Automatic material conveying system of door and window processing workshop

Abstract

The invention belongs to the technical field of logistics conveying, particularly relates to an automatic material conveying system for a door and window processing workshop, and aims to solve the problem of low efficiency caused by insufficient automation, flexibility and intelligence of an existing system. The system comprises a data acquisition and perception module, a digital twinning module, a task management and scheduling module, an intelligent path planning and obstacle avoidance module, a multi-agent cooperative control module, an execution and feedback module, a man-machine interaction and visualization module and the like. And multi-mode collaborative operation, real-time path optimization and dynamic task allocation of the material conveying equipment are realized. The production efficiency, the flexibility and the intelligent level of a door and window processing workshop are remarkably improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of logistics transportation, and particularly relates to an automatic material transportation system for a door and window processing workshop. BACKGROUND

[0002] Industrial automation and intelligent manufacturing have become the core driving force for the transformation and upgrading of modern manufacturing. Among them, an efficient material transportation system is a key link to ensure smooth production processes and improve overall production efficiency. Such a system aims to optimize the flow path and management of materials within a factory, thereby reducing operating costs and shortening production cycles, and plays a fundamental role in achieving lean and intelligent production.

[0003] Among them, the door and window processing industry has higher requirements for the automation, flexibility and precision of the material transportation system due to the diversity of its products, the non-standardization of profile sizes and the complexity of processing technology. The automatic material transportation system emerged as the times required, and its core is to realize the whole-process optimization management and efficient flow of materials from raw material storage to finished product delivery in the door and window processing workshop by integrating various automated equipment and intelligent control technology.

[0004] The existing technology still faces significant challenges in the material transportation of door and window processing workshops. The production characteristics of door and window products, such as multi-variety, small batch and customization, as well as the different forms of materials such as profiles and glass, make the traditional material handling method labor-intensive, low-efficiency and prone to material damage or mismatch due to human factors. Although some deployed automated systems have introduced mechanical equipment, they often lack integration, lack effective coordination and information sharing between subsystems, forming data silos, making it difficult to achieve precise positioning and real-time tracking of materials. In addition, their scheduling strategies are usually fixed, making it difficult to adapt to dynamic changes in production tasks and temporary order requirements, resulting in material delays, extended processing unit waiting times or imbalanced production rhythms. These problems not only constrain the overall production efficiency and flexibility of the door and window processing workshop, but also significantly increase operating costs and management difficulties, and an advanced, efficient and intelligent automatic material transportation system is urgently needed. SUMMARY

[0005] The present application aims to solve the deficiencies of existing door and window processing workshop material conveying systems in terms of automation, flexibility, intelligence, and response to complex and variable production requirements. Existing material conveying systems usually rely on fixed routes or relatively simple scheduling logic, making it difficult to effectively adapt to dynamic adjustments of production orders, real-time changes in equipment status, and complex conveying requirements brought by differences in material types, sizes, and weights. This results in low material flow efficiency, frequent path conflicts, low resource utilization, and strong dependence on manual intervention, thereby limiting the overall production efficiency and automation level of door and window processing workshops. To solve the above technical problems, the present application proposes an automated material conveying system for door and window processing workshops.

[0006] According to one aspect of the present application, an automated material conveying system for door and window processing workshops is provided, which realizes multi-modal collaborative operation, real-time path optimization, and dynamic task allocation of various material conveying equipment by constructing an integrated intelligent scheduling and control platform. The system first comprehensively perceives the material state, equipment operation, and production task demand in the workshop through multi-source data fusion technology. Then, based on the digital twin model, the physical workshop is mirrored in real time to provide accurate context information for intelligent decision-making. Subsequently, the core intelligent task management and scheduling module uses advanced artificial intelligence algorithms to dynamically allocate and prioritize conveying tasks, and combines predictive analysis to optimize material flow path and timing. At the same time, the multi-agent collaborative control module ensures the conflict-free and efficient operation of different types of conveying equipment in the shared space. Finally, through precise execution and feedback mechanisms, the automated and precise conveying of materials is realized, significantly improving the production efficiency, flexibility level, and intelligence level of door and window processing workshops.

[0007] As an embodiment of the present application, the automated material conveying system comprises: a data acquisition and perception module for real-time acquisition of various production state data, material information data, and conveying equipment operation data in the door and window processing workshop; a digital twin and state maintenance module for constructing and updating the digital twin model of the physical entities in the workshop based on the data acquired by the data acquisition and perception module; a task management and scheduling module for generating, allocating, and optimizing material conveying tasks based on real-time information provided by production orders, inventory status, and the digital twin model; an intelligent path planning and obstacle avoidance module for planning the optimal travel path for each conveying unit in the material conveying task and performing real-time dynamic obstacle avoidance; a multi-agent collaborative control module for coordinating the conflict-free operation of multiple conveying units in the shared space to optimize traffic flow; An execution and feedback module for receiving scheduling instructions and controlling specific material conveying equipment to execute conveying tasks, and feeding back execution results to the data collection and perception module; A human-computer interaction and visualization module for providing real-time monitoring of system status, manual intervention interface, and historical data query function.

[0008] As an embodiment of the present application, the data collection and perception module comprises: A production line state sensor group for real-time monitoring of the working status, processing progress, and fault information of each processing equipment; A material information recognition device for identifying the type, specification, quantity, current location, and target destination of materials such as profile, glass, hardware, etc. to be processed; A conveying equipment state sensor group for real-time acquisition of the position, speed, attitude, power, and load state of conveying units such as automated guided vehicles, mechanical arms, and conveying belts; An environmental perception sensor group for monitoring environmental parameters such as temperature, humidity, and possible dynamic obstacle information in the workshop.

[0009] Further, the production line state sensor group comprises photoelectric sensors, proximity switches, encoders, and industrial cameras for detecting the position, speed, and processing state of workpieces on the production line.

[0010] Further, the material information recognition device comprises a two-dimensional code scanner, a radio frequency identification reader, and a machine vision system for automatic identification and tracking of materials. The machine vision system uses image processing technology to identify the shape, size, surface defects, and type characteristics of materials and encodes them as digital information for system processing.

[0011] Further, the conveying equipment state sensor group comprises a global positioning system receiver, an inertial measurement unit, a laser radar sensor, and an ultrasonic sensor for providing accurate position, speed, attitude, and obstacle information of the surrounding environment of the conveying unit. The laser radar sensor constructs a high-precision two-dimensional or three-dimensional point cloud map of the workshop environment for path planning and obstacle avoidance.

[0012] Further, the environmental perception sensor group comprises temperature sensors, humidity sensors, and additional laser radars or cameras for auxiliary monitoring of changes in the workshop environment, such as detecting smoke, abnormal temperature, or sudden personnel intrusion.

[0013] As an embodiment of the present application, the digital twin and state maintenance module comprises: A workshop layout modeling unit is configured to build a static 3D model of the door and window processing workshop, including the geometric topology of processing equipment, storage area, conveying channel and safety area; A dynamic entity modeling unit is configured to create a dynamic digital model of each material conveying unit and the material to be processed, including its attributes, state and behavior rules; A real-time data mapping unit is configured to map the real-time data obtained by the data acquisition and perception module to the corresponding digital twin entity, updating its position, state and attributes; A state prediction and analysis unit is configured to predict potential events such as equipment failure, material consumption and traffic congestion based on historical data and real-time state, and perform trend analysis.

[0014] Further, the workshop layout modeling unit uses building information modeling technology or industrial computer-aided design data to generate a high-precision 3D geometric model containing the positions of all fixed infrastructure and equipment. This model serves as the basic framework for the digital twin.

[0015] Further, the dynamic entity modeling unit establishes an independent digital agent for each automated guided vehicle, each robotic arm and each batch of material to be processed. These digital agents contain real-time attributes such as speed, acceleration, power, load, task ID, destination, and material ID, type, size, weight, etc.

[0016] Further, the real-time data mapping unit periodically receives data streams from the data acquisition and perception module through a high-speed data bus and standard communication protocols such as Ethernet or industrial Internet of Things protocols, and binds and updates them with the corresponding entities in the digital twin model. The data update frequency can reach real-time level.

[0017] Further, the state prediction and analysis unit uses machine learning models such as recurrent neural networks or long short-term memory networks to analyze historical operation data and sensor data. By learning equipment wear patterns, traffic flow patterns and material consumption trends, this unit can predict equipment remaining life, potential traffic bottlenecks and future material demand, and send these predictions to the task management and scheduling module for forward-looking decision-making.

[0018] As an embodiment of the present application, the task management and scheduling module includes: A task generation unit is configured to automatically generate material conveying tasks based on production orders issued by the production management system, current inventory levels and material demand; A task decomposition and priority sorting unit is configured to decompose complex conveying tasks into smaller subtasks and assign priorities to each subtask based on the urgency of the production plan, the importance of the material and the state of the equipment; a resource allocation unit for assigning the most suitable material transport task to the transport unit according to its current location, availability, load capacity, and task priority; a scheduling strategy optimization unit for optimizing the overall scheduling scheme by considering factors such as transport efficiency, energy consumption, equipment utilization, and waiting time using multi-objective optimization algorithms.

[0019] Further, the task generation unit receives production instructions from the manufacturing execution system and enterprise resource planning system, and automatically identifies material shortages or upcoming material needs by combining inventory information in the digital twin model, and creates corresponding task descriptions such as "transport C material from point A to point B".

[0020] Further, the task decomposition and priority sorting unit decomposes large batches of material transport tasks into multiple batches or single item transport sub-tasks. Priority sorting is calculated according to a configurable weight function that considers task deadlines, production line downtime risks, material value, and current production load, among other dimensions.

[0021] Further, the resource allocation unit uses algorithms based on reinforcement learning or auction mechanisms to select the most suitable transport unit from multiple available units for the current task. For example, for heavy materials, the system will preferentially select automated guided vehicles with high carrying capacity; for urgent tasks, the closest and currently idle unit is selected. The allocation process takes into account the current task queue and estimated completion time of the transport unit.

[0022] Further, the scheduling strategy optimization unit uses heuristic search methods such as genetic algorithms or ant colony algorithms to find the optimal solution in a vast scheduling scheme space. The optimization objective function includes minimizing total transport time, minimizing transport energy consumption, maximizing average equipment utilization, and minimizing production line waiting time, among others. This unit periodically re-evaluates the scheduling scheme to adapt to dynamic changes in the state of the workshop.

[0023] As an embodiment of the present application, the intelligent path planning and obstacle avoidance module includes: a global path planning unit for calculating an initial optimal path from the starting point to the target point for the transport unit assigned to the task on the workshop layout topology of the digital twin model; a local obstacle avoidance and path correction unit for real-time detection of dynamic obstacles around the transport unit, such as other moving transport units, temporarily stacked materials, or personnel, and immediate adjustment of the current travel path to avoid collisions; a multi-path pre-computation and selection unit for pre-computing multiple alternative paths that can quickly switch to sub-optimal paths when the main path is blocked; A path conflict prediction and resolution unit is used to predict potential path intersections or congestions between different transport units and adjust their speed or path in advance to avoid conflicts.

[0024] Further, the global path planning unit adopts an improved A* algorithm or a fast search random tree algorithm to generate a theoretically optimal path in open space based on consideration of the transport unit size, maximum speed, and static obstacles in the workshop. The cost function of the algorithm considers path length and estimated transit time.

[0025] Further, the local obstacle avoidance and path correction unit combines laser radar or visual sensor data to construct a local environment grid map around the transport unit. The unit uses a dynamic window method or a potential field-based method to calculate the required steering angle and speed adjustment for obstacle avoidance in real time. When an obstacle is detected, the path correction unit will slow down, stop, or detour within a safe distance.

[0026] Further, the multi-path pre-computation and selection unit, in addition to generating the optimal path, also simultaneously calculates and stores several alternative paths that meet sub-optimal conditions during the global path planning stage. These alternative paths can be quickly activated through a simple path evaluation function to reduce decision-making delay when the main path fails.

[0027] Further, the path conflict prediction and resolution unit receives motion prediction information from the multi-agent collaborative control module for adjacent transport units and uses a time window algorithm or game theory method to identify potential collision or congestion points. For predicted conflicts, the unit sends instructions to relevant transport units to slow down, stop and wait, or detour to ensure smooth traffic flow.

[0028] As an embodiment of the present application, the multi-agent collaborative control module includes: A traffic flow management unit is used to allocate and manage the passage rights of multiple transport units at the shared transport channels and intersections in the workshop to prevent traffic congestion. A formation and coordination unit is used to coordinate multiple transport units to form a cooperative formation to complete the transport task when large or heavy materials need to be transported simultaneously. A fault tolerance and task redistribution unit is used to automatically redistribute the unfinished tasks of a transport unit to other available units and adjust the paths of affected units when the transport unit fails. An energy management and charging scheduling unit is used to monitor the power status of all automated guided vehicles and intelligently schedule them to go to the charging station for charging to ensure the continuous operation of the system.

[0029] Further, the traffic flow management unit controls the priority of the traffic on the main road and intersection between vehicles by implementing virtual traffic lights or time slice rotation mechanism. For example, when multiple delivery units approach an intersection at the same time, the system will allocate the right of way according to their task priority or first-come-first-served principle.

[0030] Further, the formation grouping and coordination unit enables multiple automated guided vehicles to maintain the preset relative position and speed accurately through cooperative sensing and distributed control algorithms. This is particularly important for scenarios where multiple vehicles need to be synchronized to lift or transport oversized profiles or large glass panels.

[0031] Further, the fault tolerance and task reassignment unit continuously receives device state information from the execution and feedback module. Once a hardware failure or communication interruption is detected in a delivery unit, the unit immediately marks its in-transit task status as abnormal and initiates an emergency dispatch process to assign the remaining tasks to the nearest idle delivery unit with matching capabilities.

[0032] Further, the energy management and charging scheduling unit predicts the charging needs of the automated guided vehicles based on their real-time power, estimated task volume, and available status of charging stations. This unit plans charging tasks in advance, guiding low-power automated guided vehicles to idle charging piles during task gaps or off-peak periods to avoid interruptions due to power depletion.

[0033] As an embodiment of the present application, the execution and feedback module includes: The instruction analysis and driving unit receives instructions from the task management and scheduling module and the multi-agent collaborative control module and converts them into bottom-layer control signals recognizable by the delivery device; The motion control unit precisely controls the driving speed, steering angle, lifting mechanism, and motion trajectory of the mechanical arm of the automated guided vehicle; The execution state monitoring unit monitors the actual execution state of the delivery device in real time, such as position, speed, load sensor data, and fault indication; The data upload unit encapsulates the data obtained by the execution state monitoring unit and sends it to the data acquisition and sensing module through the communication interface.

[0034] Further, the instruction analysis and driving unit translates abstract instructions such as "move to X, Y coordinates" or "grab the material" into specific motor speed, steering angle, or hydraulic system pressure instructions and sends them to the corresponding actuator controller.

[0035] Further, the motion control unit ensures the transport equipment can move precisely according to the planned speed and trajectory through a closed-loop control system, such as a proportional-integral-derivative controller. For robotic arms, its motion control involves inverse kinematics solving and torque control to achieve precise grasping and placing.

[0036] Further, the execution status monitoring unit continuously acquires the running parameters and status flags of the equipment using the encoders, current sensors, limit switches, and vision feedback systems carried by the transport equipment itself.

[0037] Further, the data uploading unit uses low-latency wireless communication protocols such as Wi-Fi or the fifth generation mobile communication technology to efficiently transmit real-time execution data streams back to the data collection and perception module, ensuring the freshness of the digital twin model data.

[0038] As an embodiment of the present application, the human-computer interaction and visualization module includes: A real-time monitoring interface for graphically displaying the digital twin model of the door and window processing workshop, including the real-time positions, states, material flow directions, and production line operating conditions of all transport units; A task management interface for manually creating, modifying, or canceling material transport tasks and adjusting task priorities; An alarm and event recording unit for generating and displaying alarms when the system detects abnormal conditions such as equipment failure, material jamming, or security area intrusion, while recording all event logs; A historical data query and report generation unit for querying and analyzing historical data of system operation to generate reports on efficiency, energy consumption, failure rate, etc.

[0039] Further, the real-time monitoring interface uses three-dimensional rendering technology to reproduce the physical layout of the workshop on the display. Icons or animation effects of different colors represent transport units and materials in different states, such as green for normal operation and red for failure or abnormality.

[0040] Further, the task management interface provides intuitive drag-and-drop operations, allowing operators to simply click and drag materials from one location to another and automatically trigger task generation and scheduling.

[0041] Further, the alarm and event recording unit stores all system events according to timestamps and classifies them into warnings, errors, or emergency events. Emergency alarms can notify relevant management personnel through sound and light signals or SMS / email.

[0042] Further, the historical data query and report generation unit allows users to set time ranges and query conditions, such as querying the utilization rate of a specific conveying unit or the average conveying time of a certain type of material. Reports can be exported in multiple formats to support decision analysis.

[0043] Compared with the prior art, the advantages and positive effects of the present application are: The present application constructs an automatic material conveying system for a door and window processing workshop. The system fundamentally improves the intelligence, flexibility and efficiency of material conveying by integrating multi-source heterogeneous data acquisition, digital twin technology and multi-agent collaborative control.

[0044] Specifically, the present application realizes real-time and accurate perception of the physical world of the workshop through comprehensive data acquisition and perception modules, covering multiple dimensions such as production lines, materials, conveying equipment and environment. This provides a solid data foundation for subsequent intelligent decision-making, overcoming the defects of information isolation and strong lag in traditional systems.

[0045] Further, the digital twin and state maintenance module introduced in the present application accurately maps the physical workshop in the digital space and performs real-time updating and state prediction. This technological breakthrough enables the system to conduct panoramic monitoring, in-depth analysis and trend prediction of the workshop operating state, providing forward-looking support for scheduling decisions, which is significantly superior to existing systems that rely only on historical data or simple rules.

[0046] In addition, the task management and scheduling module in the present application can dynamically generate, decompose, prioritize and optimize material conveying tasks according to production orders, inventory and real-time workshop state by using advanced artificial intelligence algorithms. This enables the system to flexibly respond to frequent adjustments in production plans and unexpected events, realizing truly flexible production and avoiding the inefficiency and rigidity caused by traditional fixed scheduling.

[0047] As an important advantage of the present application, the intelligent path planning and obstacle avoidance module not only calculates the global optimal path, but more importantly, it has the ability of real-time local obstacle avoidance, multi-path precalculation, and path conflict prediction and resolution. This ensures that the conveying unit can safely, efficiently and collision-free operate in a complex dynamic environment, greatly reducing the risk of collision and traffic congestion, thereby improving material turnover efficiency and equipment safety.

[0048] Further, the multi-agent collaborative control module realizes seamless collaboration of heterogeneous conveying equipment and system-level optimization through traffic flow management, formation and cooperation, fault tolerance and task redistribution, and energy management and charging scheduling. This solves the conflict and resource waste problems that may occur when multiple devices operate in a shared space, improves the overall system utilization and robustness, and ensures the continuity and stability of the production process.

[0049] The human-computer interaction and visualization module of the present application provides an intuitive, comprehensive real-time monitoring interface and a flexible task management interface, reduces the operation difficulty, improves the management efficiency, and supports historical data analysis, thereby providing strong data support for production management decision-making.

[0050] In summary, the present application can significantly improve the material flow efficiency of the door and window processing workshop, reduce the operating cost, improve the production flexibility and the ability to cope with complex production environment by constructing a highly integrated, intelligent and adaptive automated material conveying system, thereby providing key support for realizing intelligent manufacturing. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a schematic diagram of the overall technical scheme architecture of the present application; Figure 2 is a schematic diagram of the core principle framework of intelligent task management and scheduling in the present application. DETAILED DESCRIPTION

[0052] Please refer to Figure 1 and Figure 2 , the present application discloses an automated material conveying system applied to a door and window processing workshop, which aims to innovate the efficiency, accuracy and intelligent level of material flow in the traditional door and window manufacturing process. By integrating advanced sensing technology, digital twin modeling, artificial intelligence scheduling and multi-agent collaborative control, the system can realize adaptive management of complex and variable production environment, thereby significantly improving the overall production efficiency and flexibility.

[0053] The present automated material conveying system mainly includes a data acquisition and perception module, a digital twin and state maintenance module, a task management and scheduling module, an intelligent path planning and obstacle avoidance module, a multi-agent collaborative control module, an execution and feedback module, and a human-computer interaction and visualization module. These modules closely cooperate to form a highly integrated and functionally complete intelligent ecosystem, which collectively supports the full-process automated material conveying of the door and window processing workshop.

[0054] The data acquisition and perception module is the "senses" of the entire system, and its core function is to obtain real-time, comprehensive and accurate key production state data, material information data and conveying equipment operation data inside the door and window processing workshop. The module deploys multiple types of sensors and recognition devices to build a digital information network covering the physical entities in the workshop. Specifically, the data acquisition and perception module is subdivided into a production line state sensor group, a material information recognition device, a conveying equipment state sensor group and an environment perception sensor group.

[0055] The production line status sensor group is responsible for real-time monitoring of the working status, processing progress, and fault information of each processing equipment. This sensor group is composed of photoelectric sensors, proximity switches, encoders, and industrial cameras. Photoelectric sensors detect the presence, position, and passage of workpieces by emitting and receiving light beams. For example, at the entrance of a profile cutting machine, it detects whether the profile is in place, or at the exit of a drilling machine, it confirms that the workpiece has been removed. Its output signal is a binary state value, accurately reflecting the presence or absence of the workpiece, with a typical response time of milliseconds, ensuring real-time monitoring on high-speed production lines. Proximity switches are used for non-contact detection of the approach of metal or non-metal workpieces, such as monitoring whether the mechanical arm gripping point is ready or detecting the limit position of the feeding mechanism. Its detection range is generally a few millimeters to a few tens of millimeters, and the output signal is also a switch value. Encoders are installed on the conveyor belt motor or rotating parts of processing equipment to accurately measure their rotational speed, angular displacement, or linear displacement, thereby calculating the running speed of the processing equipment, the conveying distance of the workpiece, and the completion degree of the work cycle. The pulse signal output by the encoder can be converted into linear speed or position data after being processed by a counter, with an accuracy of microns. Industrial cameras use image acquisition and processing technology to monitor key processing links on the production line in real time, such as the accuracy of profile cutting, the positional deviation of drilling, and the quality defects of glass surfaces. Industrial cameras capture high-resolution image sequences, which are analyzed by image processing units to extract geometric dimensions, surface features, and defect information of workpieces, and quantize them into digital indicators. These indicators are compared with pre-set standards to evaluate processing quality and progress. All these sensor data are uploaded in real time to the central processing unit of the data acquisition and perception module through industrial Ethernet or fieldbus protocols (such as Profinet or EtherCAT). The data are preliminarily formatted and time-stamped before uploading to ensure data synchronization and integrity. Abnormal data or data exceeding the pre-set threshold will be immediately marked and trigger an alarm mechanism for system exception handling.

[0056] The material information recognition device focuses on identifying the type, specification, quantity, current location, and target destination of various materials such as processed profiles, glass, hardware, etc. The core components of the device include a two-dimensional code scanner, a radio frequency identification reader, and a machine vision system. The two-dimensional code scanner reads the two-dimensional code label attached to the material or material tray through optical reading, quickly obtaining pre-stored data such as batch number, production date, supplier information, and material code. The scanner usually uses CMOS or CCD image sensors, with fast decoding speed and extremely low error rate, and can adapt to different lighting conditions. The radio frequency identification reader uses radio waves to exchange data with the radio frequency identification tag embedded in the material or tray, automatically identifying material information. Radio frequency identification technology has strong penetration and can identify without alignment, making it particularly suitable for batch identification and material tracking in harsh environments. The reader transmits radio frequency signals through the antenna, receives the unique identification code and stored material attribute data from the tag, and parses these data into a structured format. The machine vision system uses high-resolution industrial cameras to capture material images and uses image processing algorithms (such as edge detection, feature extraction, pattern recognition, etc.) to identify the shape and size of the material, surface defects, and type characteristics. For example, by analyzing the cross-sectional profile of the profile to identify the specific model, and by detecting scratches or bubbles on the surface of the glass to determine its quality grade. The recognition results output by the machine vision system include the physical dimensions (length, width, height, diameter) of the material, color, texture, defect location and type, etc., which are encoded into digital information for system processing. These identification information and the current location of the material (usually provided by the encoder on the conveyor belt or external positioning system) are combined with the material requirements in the production order to automatically infer its target destination, such as entering the cutting station, drilling station, or glass cleaning station. All recognition data is transmitted through standard communication interfaces (such as RS232, Ethernet), and data verification and encryption processing are performed to ensure the accuracy and security of information transmission.

[0057] The conveyor equipment state sensor group is used to obtain the position, speed, attitude, power, and load state of the automated guided vehicle, mechanical arm, conveyor belt, and other conveying units in real time. The sensor group is composed of a global positioning system receiver, an inertial measurement unit, a laser radar sensor, and an ultrasonic sensor. The global positioning system receiver (which can be replaced by a UWB ultra-wideband positioning system or a wireless local area network fingerprint positioning system in an indoor environment) provides accurate three-dimensional position coordinates of the conveying unit. The system calculates its own position by receiving signals from multiple base stations, with an accuracy of centimeters. The inertial measurement unit includes a three-axis accelerometer and a three-axis gyroscope, which are used to measure the linear acceleration and angular velocity of the conveying unit. Through integration, real-time speed and attitude (pitch, roll, and yaw angle) information can be obtained. Its data update frequency is as high as hundreds of hertz, providing high dynamic response for high-speed motion control. The laser radar sensor obtains distance information of the surrounding environment by emitting a laser beam and measuring the time of flight of the reflected beam, constructing a high-precision two-dimensional or three-dimensional point cloud map of the workshop environment. Laser radar can accurately identify static obstacles (such as walls and fixed equipment) and dynamic obstacles (such as other conveying units and personnel), with a distance measurement accuracy of millimeters and a high scanning frequency, making it a key data source for path planning and real-time obstacle avoidance. The ultrasonic sensor detects close-range obstacles by emitting ultrasonic pulses and measuring their echo time, such as detecting whether there are sudden obstacles in the direction of the conveying unit or whether the distance to other units is too close. The detection range of the ultrasonic sensor is usually tens of centimeters to several meters, with fast response and relatively low cost. As an auxiliary obstacle avoidance means for laser radar, it is used for precise collision avoidance at close range. In addition, the conveying equipment also integrates current sensors and voltage sensors to monitor power status, as well as weighing sensors or torque sensors to obtain load information. After preprocessing by the internal controller of the equipment, all these sensor data are uploaded in real time to the central data bus of the data acquisition and perception module through wireless local area networks or cellular networks (such as the fifth generation mobile communication technology) using MQTT or HTTP protocols, ensuring data freshness and reliability.

[0058] The environmental perception sensor group is used to monitor environmental parameters in the workshop, such as temperature, humidity, and possible dynamic obstacle information. This sensor group includes temperature sensors, humidity sensors, and additional lidar or cameras. Temperature sensors and humidity sensors monitor the temperature and relative humidity of the workshop environment in real time, which is crucial for the stable operation of certain environmentally sensitive materials (such as glass storage conditions) and equipment. The data is periodically uploaded in digital signals, with a typical sampling period of seconds. Additional lidar or cameras are used for auxiliary monitoring of changes in the workshop environment. For example, when unexpected areas are intruded by personnel, cameras detect abnormal personnel activity through visual analysis and trigger safety alarms; lidar can be used to construct a real-time three-dimensional environmental model of the entire workshop, detect temporarily stacked materials or equipment, and ensure the safety and orderliness of the entire workshop environment. These environmental perception data are fused with other data to form a comprehensive and high-precision real-time perception of the physical world of the workshop, laying a data foundation for subsequent digital twin construction and intelligent scheduling decisions.

[0059] The digital twin and state maintenance module serves as the "brain" of the system, responsible for constructing and updating the digital twin model of the physical entities in the workshop based on the rich data obtained by the data acquisition and perception module. This module accurately maps the static structure and dynamic behavior of the physical world in the digital space and conducts state maintenance and predictive analysis based on this. The digital twin and state maintenance module consists of a workshop layout modeling unit, a dynamic entity modeling unit, a real-time data mapping unit, and a state prediction and analysis unit.

[0060] The workshop layout modeling unit is responsible for constructing a static three-dimensional model of the door and window processing workshop, covering the geometric topology of processing equipment, storage areas, transportation channels, and safety areas. This unit uses building information modeling technology or industrial computer-aided design data to generate high-precision three-dimensional geometric models containing the positions of all fixed infrastructure and equipment. These models are stored in parameterized three-dimensional geometric data (such as STEP, IGES format) and describe the precise positions, sizes, and geometric shapes of fixed elements such as walls, columns, production line equipment (such as cutting machines, milling machines, welding machines), workbenches, material shelves, charging stations, and safety fences. Each element is attached with its physical properties (such as material, load-bearing capacity) and logical properties (such as equipment number, functional area division). This high-precision three-dimensional model is the foundation skeleton of the entire digital twin, providing accurate physical references for subsequent path planning, space management, and visualization.

[0061] The dynamic entity modeling unit is dedicated to creating dynamic digital models of each material conveying unit and the material to be processed, including its attributes, state, and behavior rules. This unit establishes independent digital agents for each automated guided vehicle, each robotic arm, and each batch of material to be processed. These digital agents exist in the form of objects or entities in the digital twin model, with a series of real-time attributes that can be updated. For automated guided vehicles, their digital agents include attributes such as real-time speed vector, acceleration vector, current three-dimensional position coordinates, attitude (yaw angle, pitch angle, roll angle), battery power, current load mass, task identifier, target destination, and estimated arrival time. For robotic arms, their digital agents include joint angles, end effector position and attitude, grasping state, current load, etc. For materials to be processed, their digital agents include material identifier, type, size (length, width, height), weight, batch information, current position, target processing procedure, and priority, etc. These dynamic entity models not only store real-time data, but also encapsulate their behavior rules, such as the kinematics and dynamics models of automated guided vehicles, the inverse kinematics constraints of robotic arms, and the state transition logic of materials between different stations. These digital agents can accurately reflect the instantaneous state and potential behavior of physical entities, and are the key to achieving dynamic scheduling and control.

[0062] The real-time data mapping unit is responsible for mapping real-time data obtained by the data acquisition and perception module to the corresponding digital twin entities and updating their positions, states, and attributes. This unit periodically receives various real-time data streams from the data acquisition and perception module through high-speed data buses and standard communication protocols, such as Ethernet or industrial Internet of Things protocols (such as OPC UA, MQTT). These data streams include but are not limited to production line status, material identification information, conveying equipment position, speed, attitude, power, load, and environmental parameters. The real-time data mapping unit contains a data analysis and verification engine inside, which performs format conversion, noise reduction, outlier detection, and consistency checking on the received raw data. The verified data is bound to the corresponding entities in the digital twin model through predefined mapping rules. For example, the position data from the automated guided vehicle global positioning system receiver will be accurately updated to the "current three-dimensional position coordinates" attribute of the corresponding automated guided vehicle digital agent; the material type and quantity data from the material information identification device will update the "type" and "quantity" attributes of the corresponding batch of material digital agents. The data update frequency can reach real-time level, with a typical update period of 50 to 100 milliseconds, ensuring that the digital twin model is always highly synchronized with the physical world and provides high-freshness data support.

[0063] The state prediction and analysis unit predicts potential events such as equipment failure, material consumption, and traffic congestion based on historical data and real-time states, and performs trend analysis. This unit uses machine learning models such as recurrent neural networks or long short-term memory networks to analyze historical operation data and sensor data. By learning equipment wear patterns (such as abnormal motor current, vibration spectrum changes), traffic flow patterns (such as the efficiency of a specific intersection during peak hours), and material consumption trends (such as daily consumption of different types of profiles), this unit can predict equipment remaining life, potential traffic bottlenecks, and future material requirements. For example, by analyzing historical battery discharge curves and task loads, it can predict when an automated guided vehicle will have a battery level below a preset threshold under the current task queue. These predictions are sent to the task management and scheduling module in structured data form (such as predicted events, probability of occurrence, and prediction time window) to make forward-looking decisions, such as scheduling equipment maintenance in advance, adjusting material procurement plans, or optimizing delivery routes to avoid production interruptions or efficiency declines. This unit also performs trend analysis to identify long-term operating patterns and potential risks, generating reports for management reference.

[0064] The task management and scheduling module is the "decision center" of the system, generating, assigning, and optimizing material delivery tasks based on production orders, inventory status, and real-time information provided by the digital twin model. The degree of intelligence of this module directly determines the flexibility and efficiency of the entire system. The task management and scheduling module consists of a task generation unit, a task decomposition and priority sorting unit, a resource allocation unit, and a scheduling strategy optimization unit.

[0065] The task generation unit automatically generates material delivery tasks based on production orders issued by the production management system, current inventory levels, and material requirements. This unit receives production instructions from the manufacturing execution system and enterprise resource planning system, such as "Production Order No. XXX, 100 sets of window frames need to be processed, materials A, B, and C are needed." At the same time, it combines real-time inventory information in the digital twin model (such as the real-time quantity of profile A in the warehouse and the quantity of material B in the current production line buffer area) to automatically identify material shortages or upcoming material requirements. Once it detects that the material requirements do not match the existing inventory, or that a new material is needed for a certain processing station to complete the current batch processing, the task generation unit will automatically create a task description for "transporting material C from point A to point B." This task description includes detailed information such as material identifier, quantity, starting location (current material location or warehouse location), target location (such as cutting machine loading port), required completion time, material type, size, weight, and task creation time.

[0066] The task decomposition and prioritization unit breaks down complex material transport tasks into smaller sub-tasks and assigns a priority to each sub-task based on the urgency of the production plan, the importance of the material, and the state of the equipment. Large volume material transport tasks can be broken down into multiple batches or single item transport sub-tasks to accommodate the load capacity of the transport units and the pace of the production line. For example, a task to transport 100 profiles can be broken down into 10 sub-tasks of transporting 10 profiles each. The prioritization is calculated based on a configurable weight function that considers multiple dimensions such as task deadline, risk of production line downtime, material value, and current production load. For example, an urgent material demand that would cause production line downtime is given a significantly higher priority. The priority score of a task can be calculated by the following formula: where, represents the priority score of the task; represents the urgency of the task, such as the inverse of the deadline or a quantified value of the risk of production line downtime; represents the criticality of the material to the production process, such as whether it is a unique material required for a bottleneck process; represents the value of the material; represents the demand of the current production load for the task; , , , are configurable weight parameters to balance the importance of different factors. The unit periodically re-evaluates the priority of generated tasks to adapt to the dynamic changes in the production environment.

[0067] The resource allocation unit assigns the most suitable material transport tasks to the transport units based on their current location, available state, load capacity, and task priority. The unit employs algorithms based on reinforcement learning or auction mechanisms to select the most suitable transport unit for the current task demand among multiple available units. The reinforcement learning model learns and optimizes the task allocation strategy through continuous interaction with the digital twin environment to maximize the long-term cumulative reward (e.g., transport efficiency, equipment utilization). The auction mechanism allows each transport unit (agent) to "bid" for unassigned tasks based on their current state (such as location, battery level, current task queue length, load capacity), and the task management and scheduling module selects the optimal bidder as the "auctioneer." For example, for heavy materials, the system will preferentially select automated guided vehicles with strong carrying capacity; for urgent tasks, the system will select the closest and currently idle unit. The allocation process takes into account the current task queue and estimated completion time of the transport units to avoid overloading or idling of individual vehicles, ensuring balanced and efficient use of resources.

[0068] The scheduling strategy optimization unit utilizes multi-objective optimization algorithms to consider factors such as transportation efficiency, energy consumption, equipment utilization, and waiting time to optimize the overall scheduling plan. This unit uses heuristic search methods such as genetic algorithms or ant colony algorithms to search for the optimal solution in a vast scheduling plan space. The optimization objective function can be represented as: where, represents the total transportation time, represents the total transportation energy consumption, represents the average equipment utilization, represents the average production line waiting time; , , , are weight coefficients used to balance the importance of different optimization objectives. This optimization algorithm generates a series of scheduling plans that meet the constraint conditions through iterative search, and selects the optimal plan that minimizes the objective function value. The scheduling plan includes the task sequence, start time, end time, and expected path of each transportation unit. This unit periodically re-evaluates the scheduling plan to adapt to dynamic changes in the workshop state, such as new emergency task insertion, equipment failure, or traffic congestion, thereby achieving adaptive optimization.

[0069] The intelligent path planning and obstacle avoidance module serves as the "navigation system" for the transportation units, planning the optimal travel path for each transportation unit in the material transportation task and performing real-time dynamic obstacle avoidance to ensure the safety and efficiency of the transportation process. This module consists of a global path planning unit, a local obstacle avoidance and path correction unit, a multi-path pre-computation and selection unit, and a path conflict prediction and resolution unit.

[0070] The global path planning unit calculates an initial optimal path from the starting point to the target point for the transportation unit assigned with a task on the digital twin model of the workshop layout topology. This unit uses an improved A algorithm or a fast search random tree algorithm. The A algorithm uses a heuristic function to guide the search and can quickly find the shortest path in a complex static environment. Its cost function not only considers the path length, but also incorporates estimated transit time, transportation unit size, and maximum speed constraints. The fast search random tree algorithm explores the state space through random sampling and is suitable for high-dimensional continuous spaces, with advantages for nonholonomic constraint systems (such as the minimum turning radius of automated guided vehicles). This algorithm models fixed obstacles (walls, equipment) in the workshop layout as non-passable areas, generating a theoretically optimal path in open space. The path is represented as a series of waypoints and corresponding speed instruction sequences.

[0071] The local obstacle avoidance and path correction unit detects dynamic obstacles around the transport unit in real-time, such as other moving transport units, temporarily stacked materials, or personnel, and immediately adjusts the current travel path to avoid collisions. This unit combines lidar or vision sensor data to construct a local environmental grid map around the transport unit. This local map is centered on the transport unit and updates the surrounding environment information in real-time. The unit uses dynamic window approach or potential field-based method to calculate the required steering angle and speed adjustment for obstacle avoidance in real-time. The dynamic window approach samples within the current speed space of the transport unit, predicts the future trajectories of the transport unit under different speed commands, and evaluates the safety, distance to the target point, and distance to obstacles of these trajectories to select the optimal speed command. The potential field-based method treats the target point as a source of attractive force and obstacles as a source of repulsive force, and the transport unit moves under the action of these virtual forces to avoid obstacles and approach the target. When obstacles are detected, the path correction unit will slow down, stop, or detour within a safe distance. The obstacle avoidance strategy needs to ensure that while the path is corrected, the stability of the transport unit and the continuity of the task are maintained. A simplified dynamic obstacle avoidance strategy can be represented as: where, is the command velocity vector of the transport unit, including linear and angular velocities; is the target velocity vector; is the repulsive force vector from the obstacle, whose magnitude is inversely proportional to the distance between the transport unit and the obstacle; is the attractive force vector from the target point; , , are weight coefficients used to adjust the influence of different factors on the command velocity.

[0072] The multi-path pre-computation and selection unit, in the global path planning stage, not only generates the optimal path, but also simultaneously calculates and stores several alternative paths that meet sub-optimal conditions. These alternative paths can be quickly activated through a simple path evaluation function when the main path is blocked, reducing decision-making delays. The generation of alternative paths can be achieved by modifying the cost function of the A* algorithm or introducing random perturbations to generate paths with different topologies. Each pre-computed path is accompanied by metadata such as estimated transit time, risk level, and required energy consumption. When the main path is impassable due to sudden obstacles or traffic congestion, the system can quickly select a currently optimal alternative path from the alternative path library based on real-time conditions.

[0073] The path conflict prediction and resolution unit receives motion prediction information from the multi-agent collaborative control module for adjacent conveying units and identifies potential collision or congestion points using time window algorithms or game theory methods. The time window algorithm assigns an estimated time window to each conveying unit for each road segment in the shared space, predicting conflicts by detecting overlaps in time windows. The game theory method treats each conveying unit as a rational agent, predicting and avoiding potential conflicts by analyzing their interactions. For predicted conflicts, the unit sends instructions to related conveying units to slow down, stop and wait, or detour to ensure smooth traffic flow. For example, at an intersection, if two automated guided vehicles are expected to arrive at the same time, the system will instruct one to slow down or stop and wait based on task priority or first-come-first-served principles, while the other passes through first.

[0074] The multi-agent collaborative control module is the key to realizing group intelligence in the system, coordinating the conflict-free operation of multiple conveying units in the shared space, optimizing traffic flow, forming formations, fault-tolerant, and energy management. The module consists of a traffic flow management unit, a formation formation and collaboration unit, a fault tolerance and task redistribution unit, and an energy management and charging scheduling unit.

[0075] The traffic flow management unit allocates and manages the passage rights of multiple conveying units at the conveying channel shared by the workshop and the intersection, preventing traffic congestion. The unit implements virtual traffic lights or time slice rotation mechanisms to control the priority of passing through the main roads and intersections of the workshop. Virtual traffic lights can dynamically adjust the red and green light cycle according to real-time traffic flow and the task priority of each conveying unit. The time slice rotation mechanism divides the passing time of the shared road segment into multiple time slices and assigns them to different conveying units. For example, when multiple conveying units approach an intersection at the same time, the system will allocate passage rights to them based on their task priority, distance to the intersection, or first-come-first-served principles. Units that fail to obtain passage rights will be instructed to wait outside a safe distance.

[0076] The formation formation and collaboration unit is used to coordinate multiple conveying units to form a collaborative formation when conveying large or heavy materials simultaneously. For example, when conveying ultra-long profiles or ultra-large glass panels, two or more automated guided vehicles may need to be synchronized to lift or carry. The unit uses collaborative sensing and distributed control algorithms to enable multiple automated guided vehicles to accurately maintain the preset relative position and speed. Collaborative sensing allows each unit in the formation to share its sensor data (such as position, speed, and obstacle information), establishing a common understanding of the environment and formation members. Distributed control algorithms ensure that each unit makes local decisions based on the overall goal of the formation and its own sensor data, while maintaining synchronization with other members, such as through a leader-following control strategy or virtual structure method to achieve precise control of the formation posture.

[0077] The fault tolerance and task reassignment unit is used to automatically reassign the unfinished tasks of a failed transport unit to other available units and adjust the path of the affected units when a certain transport unit fails. The unit continuously receives device status information from the execution and feedback module. Once a hardware failure (such as motor stall, sensor failure) or communication interruption of a certain transport unit is detected, the unit immediately marks the status of its in-transit tasks as abnormal and initiates an emergency scheduling process. First, it notifies other units around the failed unit to avoid secondary accidents. Then, the remaining tasks of the failed unit, including unfinished transport sub-tasks, are assigned to the nearest idle transport unit that matches the capabilities. At the same time, the path of the affected transport unit is updated to guide it to the nearest safe parking area or maintenance station.

[0078] The energy management and charging scheduling unit is used to monitor the power status of all automated guided vehicles and intelligently schedule their charging at charging stations to ensure the continuous operation of the system. Based on the real-time power of the automated guided vehicles, the estimated task volume (obtained from the task management and scheduling module), and the available status of the charging stations, the unit predicts the charging needs. The unit plans ahead for charging tasks, guiding low-power automated guided vehicles to idle charging piles during task gaps or off-peak hours. The charging scheduling strategy takes into account the capacity of the charging station, the charging time, and the impact on overall transport efficiency, aiming to maximize the availability of automated guided vehicles without affecting the production plan. For example, the system will prioritize charging at night or during lunch breaks, or perform a quick power-up on the way to the next task point after completing a short task.

[0079] The execution and feedback module is the "executor" of the system, responsible for receiving scheduling instructions and controlling specific material transport equipment to execute transport tasks, while feeding back execution results to the data collection and perception module. This module ensures that scheduling decisions in the digital space can be accurately executed in the physical world. The execution and feedback module consists of an instruction parsing and driving unit, a motion control unit, an execution state monitoring unit, and a data uploading unit.

[0080] The instruction parsing and driving unit receives high-level instructions from the task management and scheduling module and the multi-agent collaborative control module, and converts them into bottom-level control signals recognizable by the transport equipment. For example, abstract instructions such as "move to X, Y coordinates" or "grab materials" are translated into specific motor speed, servo angle, hydraulic system pressure instructions, or pneumatic valve switch instructions, and are sent to the corresponding actuator controller. This unit is responsible for unpacking, verifying, and formatting instructions to ensure that the instructions can be correctly understood and executed by the transport equipment hardware.

[0081] The motion control unit precisely controls the travel speed, steering angle, lifting mechanism, and motion trajectory of the automated guided vehicle. This unit ensures that the transport equipment moves accurately according to the planned speed and trajectory through a closed-loop control system, such as a proportional-integral-derivative controller. For the automated guided vehicle, the motion control unit adjusts the output of the drive motor based on real-time position and speed feedback to achieve precise speed tracking, path tracking, and posture maintenance. For the robotic arm, its motion control involves inverse kinematics solving and torque control. Inverse kinematics solving converts the target position and posture of the end effector into the required angles of each joint, and torque control ensures that the robotic arm can accurately grasp, place, and transport materials with the required force and speed, while avoiding overload and vibration.

[0082] The execution state monitoring unit monitors the actual execution state of the transport equipment in real time, such as position, speed, load sensor data, and fault indications. This unit uses the transport equipment's own onboard encoders (to measure motor speed and position), current sensors (to monitor motor load and abnormalities), limit switches (to detect mechanism motion range), force sensors (to measure gripping force or load weight), and visual feedback systems (such as cameras installed at the end of the robotic arm for visual servoing and grasping confirmation) to continuously obtain the equipment's operating parameters and state indicators. These data are collected in real time, preliminarily processed, and time-stamped to form a continuous equipment operation state data stream.

[0083] The data upload unit encapsulates the data obtained by the execution state monitoring unit and sends it to the data acquisition and perception module through the communication interface. This unit uses low-latency wireless communication protocols such as Wi-Fi or the fifth generation mobile communication technology to efficiently transmit real-time execution data streams back to the data acquisition and perception module, ensuring the freshness of the digital twin model data. Data compression and encryption techniques are used during data transmission to reduce bandwidth usage and ensure data security. This real-time feedback mechanism is the basis for synchronously updating the digital twin model with the physical world and is a key input for state prediction and analysis, fault tolerance, and task redistribution.

[0084] The human-machine interaction and visualization module is a bridge between the user and the system, providing real-time monitoring of system status, manual intervention interfaces, and historical data query functions. This module aims to improve the usability, manageability, and transparency of the system. The human-machine interaction and visualization module consists of a real-time monitoring interface, a task management interface, an alarm and event recording unit, and a historical data query and report generation unit.

[0085] The real-time monitoring interface displays the digital twin model of the door and window processing workshop in a graphical manner, including the real-time positions, states, material flow directions, and production line operating conditions of all conveying units. The interface uses three-dimensional rendering technology to highly replicate the physical layout of the workshop on the display, including production equipment, shelves, conveying channels, etc. Icons or animation effects of different colors represent conveying units and materials in different states, such as green icons representing normal operation and task execution, yellow representing standby or idle, and red representing failure or abnormality. The material flow direction is visually displayed through dynamic arrows or path tracks, allowing operators to easily understand the operating situation and material distribution of the entire workshop. The monitoring interface also provides zooming, panning, and rotating functions, allowing users to observe the workshop from different angles and levels of detail.

[0086] The task management interface provides functions for manually creating, modifying, or canceling material conveying tasks, and adjusting task priorities. The interface provides intuitive drag-and-drop operations, allowing operators to simply click and drag material icons from one storage area or processing station to another target location, and automatically trigger task generation and scheduling processes. Operators can also directly modify the targets, quantities, or priorities of existing tasks to respond to sudden production changes or urgent needs. All manual interventions are verified by the system to ensure the effectiveness and safety of the operations.

[0087] The alarm and event recording unit generates and displays alarms when the system detects abnormal conditions, such as device failure, material jamming, or security area intrusion, and records all event logs. All system events are stored according to timestamps and classified as warnings, errors, or emergency events, such as low battery level warnings for conveying units, mechanical arm grabbing failure errors, security grating triggered emergency shutdowns, etc. Emergency alarms can notify relevant management personnel through sound and light signals, interface pop-ups, SMS notifications, or email notifications to ensure that abnormal conditions can be discovered and handled in a timely manner. Event logs record the time, type, location, involved devices, processing results, and relevant operator information of the event, providing a basis for subsequent problem analysis and system optimization.

[0088] The historical data query and report generation unit allows users to query and analyze historical data of system operation and generate reports on efficiency, energy consumption, failure rate, etc. Users can set time ranges and query conditions, such as querying the utilization rate of a specific conveying unit in the past week, the average conveying time of a certain type of material, or the material turnover frequency of a certain processing station. Reports can be displayed in various visual forms such as charts and tables, and can be exported in formats such as PDF and Excel to support production management decision analysis and continuous improvement. These reports can help managers evaluate system performance, identify bottlenecks, and optimize production processes.

[0089] To sum up, the automatic material conveying system of the embodiment, through multi-dimensional data acquisition, digital twin model construction and real-time updating, task management and scheduling based on artificial intelligence, intelligent path planning and dynamic obstacle avoidance, multi-agent collaborative control, and intuitive human-computer interaction and visualization, builds a highly integrated, intelligent and self-adaptive material conveying solution for door and window processing workshops. The system fundamentally solves the shortcomings of traditional systems in automation, flexibility and intelligence, significantly improves material flow efficiency, reduces operating costs, improves production flexibility and the ability to cope with complex production environments, thereby providing key support for realizing intelligent manufacturing in the door and window processing industry. The system can effectively improve the overall production efficiency, automation level and intelligence level of the door and window processing workshop, bringing significant economic and social benefits to the enterprise.

Claims

1. An automated material conveying system for a door and window processing workshop, characterized in that: include: The data acquisition and perception module obtains production line status data, material information identification data, conveying equipment operation data, and environmental perception data in the door and window processing workshop; The digital twin and condition maintenance module is used to build and dynamically update the digital twin model of the physical entity of the workshop based on the real-time multi-source data obtained by the data acquisition and perception module, and to perform status prediction and trend analysis of potential events; The task management and scheduling module intelligently generates, decomposes, prioritizes, and optimizes the allocation and scheduling of material transportation tasks based on production orders, current inventory status, and precise contextual information provided by the digital twin model. The intelligent path planning and obstacle avoidance module plans the global optimal travel path for each conveying unit in the material conveying task and detects dynamic obstacles in real time based on sensor data; A multi-agent collaborative control module is used to coordinate the conflict-free and efficient operation of multiple heterogeneous transport units in a shared space. Specifically, it implements traffic flow management, multi-unit formation formation and collaboration, fault tolerance and task redistribution, as well as energy management and charging scheduling; The execution and feedback module is used to receive scheduling instructions and convert them into underlying control signals, thereby accurately controlling specific material conveying equipment to perform conveying tasks; The human-computer interaction and visualization module is used to provide real-time graphical monitoring of the system's operating status and a manual intervention interface for manually creating or modifying conveying tasks.

2. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The data acquisition and perception module includes: Production line status sensor group, used to monitor the working status, processing progress and fault information of each processing equipment in real time. The production line status sensor group includes photoelectric sensors, proximity switches, encoders and industrial cameras; A material information recognition device for identifying the type, specification, quantity, current location, and destination of the profiles, glass, and hardware to be processed. The material information recognition device includes a QR code scanner, a radio frequency identification reader, and a machine vision system. A conveyor equipment status sensor group, which is used to obtain the position, speed, posture, power level, and load status of the automated guided vehicle, robotic arm, and conveyor belt conveyor unit in real time. The conveyor equipment status sensor group includes a global positioning system receiver, an inertial measurement unit, a lidar sensor, and an ultrasonic sensor; The environmental perception sensor group is used to monitor the environmental parameters and dynamic obstacle information in the workshop. The environmental perception sensor group includes a temperature sensor, a humidity sensor, and an additional lidar or camera.

3. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The digital twin and state maintenance module includes: A workshop layout modeling unit, used to construct a static three-dimensional model of the door and window processing workshop, including the geometric topological structure of processing equipment, storage areas, conveying channels and safety areas; Dynamic solid modeling unit, used to create dynamic digital models of each material handling unit and the material to be processed, including their attributes, states and behavior rules; A real-time data mapping unit, used to map the real-time data acquired by the data acquisition and perception module to the corresponding digital twin entity and update its position, status, and attributes; The status prediction and analysis unit is used to predict potential events such as equipment failure, material consumption, and traffic congestion based on historical data and real-time status using machine learning models, and perform trend analysis.

4. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The task management and scheduling module includes: The task generation unit is used to automatically generate material delivery tasks based on the production orders issued by the production management system, the current inventory level and material requirements; Task decomposition and prioritization unit, used to decompose complex conveying tasks into smaller subtasks and assign priorities to each subtask based on the urgency of the production plan, material importance, and equipment status; The resource allocation unit is used to allocate the most suitable material transportation tasks to the transportation units based on their current location, availability, load capacity, and task priority, using an algorithm based on reinforcement learning or an auction mechanism; The scheduling strategy optimization unit is used to adopt a multi-objective optimization algorithm to comprehensively consider factors such as transportation efficiency, energy consumption, equipment utilization, and waiting time to optimize the overall scheduling plan.

5. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The intelligent path planning and obstacle avoidance module includes: A global path planning unit, configured to calculate an initial optimal path from a starting point to a target point for a transport unit to which a task has been assigned, based on the workshop layout topology of the digital twin model, wherein the global path planning unit adopts an improved A* algorithm or a fast search random tree algorithm; Local obstacle avoidance and path correction unit, which detects dynamic obstacles around the conveyor unit in real time and immediately adjusts the current travel path to avoid collisions; Multi-path pre-calculation and selection unit, used to pre-calculate multiple alternative paths and quickly switch to the suboptimal path when the primary path is blocked; The path conflict prediction and resolution unit is used to predict potential path intersections or congestion between different transport units and adjust their speeds or paths in advance to avoid conflicts.

6. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The multi-agent collaborative control module includes: Traffic flow management unit, used to allocate and manage access rights for multiple transport units in shared transport corridors and intersections between workshops by implementing virtual traffic lights or time-slice rotation mechanisms to prevent traffic congestion; The formation and collaboration unit is used to coordinate multiple conveying units to form a collaborative formation through collaborative sensing and distributed control algorithms to jointly complete the conveying task when large or heavy materials need to be transported simultaneously; Fault tolerance and task reallocation unit, which is used to automatically reallocate unfinished tasks to other available units when a transport unit fails, and adjust the paths of the affected units; The energy management and charging scheduling unit is used to monitor the power status of all automated guided vehicles and intelligently schedule them to go to charging stations for charging to ensure the system's continuous operation capability.

7. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The execution and feedback module includes: An instruction parsing and driving unit, configured to receive instructions from the task management and scheduling module and the multi-agent collaborative control module, and convert them into underlying control signals recognizable by the conveying equipment; The motion control unit is used to accurately control the AGV's speed, steering angle, lifting mechanism, and the trajectory of the robotic arm through a closed-loop control system; Execution status monitoring unit, used to monitor the actual execution status of the conveying equipment, position, speed, load sensor data and fault indication in real time; The data uploading unit is used to encapsulate the data acquired by the execution status monitoring unit and send it to the data acquisition and perception module through a low-latency wireless communication protocol.

8. The automated material conveying system for a door and window processing workshop according to claim 1, characterized in that: The human-computer interaction and visualization module includes: A real-time monitoring interface, which uses 3D rendering technology to graphically display the digital twin model of the door and window processing workshop, including the real-time position, status, material flow direction, and production line operation status of all conveying units; Task management interface, which provides intuitive drag-and-drop operations to manually create, modify, or cancel material transport tasks and adjust task priorities; Alarm and event recording unit, used to generate and display alarms when the system detects abnormal conditions, and record all event logs; The historical data query and report generation unit is used to query and analyze the historical data of system operation and generate reports on efficiency, energy consumption and failure rate.

9. The automated material conveying system for a door and window processing workshop according to claim 4, characterized in that: The task decomposition and priority sorting unit calculates its priority based on a configurable weight function, which comprehensively considers multiple dimensions such as task deadline, shutdown risk of the production line involved, material value, and current production load. The scheduling strategy optimization unit uses a genetic algorithm or ant colony algorithm heuristic search method to find the optimal scheduling solution through a multi-objective optimization function, which includes minimizing total delivery time, minimizing delivery energy consumption, maximizing average equipment utilization, and minimizing production line waiting time.

10. The automated material conveying system for a door and window processing workshop according to claim 5, characterized in that: The local obstacle avoidance and path correction unit, in combination with lidar or visual sensor data, constructs a local environment grid map around the transport unit, and uses a dynamic window method or a potential field-based method to calculate in real time the steering angle and speed adjustment required for obstacle avoidance. When an obstacle is detected, it slows down, stops, or detours within a safe distance. The path conflict prediction and resolution unit receives motion prediction information of adjacent transport units from the multi-agent collaborative control module, and uses a time window algorithm or game theory method to identify potential collisions or congestion points. For predicted conflicts, the unit sends instructions to the relevant transport units to slow down, stop, wait, or detour.

Images