamically allocated based on the order requirements of the MES system, and the sorting task priority queue is generated. The specific steps include: Inputting the feature dataset into a lightweight classification model, performing semantic parsing on the model identification characters of the PCB board using a pre-trained PCB model feature matching network, generating an initial model classification label set, and associating the corresponding process parameter threshold ranges; Based on a preset process grade determination rule library, each PCB board in the initial model classification label set is classified into a quality grade, including: calculating a quality score based on pad distribution uniformity, copper layer thickness deviation value, and microcrack defect density, and mapping it to a preset process grade interval to generate an enhanced classification result with a quality grade identification; Acquire order demand data from the MES system in real time, extract order urgency coefficient, storage location status, and customer customization parameters, build a multi-objective optimization function, dynamically calculate the sorting priority weight of each PCB board, and generate a priority weight matrix; The enhanced classification results are associated and matched with the priority weight matrix, and a sorting task priority queue is generated through a dynamic weighted sorting algorithm, and is aggregated into a task instruction package and transmitted to the path planning module.

5. The PCB intelligent sorting process based on an unmanned factory according to claim 1 is characterized in that: According to the sorting task priority queue, the optimal grasping path of the robot arm and the AGV delivery route are calculated through the path planning algorithm, and a sorting instruction set is generated and sent to the execution unit through the industrial communication network. Specifically, the following steps are included: Based on the sorting task priority queue, the task parsing module extracts the target storage location, process level constraints, and time window requirements of the PCB panels, and generates a sorting task parameter matrix containing time and space constraints; A hybrid path planning algorithm is used, combining the robot arm's kinematic model with the AGV's dynamic obstacle avoidance rules. With the optimization goal of minimizing the total sorting time and energy consumption, the robot arm's grasping trajectory sequence and the AGV's conveying path map are calculated to generate an initial path planning solution. The initial path planning scheme is input into the instruction compilation engine, and the robot arm joint angle, AGV speed curve and collaborative timing are encoded based on the preset industrial protocol standard to generate an executable sorting instruction set and embed the equipment status self-check code; The sorting instruction set is transmitted to the execution unit through the industrial communication network, the instruction confirmation signal of the robot arm and AGV is received in real time, and the instruction timing offset is dynamically adjusted according to the network delay to complete the issuance of the sorting instruction.

6. The PCB intelligent sorting process based on an unmanned factory according to claim 1 is characterized in that: Based on the sorting instruction set, the robot arm is controlled to complete the grabbing and placement operations of the PCB panels. The visual verification module is used to perform real-time quality inspection on the sorting results. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality verification data are output. The specific steps include: Parse the robotic arm grabbing coordinate sequence and AGV target placement coordinates in the sorting instruction set, drive the robotic arm end effector to the specified position through the motion control module, and complete non-destructive grabbing of PCB panels based on the composite grabbing mechanism of electromagnetic suction cup and flexible gripper, and generate a grabbing action log; The robot arm joint torque data and the pressure feedback signal of the end effector are collected in real time, and the grasping force is dynamically adjusted based on the preset mechanical tolerance threshold. If the grasping offset or abnormal pressure fluctuation is detected, the grasping retry mechanism is triggered and the grasping action log is updated.

7. The PCB intelligent sorting process based on an unmanned factory according to claim 6 is characterized in that: If an abnormal fluctuation in the grasping offset or pressure is detected, the grasping retry mechanism is triggered and the grasping action log is updated, and the following is also included: The AGV collaborative scheduling module delivers the grabbed PCB panels to the target storage location, and simultaneously activates the visual verification module deployed on the AGV carrier to perform real-time scanning of the surface integrity and placement posture of the PCB panels, generating an original verification image data stream. Performing anomaly analysis on the original verification image data stream based on a dynamic adaptive detection algorithm, including: identifying scratches, warpage, and pad oxidation defects using a defect detection model optimized by transfer learning, comparing process parameter thresholds in a standardized feature data set, and outputting anomaly determination results and defect type labels; If the abnormal judgment result is marked as a defective panel, a rework instruction is sent to the MES system through the rework signal generation module, triggering the AGV to automatically return to the sorting starting station.

8. The PCB intelligent sorting process based on an unmanned factory according to claim 1 is characterized in that: The synchronous visual inspection module performs real-time quality inspection on the sorting results. If an abnormal panel is detected, a rework signal is triggered, and a sorting completion status report and quality inspection data are output. The following also includes: The sorting completion status report and quality verification data are sent back to the MES system. The sorting efficiency and error types are statistically analyzed through the data analysis module, and the parameter library of the lightweight classification model and the optimization weight of the path planning algorithm are dynamically updated to form a closed-loop iterative mechanism.

9. A PCB intelligent sorting system based on an unmanned factory, characterized in that: Sorting is achieved by using the unmanned factory-based PCB intelligent sorting process according to any one of claims 1 to 8, wherein the PCB intelligent sorting system specifically comprises: Sorting equipment, wherein the sorting equipment is provided with an industrial camera and a scanning module for obtaining surface image data and internal structural parameters of PCB boards in real time; A robotic arm equipped with an end effector comprising an electromagnetic suction cup and a flexible gripper, for supporting non-destructive grasping of PCBs of varying thicknesses; wherein the robotic arm is integrated with a joint torque sensor and an end pressure sensor; AGV conveying components, equipped with visual verification modules and positioning devices, enable directional conveying of panels; Industrial communication network for transmitting sorting instruction sets; The control center integrates lightweight classification models, dynamic path planning algorithms, and multi-objective optimization engines to drive the coordinated operation of the sorting system; The closed-loop management module realizes the feedback of sorting data and iterative update of process parameters through the MES system interface.

Images