Cooperative scheduling method, device and equipment for digital manufacturing platform

Abstract

The invention relates to the technical field of production management, and discloses a collaborative scheduling method, device and equipment for a digital manufacturing platform, and the method comprises the steps: generating a real-time sensing matrix through simulating the operation condition of a factory in real time, predicting material shortage information according to a real-time production work order and a material demand, obtaining the positioning data of an automatic carrying unit, and carrying out the real-time monitoring of the real-time sensing matrix; according to the method, a transport target map and a material carrying characteristic matrix are constructed, a graph neural network is utilized to simulate a transport path and optimize and drive an automatic carrying unit to carry out transport, risk assessment is carried out through multi-modal sensing data, an optimal adjustment strategy is generated, and through real-time data driving and path optimization, the precision and efficiency of production scheduling are improved, and the production efficiency is improved. The risk in the transportation process is reduced, the utilization rate of production resources is increased, and the problem that in the prior art, material demand changes are difficult to respond rapidly is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of production management, and in particular to a collaborative scheduling method, device and equipment for a digital manufacturing platform. Background Art

[0002] The core goal of the digital manufacturing platform is to highly integrate the physical production process and digital management to form an intelligent production environment with comprehensive perception, real-time feedback and precise control. By real-time perception and analysis of production site data, the platform can simulate the operating conditions of the factory, grasp key factors such as production work orders, material requirements, and transportation conditions, thereby realizing dynamic optimization of production scheduling. In modern manufacturing processes, material management and transportation scheduling are key factors to ensure smooth production. Traditional material scheduling relies on static plans and manual adjustments, which makes it difficult to quickly respond to changes in material demand during the production process and is prone to delays due to material shortages or unreasonable transportation routes. Summary of the Invention

[0003] The purpose of the present invention is to provide a collaborative scheduling method, device and equipment for a digital manufacturing platform, aiming to solve the problem in the prior art that it is difficult to quickly respond to changes in material demand.

[0004] The present invention is implemented as follows: In a first aspect, the present invention provides a collaborative scheduling method for a digital manufacturing platform, comprising: Use the digital manufacturing platform to simulate the factory's operating conditions in real time to generate a real-time perception matrix; Synchronously obtain real-time production work orders, and perform material demand forecasting on the real-time production work orders based on the real-time sensing matrix to obtain material shortage information; Obtaining the positioning data and status data of each automatic transport unit, and combining the material shortage information with the real-time perception matrix to construct a transport target map and a corresponding material transport characteristic matrix; The transport path of the transport target map is simulated by a graph neural network model, and the transport path is iteratively optimized using the material transport characteristic matrix as a supervision condition to obtain the expected transport path of each automatic transport unit; According to the expected transportation path, each automatic transport unit is driven to transport materials, and multimodal perception data of each automatic transport unit is simultaneously collected to calculate potential risk factors in real time for interactive constraint analysis to obtain the optimal adjustment strategy.

[0005] In a second aspect, the present invention provides a collaborative scheduling device for a digital manufacturing platform, configured to implement the collaborative scheduling method for a digital manufacturing platform described in any one of the first aspects, comprising: A real-time perception module, which simulates the factory's operating conditions in real time through a digital manufacturing platform to generate a real-time perception matrix; A demand forecasting module is used to synchronously obtain real-time production work orders and perform material demand forecasting on the real-time production work orders based on the real-time sensing matrix to obtain material shortage information; A target analysis module is used to obtain the positioning data and status data of each automatic transport unit, and to construct a transport target map and a corresponding material transport characteristic matrix by combining the material shortage information with the real-time perception matrix; A path analysis module is used to simulate the transportation path of the transportation target map through a graph neural network model, and iteratively optimize the transportation path using the material transportation characteristic matrix as a supervision condition to obtain the expected transportation path of each automatic transportation unit; The strategy adjustment module is used to drive each automatic transport unit to transport materials according to the expected transportation path, and simultaneously collect multimodal perception data of each automatic transport unit to calculate potential risk factors in real time for interactive constraint analysis to obtain the optimal adjustment strategy.

[0006] In a third aspect, the present invention provides a collaborative scheduling device for a digital manufacturing platform, comprising a memory and a processor, wherein the memory stores a collaborative scheduling program for the digital manufacturing platform that can run on the processor, and when the processor executes the collaborative scheduling program for the digital manufacturing platform, it implements a collaborative scheduling method for a digital manufacturing platform as described in any one of the first aspects.

[0007] In a fourth aspect, the present invention provides a readable storage medium for collaborative scheduling of a digital manufacturing platform, on which a collaborative scheduling program of the digital manufacturing platform is stored. When the collaborative scheduling program of the digital manufacturing platform is run by a processor, the processor executes a collaborative scheduling method for a digital manufacturing platform as described in any one of the first aspects.

[0008] The present invention provides a collaborative scheduling method for a digital manufacturing platform, which has the following beneficial effects: The present invention generates a real-time perception matrix through real-time simulation of factory operation conditions, predicts material shortage information based on real-time production work orders and material requirements, obtains positioning data of automatic transport units, constructs a transportation target map and a material transportation characteristic matrix, uses graph neural networks to simulate and optimize transportation routes, drives automatic transport units to carry out transportation, and simultaneously conducts risk assessment through multimodal perception data to generate the optimal adjustment strategy. This method improves the accuracy and efficiency of production scheduling through real-time data drive and route optimization, reduces risks in the transportation process, improves the utilization rate of production resources, and solves the problem of difficulty in quickly responding to changes in material demand in existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a schematic diagram of the steps of a collaborative scheduling method for a digital manufacturing platform provided by an embodiment of the present invention; Figure 2 It is a structural diagram of a collaborative scheduling device for a digital manufacturing platform provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0010] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0011] The implementation of the present invention is described in detail below with reference to specific embodiments.

[0012] Reference Figure 1 、 Figure 2 As shown, a preferred embodiment of the present invention is provided.

[0013] In a first aspect, the present invention provides a collaborative scheduling method for a digital manufacturing platform, comprising: S1: Use the digital manufacturing platform to simulate the factory's operating conditions in real time to generate a real-time perception matrix; S2: synchronously obtain a real-time production work order, and perform material demand forecasting on the real-time production work order according to the real-time sensing matrix to obtain material shortage information; S3: Acquire positioning data and status data of each automatic transport unit, and construct a transport target map and a corresponding material transport characteristic matrix by combining the material shortage information with the real-time perception matrix; S4: simulating the transport path of the transport target map using a graph neural network model, and iteratively optimizing the transport path using the material transport characteristic matrix as a supervision condition to obtain the expected transport path of each automatic transport unit; S5: Drive each automatic transport unit to transport materials according to the expected transport path, and simultaneously collect multimodal perception data of each automatic transport unit to calculate potential risk factors in real time for interactive constraint analysis to obtain the optimal adjustment strategy.

[0014] Specifically, in step S1 of the embodiment provided by the present invention, the automatic transport unit (such as an automatic transport vehicle, an unmanned transport vehicle, etc.) is equipped with sensors and positioning systems (such as GPS, lidar, a vision system, etc.) to collect its current position and status information in real time. The status data of the automatic transport unit includes battery power, carrying capacity, fault status, temperature, speed, etc. All of this information can be monitored in real time through internal sensors.

[0015] More specifically, real-time environmental data from production workshops, storage sites, roads, etc. is collected to obtain information on temperature, humidity, air quality, and the placement of materials in the workshop. These data can help determine the impact of environmental factors on material transportation and production, and monitor the operating status of factory equipment (such as conveyor belts, processing equipment, sensors, etc.) in real time, including whether it is working normally, load conditions, and whether there are any faults.

[0016] More specifically, through sensors and IoT devices, information such as roads, passages, slopes, traffic signals, obstacles, etc. within the factory is collected, with the focus on analyzing the road's traffic capacity, whether there is congestion, whether there are obstacles, etc. Sensors can also monitor the traffic status of different road sections, obtain road congestion and traffic restriction data, and help optimize the path selection of automatic transport units.

[0017] More specifically, digital twin technology is used to create a virtual model of the factory, combining the factory's geographic information and real-time collected data. By mapping the data of various parts of the factory (production areas, storage sites, roads, etc.) into the digital twin model, a real-time model that accurately reflects the actual operation of the factory can be constructed. Based on these real-time collected data and the virtual model, a real-time perception matrix is ​​generated. This matrix contains multi-dimensional real-time data such as the operating status, location, environmental changes, and road conditions of each automatic transport unit in the factory.

[0018] More specifically, various data sources (such as the status of automatic transport units, environmental sensor data, road information, equipment status, etc.) are integrated into a unified perception matrix through multi-dimensional analysis and model algorithms. This matrix not only includes physical-level data, but can also be analyzed based on machine learning and artificial intelligence technologies to predict potential production problems, material requirements, equipment failures, etc.

[0019] More specifically, the generated real-time perception matrix will reflect the overall operating status of the factory in real time. Through the real-time feedback mechanism, the platform transmits key information in the matrix to the scheduling system, helping the system optimize production scheduling and control of automated equipment. Through the real-time feedback mechanism, the system can dynamically adjust the production and transportation processes according to the perception matrix.

[0020] As can be understood, the perception matrix generated by executing the above steps can reflect the production and transportation status of the entire factory in real time, including the location of the automated transport units, power levels, transport tasks, material conditions within the factory, road conditions, and more. This allows managers to monitor every detail of the production and transportation process at any time. Real-time monitoring data provides a basis for accurate production scheduling. The scheduling system can dynamically adjust production and transportation plans based on the data in the real-time perception matrix to avoid production line stagnation or material shortages. Based on the real-time perception matrix, the system can promptly identify potential risks such as automated transport unit failures, material transportation delays, and road congestion, and take appropriate measures through early warning mechanisms to reduce the risk of production interruptions.

[0021] More specifically, through real-time analysis of the status of roads and automated transport units within the factory, the real-time perception matrix provides accurate data support for path planning and scheduling. The system can calculate the optimal path, reduce transportation time, avoid transportation conflicts, and improve transportation efficiency. As the factory environment changes (such as equipment failures, changes in production tasks, road congestion, etc.), the real-time perception matrix can adjust simulation and scheduling strategies in real time to ensure the efficiency and flexibility of the entire factory operation. By using machine learning, artificial intelligence and data analysis technologies, the real-time perception matrix provides decision-making support for the digital manufacturing platform, helping enterprises make decisions such as production scheduling, material allocation, and equipment maintenance based on real-time data. The matrix can also be used for collaborative scheduling between automated transport units. Through real-time data analysis and collaborative optimization, automated transport units can collaborate efficiently in complex environments and complete multiple transportation tasks.

[0022] Specifically, in step S2 of the embodiment provided by the present invention, real-time production work orders are synchronously obtained through the enterprise resource planning (ERP) system or the production scheduling system, and the content includes current production tasks, production goals, time plans, required materials, process requirements, etc. The update frequency of the work order is adapted to the production rhythm to ensure that the latest production needs are obtained. The system will analyze the content of the production work order and identify information such as the type, quantity and delivery time of materials required for the work order. The work order contains multiple material requirements, and the system will organize these requirements and sort them by priority.

[0023] More specifically, the work order requirements are connected to the real-time perception matrix. Based on the generated real-time perception matrix, the system will match the material requirement information in the production work order with the current material inventory, production progress, transportation status, etc. in the factory. The real-time perception matrix provides real-time material inventory, transportation status, material flow, equipment status and other information in the factory. The system will analyze the requirements in the work order and combine the current material inventory and transportation conditions to predict whether the materials can be supplied on time.

[0024] More specifically, advanced machine learning or data mining techniques (such as time series forecasting, regression analysis, etc.) are used to forecast material demand based on historical production data, current production work orders, inventory data, material transportation routes and other information. Future material demand is predicted through regression models or neural network models, and the consumption of each material within a certain period of time is calculated. As the production process progresses, changes in work orders will be fed back to the forecasting model in real time. The system can dynamically adjust material demand forecasts to accurately reflect material supply gaps.

[0025] More specifically, the work order demand forecast is compared with the current material inventory. The system will analyze whether there is a material shortage based on real-time material inventory data. If the predicted material demand exceeds the existing inventory, the system will mark it as a material shortage and generate a material shortage warning message. The material shortage information will include detailed data such as the type of missing material, the quantity of missing materials, and the expected time of shortage. The system will automatically issue a material shortage reminder through the early warning mechanism to notify production managers, warehouse managers or supply chain teams. Based on the material shortage information, production managers can take corresponding replenishment, procurement or production adjustment measures to ensure that production tasks are not affected.

[0026] More specifically, the system will automatically generate a material shortage report, which includes the name of the missing material, the quantity of the missing material, the expected time period of the shortage, and the possible impact of the shortage. For material shortage problems, the system can automatically generate emergency response plans based on the urgency of the current production task, the material procurement cycle, supplier data, etc. For example, it can recommend allocating materials from other workshops, contacting suppliers to speed up delivery, and temporarily adjusting production plans.

[0027] More specifically, based on material shortage information, the system can adjust the production plan, postpone non-urgent work orders, speed up the transportation of urgent materials, and even adjust the order of material use on the production line. The production scheduling system can flexibly reallocate resources and work orders based on the material shortage warning results to ensure that key production tasks are not affected.

[0028] It can be understood that through the above steps, the system can efficiently predict material demand, identify material shortage information, and provide intelligent support for production scheduling. By combining historical production data, real-time inventory data, work order requirements, etc., and adopting advanced forecasting models, the system can accurately predict future material demand, dynamically adjust the forecast results, and make adjustments at any time according to changes in the production site to ensure the real-time and accuracy of the forecast.

[0029] More specifically, the system can automatically compare work order requirements with inventory status and promptly detect material shortages. Material shortage information can be identified and an early warning can be generated in the first place, reducing the risk of manual intervention and delayed response. By providing early warning of material shortages, production managers can take remedial measures to avoid production stagnation and equipment idleness caused by material shortages, thereby ensuring the continuity of the production process. The system can adjust production plans in real time to reduce unnecessary delays and waste of resources. Material shortage information is closely integrated with production scheduling, and can ensure that the most urgent production tasks can be guaranteed with materials through measures such as allocating resources and adjusting priorities.

[0030] More specifically, the system can also be connected to the supply chain management system to quickly respond to the delivery arrangements of external suppliers or other supply chain optimization measures. The material shortage reports and emergency response suggestions provided by the system help decision makers respond quickly and take effective measures to deal with material shortages. Through machine learning and data analysis, the system can also continuously optimize emergency response strategies and enhance future response capabilities. Based on real-time material shortage information, the system can optimize the allocation of production resources, improve material utilization efficiency, reduce inventory backlogs and excess materials, and can flexibly schedule production lines, equipment and personnel to ensure that production goals are completed on time.

[0031] Specifically, in step S3 of the embodiment provided by the present invention, the automated transport unit (AGV, AMR, etc.) is equipped with GPS, lidar, visual sensors, and other equipment to collect real-time positioning information, including current location, target location, current route, and driving status. This positioning data is continuously updated and used to build a real-time transportation route and scheduling optimization model. The automated transport unit also monitors other key information, such as battery charge, load, fault status, speed, and fault alarms. This data is transmitted to the system in real time via onboard sensors. This status data helps determine whether the automated transport unit can complete its mission according to the planned route or whether maintenance or battery replacement is required.

[0032] More specifically, the previous step has generated material demand forecasts and material shortage information. The system will identify material shortages in real time based on work order requirements, inventory, transportation status and other information. Material shortage information will affect the priority of transportation tasks and the order of material distribution. The real-time perception matrix integrates multi-dimensional information such as production environment, material status, and automatic transport unit status, and can reflect the real-time operation of the entire factory. By combining material shortage information with the real-time perception matrix, the system can obtain information about the specific material type, location, required time and transportation route of the shortage, providing a basis for the planning of subsequent transportation tasks.

[0033] More specifically, based on work orders and production progress, combined with material shortage information, the system automatically determines the type, quantity and transportation destination of materials that need to be transported (for example, materials that need to be transported from warehouse A to production line B). The transportation target map can calculate the transportation route of each target material through graph algorithms (such as the shortest path algorithm, minimum cost path algorithm, etc.). The nodes of the transportation target map represent the source and destination of each material (such as raw material warehouse, finished product warehouse, each production line, automatic transport unit, etc.), and the edges represent the transportation path of materials from one node to another. The weight of the edge can be weighted according to the distance of the path, transportation time, load capacity of the automatic transport unit, traffic conditions during the transportation period, etc. The transportation target map can be dynamically updated and adjusted when real-time material demand or production scheduling changes.

[0034] More specifically, material types and characteristics: different types of materials have different transportation requirements. For example, heavy materials require special transportation tools, and fragile materials need to be transported slowly. The system needs to determine the appropriate transportation method by classifying and analyzing the characteristics of the materials. Each automatic transport unit has different load capacity, speed, and path adaptability (such as whether it can pass through certain narrow passages, etc.). Therefore, the material transport characteristic matrix needs to combine the characteristics of each automatic transport unit to evaluate whether it is suitable for performing specific transportation tasks. Each row represents a material, and each column represents an automatic transport unit. Each element in the matrix indicates whether the automatic transport unit is suitable for transporting the material, as well as its transportation efficiency (for example, weight bearing capacity, transportation speed, time required for transportation, etc.). The transport characteristic matrix will be dynamically adjusted according to the weight, volume, transportation distance, and transportation capacity of the automatic transport unit of each material.

[0035] More specifically, based on the transportation target map, combined with real-time material demand, transportation goals and the status of automatic transport units (such as power, failure, etc.), the system automatically optimizes the transportation path of each automatic transport unit, and uses heuristic algorithms (such as A* algorithm, Dijkstra algorithm, etc.) or optimal path planning algorithms (such as genetic algorithms, simulated annealing, etc.) to calculate the shortest path or optimal transportation path. According to the priority of materials (such as urgent materials) and the idle time of automatic transport units, path conflicts, etc., the system optimizes the execution order of transportation tasks through scheduling algorithms. According to the status of automatic transport units during transportation, path changes or emergencies (such as equipment failures, traffic jams, etc.), the system will provide instant feedback and scheduling adjustments.

[0036] It can be understood that by constructing the transportation target map and material transportation characteristic matrix through the above steps, the system can effectively optimize the scheduling of automatic transportation units. The transportation target map can provide the optimal transportation route for each transportation task, ensuring that materials can be transported in the shortest time and at the lowest cost. The system can dynamically adjust the transportation target according to material shortage information, real-time production needs and the status of the automatic transportation unit to avoid production bottlenecks caused by material shortages. Through the material transportation characteristic matrix, the system can evaluate the matching degree between different materials and automatic transportation units to ensure that the most suitable transportation tools are used for each transportation task. The system can flexibly select the most suitable transportation route and automatic transportation unit according to the characteristics and transportation needs of each material, thereby improving overall transportation efficiency.

[0037] More specifically, the system can perform intelligent scheduling based on real-time data (such as the status of automatic transport units, path conditions, material requirements, etc.), reduce manual intervention, and optimize production scheduling and material flow. During the transportation process, the system can adjust the path and scheduling based on real-time feedback to improve flexibility and response speed. Combining material shortage information with the real-time perception matrix, the system can effectively identify material shortage risks and adjust transportation tasks and priorities in a timely manner to ensure that critical materials can be delivered in time and avoid production line shutdowns. Based on the transportation target map and material transportation characteristic matrix, the system can provide production managers with detailed decision support to help make more efficient resource allocation, transportation scheduling and emergency response. By optimizing the task allocation and path planning of each automatic transport unit, the system can maximize the utilization efficiency of the automatic transport unit and reduce idle time and waiting time.

[0038] Specifically, in step S4 of the embodiment provided herein, a transport target map is first constructed into a graph structure based on the factory layout, the initial positions of automated transport units (AGVs, AMRs, etc.), and the material requirements and destinations. Each node represents a transport target (e.g., a material source, a destination, an automated transport unit, etc.). Transport targets may include raw material warehouses, production lines, finished product warehouses, etc. Each edge represents a feasible path between two nodes. Edge weights can be assigned based on factors such as distance, time, and traffic conditions. Different material transport vehicles (e.g., different types of AGVs) may have different suitable paths.

[0039] More specifically, the characteristics of each node include the type of material at that location, the required quantity, material characteristics, equipment status, etc. The node characteristics of the automatic transport unit also include information such as the current location, battery power, load, speed, etc. The characteristics of the edge usually represent the specific information of the transportation path, such as path length, estimated transportation time, traffic conditions, congestion, etc. The material transportation characteristic matrix is ​​used as supervision information and includes the requirements of different materials for transportation equipment, such as weight, size, transportation method, etc.

[0040] More specifically, a graph convolutional network (GCN) is used to model the graph structure. Each node updates its features by communicating with its neighbors, thereby obtaining global information. In each layer of graph convolution, the node's features are weighted and summed with information from its neighbors, reflecting factors such as material demand and transportation tool conditions in the graph. Graph pooling can help compress the graph size and extract key information, enabling the network to more effectively plan routes. The path planning layer predicts transportation routes based on node features processed by the graph neural network.

[0041] More specifically, the output layer of the network predicts the transportation path from the source to the target based on the characteristics of each node. Path optimization is a process of adjustment based on the back propagation of the graph convolution operation. The network will combine the weight, timeliness, material transportation characteristics and other information of each path to calculate the optimization target of the path (such as the shortest path, minimum transportation time, etc.). The material transportation characteristic matrix is ​​embedded in the loss function as a supervision condition to ensure that the matching degree between material characteristics and transportation equipment is taken into account during the path optimization process. The loss function can be designed to be weighted according to the total transportation time, path reliability, equipment load, etc. of each transportation path. The ultimate goal is to minimize the transportation time and cost of all paths while meeting the material transportation needs. Through the back propagation algorithm, the parameters of the graph neural network model are optimized so that the network can predict the transportation path more accurately.

[0042] More specifically, during training, the graph neural network model is optimized through multiple rounds of iterations, with the model weights continuously adjusted, gradually bringing the transport path of the automated transport unit closer to the optimal path. After each iteration, the model adjusts parameters based on the error between the current predicted path and the actual path (measured by a loss function) and recalculates the next path. After each iteration, the path is updated based on the current optimization results. As training progresses, the path prediction becomes increasingly precise, and the model can adaptively adjust the path to respond to new transportation needs and environmental changes.

[0043] More specifically, the final model will output the expected transportation path for each automatic transport unit. Based on the inference results of the graph neural network model, each path can include information such as the specific location of the path, the nodes passed through, the time required, etc. The output path will take into account the transportation target, the status of the automatic transport unit, the special needs of the materials, and real-time feedback.

[0044] It is understandable that by using the graph neural network model to simulate the path of the transportation target map and iteratively optimize it with the material transportation characteristic matrix as the supervision condition, the graph neural network can effectively learn the material transportation mode and the adaptability of the automatic transport unit from the complex transportation target map, and predict a more accurate transportation path. Compared with traditional path planning methods, the graph neural network can take into account more factors (such as material characteristics, equipment status, real-time traffic conditions, etc.), making path planning more intelligent. The graph neural network has strong adaptive capabilities and can continuously adjust the transportation path according to real-time data (such as the status of the automatic transport unit, production demand, traffic congestion, etc.). Through the backpropagation mechanism, the system can continuously optimize the path planning results in each round of iteration to adapt to complex and dynamic production environments.

[0045] More specifically, through iterative optimization, the transportation path can complete the material transportation task in the shortest time and maximize the utilization efficiency of the automatic transport unit. The introduction of the material transportation characteristic matrix ensures that the most suitable automatic transport unit is used for each transportation task, effectively avoiding resource waste. The material transportation characteristic matrix serves as a supervision condition to ensure that the transportation needs of each material are fully considered, making the selection of transportation tools more reasonable and more efficient. Factors such as material type, weight, and volume will affect the optimization of the transportation path. Therefore, this matrix can ensure that the special needs of material transportation are met.

[0046] More specifically, through the online learning and optimization capabilities of graph neural networks, the system can respond to changes in factory production needs and logistics environment in real time to ensure the efficient execution of transportation tasks. The system no longer relies on manual intervention for path adjustment, but automatically adjusts the path through a deep learning model, greatly improving the system's automation level. Through efficient path planning and material transportation optimization, the system can reduce redundant time and resource waste during transportation, thereby reducing production and transportation costs.

[0047] Specifically, in step S5 of the embodiment provided by the present invention, based on the prediction results of the graph neural network model, the expected transportation paths for each automatic transport unit are generated. These paths include the starting point, destination, and each important node along the path. The automatic transport unit begins to transport materials according to these path instructions. The path information includes data such as travel time, path traversed, and estimated arrival time. After receiving the instructions, the automatic transport unit navigates and transports according to the predetermined transportation path. The real-time status monitoring module records and provides feedback on the unit's current position, speed, load, and other information.

[0048] More specifically, a variety of sensors are deployed on each automatic transport unit, such as lidar, cameras, infrared sensors, inertial measurement units (IMUs), temperature and humidity sensors, etc. These sensors are responsible for collecting multimodal data about the environment, materials, and transportation tool status. The sensor data collection covers the following aspects: Environmental data: including obstacle detection, ambient temperature and humidity, light intensity, etc. Equipment status: including the speed, battery power, load condition, temperature, vibration, etc. of the automatic transport unit; Path monitoring: real-time monitoring of path conditions through lidar or cameras to detect whether there are new obstacles or unfavorable conditions on the path.

[0049] More specifically, dynamic obstacle risk refers to dynamic obstacles (such as pedestrians and other vehicles) detected by sensors that may affect the safety of the transport route. Equipment failure risk includes the risk of ATU failure, such as low battery, sensor failure, and abnormal speed. Traffic condition risk refers to traffic congestion, narrow roads, and traffic signal problems along the transport route. Environmental risk refers to environmental changes such as excessive temperature and humidity and unstable slideways that may affect the stability and safety of the transport vehicle. The system evaluates and calculates potential risk factors in real time based on the perception data of each ATU. Each source of perception data influences the calculation of risk factors. For example: Dynamic obstacle risk: Cameras and lidar are used to detect surrounding obstacles and calculate the distance to them to determine whether there is a collision risk. Low battery: Remaining operating time is calculated based on battery power monitoring data, and the risk of battery depletion to the transport mission is assessed. Path congestion: The path is checked in real time for new obstacles or unsuitable conditions, and the route is calculated to determine whether route adjustments are needed. Risk factors based on perception data are input into a real-time risk assessment model, which outputs possible risk levels (such as high, medium, and low) and assigns different weights to different risk factors.

[0050] More specifically, the driving of the automatic transport unit is affected by multiple constraints, including: physical constraints: such as the maximum load, maximum speed, maximum turning radius of the transport vehicle, environmental constraints: such as the width of the path, turning angle, ground conditions, etc., task constraints: such as material priority, transportation time, etc. These constraints may interact during the transportation process, especially when multiple automatic transport units perform tasks at the same time. The system collects the status information (position, speed, path, etc.) of all automatic transport units and analyzes the mutual influence between different units. During the transportation process, if multiple units are found to have path intersections, traffic congestion, equipment failures, etc., the system will perform constraint analysis in real time to determine whether certain transportation paths need to be adjusted or tasks need to be rescheduled. The interactive constraint analysis not only considers the status of a single automatic transport unit, but also includes the collective behavior, work efficiency, traffic smoothness and other factors of multiple units.

[0051] More specifically, ensure that the transportation path of each automatic transport unit has no potential safety hazards in terms of the environment, obstacles, equipment status, etc., optimize the transportation path while meeting safety and task constraints, reduce time and resource waste, reasonably schedule the automatic transport units, ensure their efficient operation, and avoid idle or overloaded resources. According to the results of the interactive constraint analysis, the system will generate the optimal adjustment strategy, including: path replanning: when obstacles or blocked paths are detected, the system will adjust the path of the automatic transport unit in real time, task scheduling optimization: reschedule tasks according to the real-time status of each unit (such as battery power, speed, load) to avoid unit overload or idleness, collaborative work optimization: when multiple automatic transport units collaborate, adjust the coordination strategy between units to avoid path conflicts and ensure efficient collaborative transportation.

[0052] More specifically, the system will send adjustment strategies to each automatic transport unit and monitor its execution. During the execution process, the automatic transport unit will provide real-time feedback on the current transportation status. If new risks or constraints arise, the system will continue to make dynamic adjustments.

[0053] It is understandable that through real-time multimodal perception and risk calculation, potential safety hazards such as obstacles, insufficient batteries, and poor paths can be discovered in a timely manner, thereby effectively preventing accidents. Real-time interactive constraint analysis and optimal adjustment strategy generation can reduce the time wasted due to obstacles, traffic, etc. during transportation, and improve transportation efficiency. The system has strong adaptive capabilities and can dynamically adjust transportation paths and task scheduling according to the real-time status of the automatic transport unit, environmental changes, path conflicts, etc., to ensure the smooth operation of the system under complex and uncertain conditions. When multiple automatic transport units are in collaborative transportation tasks, the system can optimize their collaborative work through interactive constraint analysis, avoid conflicts, and improve resource utilization.

[0054] More specifically, through real-time calculation of potential risk factors and interactive constraint analysis, system failures can be effectively prevented and the stability and operational reliability of the automatic transport unit can be improved. The system can adjust the transportation strategy in real time based on the status and environmental feedback of the automatic transport unit to ensure that the transportation task can be completed on time and safely.

[0055] The present invention provides a collaborative scheduling method for a digital manufacturing platform, which has the following beneficial effects: The present invention generates a real-time perception matrix through real-time simulation of factory operation conditions, predicts material shortage information based on real-time production work orders and material requirements, obtains positioning data of automatic transport units, constructs a transportation target map and a material transportation characteristic matrix, uses graph neural networks to simulate and optimize transportation routes, drives automatic transport units to carry out transportation, and simultaneously conducts risk assessment through multimodal perception data to generate the optimal adjustment strategy. This method improves the accuracy and efficiency of production scheduling through real-time data drive and route optimization, reduces risks in the transportation process, improves the utilization rate of production resources, and solves the problem of difficulty in quickly responding to changes in material demand in existing technologies.

[0056] Preferably, the step of simulating the factory's operating conditions in real time through the digital manufacturing platform to generate a real-time perception matrix includes: S11: Based on the factory layout information collected in advance, the factory layout is digitally simulated on the digital manufacturing platform to generate a digital twin that reflects the factory layout; S12: Sensor data is collected from the factory through sensor modules pre-deployed at the production site and material storage site to obtain the factory's original monitoring data; S13: performing information parsing on the digital twin according to the original monitoring data to generate dense feature vectors of key areas and sparse feature vectors of edge areas based on the digital twin; wherein the key areas correspond to production equipment in the production site and material storage structures in the material storage site, and the edge areas correspond to passages in the production site and the material storage site; S14: Generate timestamps for the dense feature vectors and sparse feature vectors at each moment, and perform temporal correlation analysis on the dense feature vectors and sparse feature vectors within a specified time window based on the timestamps to generate potential feature vectors, thereby jointly constructing a real-time perception matrix.

[0057] Specifically, through a detailed investigation of the factory, key information including production line layout, equipment location, storage space and channels is collected. This information is usually obtained through architectural drawings, factory site measurements or existing three-dimensional modeling data. This layout information is modeled using a digital manufacturing platform (such as CAD software, BIM platform or 3D modeling tools) to generate a complete factory digital twin. This digital twin is not just a static model, but also needs to be able to update its status according to real-time data, such as equipment status, environmental changes, etc.

[0058] More specifically, various types of sensor modules are deployed in production sites, storage sites, and factory passages. These sensors include temperature and humidity sensors, light sensors, vibration sensors, temperature sensors, pressure sensors, RFID, positioning sensors, etc., to obtain real-time data on production line equipment, storage areas, and material flow processes. The sensors continuously collect raw data from the factory site. The data content includes equipment status (such as on / off, operating speed, load conditions, etc.), environmental parameters (such as temperature, humidity, air quality), the location of personnel and materials, and the inventory status of storage sites.

[0059] More specifically, the collected raw monitoring data is processed and analyzed. This step involves cleaning and filtering the data to remove noise and extract meaningful information. Appropriate algorithms (such as filtering, data standardization, and time series data analysis) are employed to ensure data quality. The digital twin is dynamically updated based on the raw monitoring data. This means that every data change (such as equipment startup, shutdown, or location change) will also update the state of the digital twin, providing real-time insights into the factory's operational status.

[0060] More specifically, by collecting and analyzing data from key areas of production sites and storage sites (such as production equipment, storage structures, etc.), dense feature vectors of these areas are extracted. Dense feature vectors reflect the status information of these key areas, such as equipment operation status, storage area capacity utilization, etc. Dense feature vectors may include equipment operation temperature, vibration intensity, load conditions, production progress, equipment failure frequency, etc. Sparse feature vectors are mainly aimed at the channel areas in production sites and storage sites. The feature vectors of the channel areas are relatively sparse, usually including information such as personnel flow, material flow, and channel congestion. Sparse feature vectors include data such as channel idleness, traffic flow, and personnel density.

[0061] More specifically, each time new dense feature vectors and sparse feature vectors are collected, the system adds a timestamp to these feature vectors to identify the time point when the feature data was generated. Based on the timestamp, the system performs time series correlation analysis on the dense feature vectors and sparse feature vectors within the specified time window. By using time series data analysis algorithms (such as time series regression analysis, sliding window analysis, correlation analysis, etc.), the change patterns of these feature vectors in the time dimension are found. This analysis helps to identify regular problems, bottlenecks or potential risks that may arise in factory operations. For example, production equipment may be under high load operation for a period of time, causing congestion in subsequent production lines.

[0062] More specifically, based on the results of time series correlation analysis, dense and sparse feature vectors are combined to generate potential feature vectors. These potential feature vectors can reveal potential problems in factory operations, such as equipment failure warnings, high load on production lines, channel congestion, etc. These potential feature vectors are combined with the real-time status of the factory to generate a real-time perception matrix. This matrix comprehensively reflects the operation of the factory at all times and can provide accurate decision-making support for production managers. The real-time perception matrix can not only serve as basic data for real-time monitoring, but can also be used to predict and optimize the factory's production scheduling, equipment maintenance plans, etc.

[0063] It is understandable that through real-time simulation of factory operations, it is possible to accurately capture dynamic changes in the production process, monitor the status of production equipment and material flow in real time, and through the generation and analysis of potential feature vectors, it is possible to predict equipment failures, material shortages, production line bottlenecks and other problems in advance, thereby achieving early warning and timely adjustments. Through the construction and real-time updating of digital twins, the actual conditions of each area of ​​the factory can be accurately reflected.

[0064] More specifically, through multi-dimensional and multi-source data collection and analysis (such as sensor data, time series data, factory layout information, etc.), different types of data can be effectively integrated and valuable information can be extracted. The real-time perception matrix can provide support for production scheduling, material allocation, equipment maintenance and other decisions based on real-time data feedback, thereby optimizing the factory's production efficiency and resource utilization. Through digital twins and real-time data analysis, the factory can quickly respond to external changes (such as order changes, equipment failures, etc.), and adjust production plans and work arrangements in a timely manner to enhance the factory's flexibility and adaptability. By generating and analyzing potential feature vectors, the factory can simulate and predict future production operations, further improve the intelligence level of the decision support system, and help managers make more accurate decisions.

[0065] Preferably, the steps of synchronously acquiring a real-time production work order and performing material demand forecasting on the real-time production work order according to the real-time sensing matrix to obtain material shortage information include: S21: Obtain the latest real-time production work order through the work order registration device of the digital manufacturing platform; S22: Analyzing the production order processing performance of each production equipment in the factory production site according to the real-time perception matrix to obtain the production performance adaptability of each production equipment to the real-time production order; S23: Allocating production tasks to the real-time production work order based on the production performance adaptability of each production device to the real-time production work order; S24: Splitting and constructing production events for the assigned production tasks based on the currently allocated materials of each production device fed back by the real-time perception matrix, so as to obtain an information flow of production events expected to be executed by each production device in a future time period; S25: analyzing the consumption trend of allocated materials for each production device in the production event information flow to obtain the expected time when the allocated materials of each production device will be insufficient to meet the real-time production work order; S26: Based on the expected time of material shortage of each production equipment, the material shortage events of the production event information flow are sequentially constructed to obtain the material shortage event flow of each production equipment, and the material shortage event flow of each production equipment is collectively used as the material shortage information.

[0066] Specifically, in the digital manufacturing platform, the work order registration device (such as the work order management system, ERP system or production scheduling system) is responsible for obtaining and storing real-time production work order information. The system obtains the production plan and real-time production work order from the database or work order system based on the latest time (such as the current date and time), including key information such as work order number, production product, production quantity, delivery time, etc.

[0067] More specifically, based on the real-time data of production equipment contained in the live perception matrix, a production work order processing performance analysis is conducted. This analysis requires evaluating the adaptability of each production equipment to the current production work order, such as the equipment type, current status (such as whether it is idle, faulty or running), production capacity (such as production speed, load), etc. For example, if a certain equipment is heavily loaded or is under maintenance, it may not be able to process certain production work orders efficiently; on the contrary, when the equipment is idle and in good condition, it has strong adaptability and can take on more production tasks.

[0068] More specifically, production tasks are assigned to real-time production work orders based on the adaptability analysis results of the production equipment. This step will split each production work order into multiple production tasks and assign these tasks to suitable production equipment. Task allocation needs to consider factors such as the equipment's production capacity, current load, and equipment maintenance status. For example, equipment with a lighter load and normal status will be given priority to be assigned more production tasks.

[0069] More specifically, on each piece of production equipment, an information flow of split events for production tasks is generated based on the currently available allocated materials for the equipment. Each production task involves material consumption, production time, processing technology, etc. The production tasks assigned to each piece of equipment are split into time periods to obtain the production events that each piece of equipment is expected to execute in the future time period. Each event represents a specific production activity that the equipment will perform at a certain point in the future (such as the processing of a certain process, the consumption of a certain material, etc.).

[0070] More specifically, the production event information flow of each production equipment is analyzed to analyze the material consumption trend. The goal of material consumption trend analysis is to predict how much material (such as raw materials, semi-finished products, etc.) each production equipment will consume within a specified time period. This step is usually based on historical data, process requirements and production tasks to estimate material consumption. For example, if the production tasks performed by a certain equipment require a large amount of raw materials, the material consumption of the equipment in the future period is calculated based on the number of these tasks and the required raw materials.

[0071] More specifically, based on the results of the material consumption trend analysis and combined with the factory's real-time material inventory data, the expected time when materials may be out of stock for each production equipment in the future is calculated. The expected time of material shortage refers to the time at a certain point in the future when the required materials will not be able to meet the production needs of a certain equipment when it starts to perform a specific production task.

[0072] More specifically, for the production event information flow of each device, a material shortage event flow is constructed in sequence according to the expected time of material shortage for each device. Each material shortage event contains information such as the type of missing material, the time of shortage, the production tasks involved, and the affected equipment. The material shortage event flow is a prediction and early warning of potential material shortage problems in the production process.

[0073] More specifically, the material shortage event streams of all production equipment are aggregated into final material shortage information. This information includes material shortage warnings for each production equipment, the specific time of shortage, and the type of missing material. This information can help factory managers make material scheduling and replenishment plans in advance to avoid production delays or stagnation.

[0074] It is understandable that real-time acquisition of production work orders and task allocation based on equipment adaptability can help predict the material consumption needs of each device. Through material consumption trend analysis, the material needs of each production equipment in the future time period can be accurately predicted to avoid material shortages. By analyzing the expected time of material loss, potential problems of material shortages can be identified in advance. This can provide data support for production plan adjustments, material procurement or inventory replenishment, thereby effectively preventing production stagnation. Based on real-time data and analysis results, production tasks and material allocation can be dynamically adjusted. Production task allocation not only takes into account the production capacity of the equipment, but also the material needs of the equipment and the current material inventory, thereby optimizing production scheduling.

[0075] More specifically, by accurately predicting and allocating production tasks and materials, production equipment and materials can be used more efficiently, overproduction and material waste can be avoided, thereby improving overall production efficiency and resource utilization. The material shortage event flow provides factory managers with intelligent decision-making support. Material shortage warnings and event flows make production decisions more scientific and accurate. Managers can more effectively respond to sudden material shortages. By identifying material needs and shortages in advance, they can better coordinate relationships with suppliers, replenish or schedule materials in advance, and enhance the flexibility and responsiveness of supply chain management.

[0076] Preferably, the steps of obtaining the positioning data and status data of each automatic transport unit and combining the material shortage information with the real-time perception matrix to construct a transport target map and a corresponding material transport characteristic matrix include: S31: Acquiring positioning data and status data of each automatic transport unit through the positioning function and self-test function of the automatic transport unit; wherein the status data includes power status data and transport status data; S32: Digitally reproducing the road conditions of the factory's production site and storage site based on the real-time perception matrix to generate a basic site map, and constructing corresponding transport simulation units on the basic site map based on the positioning data and status data of each automatic transport unit; S33: Using the real-time sensing matrix as an information guide, constructing a material simulation unit corresponding to the transportable material on the basic site map according to the material shortage information, so as to construct a transportation target map; S34: Analyzing and labeling the transport stability requirements, transport weight burden, and transport volume burden of each material simulation unit on the transport target map based on a pre-built material transport knowledge base, so as to configure material transport characteristics for each material simulation unit on the transport target map; S35: An interactive analysis is performed on the material transport characteristics of each material simulation unit on the transport target map to analyze the feasibility of collaborative transport and the priority of the transport sequence of each material simulation unit, so as to generate corresponding collaborative transport characteristic vectors and transport priority characteristic vectors between each material simulation unit according to the analysis results, thereby obtaining a material transport characteristic matrix.

[0077] Specifically, the location data of each automated transport unit is obtained using the unit's positioning function. This positioning function is typically based on GPS, visual recognition, or other sensor technologies, and updates the geographic location of each automated transport unit in real time. The unit's self-diagnosis function provides data on power and transport status, such as battery level, mission completion, and any malfunctions. This data helps understand the current operating status of each automated transport unit and whether it can continue to perform its mission.

[0078] More specifically, the information in the real-time perception matrix is ​​used to digitally reproduce the road conditions of the factory production site and the storage site. The real-time perception matrix provides data on the road status, such as road congestion, whether there are obstacles, the slope of the road, etc. By analyzing this data, a basic site map is generated, which includes the spatial layout of the road network, traffic conditions, and path information available for the automatic transport units to travel.

[0079] More specifically, based on the positioning data and status data of the automatic transport units, a transport simulation unit is constructed on the basic site map. The transport simulation unit refers to a virtual model representing each automatic transport unit, which contains data such as its current position, motion status, load capacity and power. By simulating the behavior of these automatic transport units, its motion trajectory and transportation capacity in the factory environment can be understood in real time.

[0080] More specifically, based on the material shortage information (including the type, quantity and time of the missing materials), corresponding material simulation units are constructed on the basic site map. The material simulation units represent the location and demand for specific materials to be transported during the production process. These material simulation units contain information such as the type of material, transportation requirements, target location, transportation time, etc., thereby forming a transportation target map.

[0081] More specifically, referring to the pre-built material transportation knowledge base, the transportation requirements of each material simulation unit on the transportation target map are analyzed, focusing on the analysis of aspects such as transportation stability requirements, transportation weight burden and transportation volume burden. Each material simulation unit is marked to indicate the transportation characteristics of the material, such as weight, volume, stability requirements, etc., to provide data support for subsequent transportation scheduling.

[0082] More specifically, based on the transportation characteristics of the material simulation units on the transportation target map, an interactive analysis is performed to evaluate the feasibility of collaborative transportation between materials. For example, issues such as path planning, load balancing, and power distribution for the simultaneous transportation of multiple materials are considered. After considering the feasibility of collaborative transportation of materials, the material simulation units are analyzed for the priority of the transportation sequence. The priority sorting takes into account factors such as the urgency of each material, the efficiency of the transportation route, and the working status of the automatic transportation unit.

[0083] More specifically, based on the above analysis results, a collaborative transportation characteristic vector and a transportation priority characteristic vector are generated for each material simulation unit. The collaborative transportation characteristic vector describes the performance of the material simulation unit in the common transportation task (such as load distribution, path selection, etc.); the transportation priority characteristic vector describes the material transportation urgency, priority and impact on transportation scheduling. These vectors provide a basis for subsequent material transportation optimization.

[0084] More specifically, by analyzing the collaborative transportation characteristic vector and transportation priority characteristic vector of each material simulation unit, a material transportation characteristic matrix is ​​finally generated. This matrix summarizes the transportation characteristics of all material simulation units and provides a basis for further transportation scheduling, resource allocation and path optimization.

[0085] It is understandable that by obtaining the real-time positioning and status data of the automatic transport unit, combined with the factory road conditions and material shortage information, accurate material transportation scheduling can be achieved, which enables the automatic transport unit to efficiently perform transportation tasks, maximize transportation efficiency, and reduce waiting time. Through the digital reproduction of road conditions and the positioning data of the automatic transport unit, the transportation route can be dynamically optimized to avoid road congestion, obstacles and other problems, and improve transportation efficiency and reliability. By analyzing the carrying characteristics of materials (such as weight, volume, stability, etc.), the transportation needs of each material can be accurately analyzed to ensure the safety and stability of materials during transportation. Material collaborative transportation analysis can optimize the transportation scheduling of multiple materials, reasonably allocate transportation resources, improve carrying capacity, and reduce transportation costs. At the same time, according to the priority of the transportation sequence, it ensures that urgent materials are transported first, and improves the stability of material supply on the production line.

[0086] More specifically, this process enables the coordinated work of automated transport units in the production process, enhances the intelligent scheduling capabilities of the production and logistics processes, and enables the automated transport units to flexibly respond based on real-time data and predictive information, reducing human intervention and improving the overall production efficiency of the factory. By analyzing the material transport characteristic matrix, it can efficiently utilize transportation resources, avoid resource waste, and ensure efficient and smooth material transportation, especially in the face of material shortages and demand fluctuations. Optimal scheduling can be achieved, and based on the analysis of material transportation characteristics and priorities, it can provide early warnings of impending material shortages and transportation delays, helping managers to adjust transportation plans in a timely manner and avoid production interruptions.

[0087] Preferably, the steps of simulating the transport path of the transport target map using a graph neural network model, iteratively optimizing the transport path using the material transport characteristic matrix as a supervision condition, and obtaining the expected transport path of each automatic transport unit include: S41: Substituting the transport target map into a pre-trained graph neural network model, and allowing the graph neural network model to allocate transport tasks to the transport target map, so as to assign a corresponding number of material simulation units to each transport simulation unit as objects of the transport task; S42: Based on the transportation task allocation of each transport simulation unit, a transport route is simulated for each transport simulation unit using a graph neural network model, and the transport route simulation is conditionally constrained according to route proximity conditions and route non-conflict conditions to generate a simulated transport path in which each transport simulation unit moves sequentially to a designated location to perform a material loading operation; S43: Evaluate the feasibility of collaborative transportation and the rationality of material transportation sequence of the simulated transportation paths of each transportation simulation unit according to the material transportation characteristic matrix; S44: extracting path nodes from the simulated transport path of each transport simulation unit using the evaluation result as a supervision condition, and analyzing the node advantage factors of the extracted path nodes to generate the node advantage factors of each path node; S45: constructing a dominant node topology network based on the node dominance factor of each path node in each of the simulated transportation paths, and performing a feasibility value assessment on each material simulation unit in the transportation target map as a transportation task object of the same transportation simulation unit based on the dominant node topology network; S46: Based on the feasibility value assessment results, the transport task objects of each transport simulation unit are adjusted, and the transport routes are adjusted accordingly to optimize the simulated transport path. S47: Repeat the evaluation and optimization of the simulated transport path until the expected transport path of each automatic transport unit that meets the preset standard is obtained.

[0088] Specifically, the transportation target graph is substituted into the trained graph neural network (GNN) model as input data. The graph neural network model can process graph structured data, where each node represents a material simulation unit, and the edges between nodes represent the relationship between transportation paths or tasks. In this step, the model first assigns tasks to the transportation target graph, associates each transport simulation unit with several material simulation units, and assigns material transportation tasks. For example, a transport unit is assigned to the path from factory A to warehouse B, responsible for transporting a specific type of material.

[0089] More specifically, based on the calculation of the graph neural network model, the transportation route of each transport simulation unit is simulated. The network predicts the path of each transport unit through the structure of the graph (such as the connection relationship between nodes and edges). In this process, the model needs to take into account the proximity conditions (i.e., the distance of the path, traffic flow, congestion, etc.) and non-conflict conditions (i.e., avoiding multiple transport units occupying the same path at the same time). These constraints are used to optimize path planning, avoid collisions or path conflicts, ensure the coordinated operation of multiple vehicles, and finally generate a simulated transportation path in which each transport simulation unit moves to the designated location in turn and performs material loading operations.

[0090] More specifically, the simulated transport routes are evaluated using a material transport characteristic matrix, which includes the transport requirements of each material, such as weight, volume, and stability. The graph neural network uses this information to determine whether each transport route meets the requirements of the transport task. The evaluation focuses on analyzing whether the transport sequence of each transport simulation unit is reasonable and whether the material priority, stability requirements, and transport load are taken into account.

[0091] More specifically, based on the evaluation results, the path nodes in each simulated transportation path are extracted, and the node advantage factor of each path node is analyzed. The node advantage factor represents the importance or superiority of the node in the path, and usually includes factors such as the node's transportation capacity, load requirements, and path congestion. The analysis of the node advantage factor helps identify the optimal path node and ensure the accuracy and efficiency of path planning.

[0092] More specifically, the extracted advantageous nodes and the relationships between them are constructed into a topological network, which represents the key nodes in the transportation task and their connection relationships. The topological structure of the advantageous nodes helps to optimize path planning and enhance the intelligence of path selection. Through the advantageous node topological network, the priority of each node, the feasibility of path selection, and the key decision points in the transportation task can be more clearly displayed.

[0093] More specifically, based on the advantageous node topology network, a feasibility value assessment is conducted on each material simulation unit in the transportation target map, that is, whether the material simulation unit is suitable for jointly undertaking the transportation task with the current carrying simulation unit. The feasibility value assessment not only considers the node advantages, but also takes into account factors such as the time requirements, load capacity and path conflicts of the transportation task to ensure that the evaluation results are scientific and reasonable.

[0094] More specifically, based on the feasibility assessment results, the transportation task object of each transport simulation unit is adjusted, which means that the material simulation units may be reallocated to ensure the rationality and efficiency of the transportation task. The adjustment of the transportation route is also based on the results of the feasibility assessment. By adjusting the path, conflicts can be avoided, delays can be reduced, and the efficiency of the entire transportation process can be improved.

[0095] More specifically, the adjusted transport path is re-input into the graph neural network model for re-evaluation and optimization. Through a continuous iterative process, the model will gradually adjust the path and improve the transport sequence until all transport paths meet the preset optimization criteria. The optimization goals include reducing transport time, improving transport efficiency, avoiding path conflicts, and optimizing transport sequence, etc., to ensure that the transport path of each transport simulation unit is optimal.

[0096] More specifically, after several rounds of evaluation and optimization, the expected transportation paths for each automated transport unit are finally obtained. These paths can achieve efficient and coordinated transportation of materials and meet the requirements of the material transportation characteristic matrix for weight, volume, stability, etc.

[0097] It is understandable that graph neural networks can conduct deep learning based on the graph structure of the transportation target map, intelligently plan transportation routes, and take into account multiple factors such as route congestion, load, stability, etc., to generate the optimal transportation route. By combining the material transportation characteristic matrix with the graph neural network model, accurate transportation task allocation can be achieved to ensure that each transportation simulation unit performs tasks at the appropriate time and place, reduce resource waste, and improve transportation efficiency. By evaluating the feasibility of collaborative transportation and the rationality of transportation sequence, it ensures that multiple transportation units can collaborate efficiently. Path adjustment and task redistribution can avoid conflicts, balance loads, and improve overall system performance.

[0098] More specifically, this method can continuously adjust path planning according to real-time data and feedback through continuous iteration and optimization processes. It has strong adaptability and can make flexible adjustments according to different transportation tasks and real-time environments. Through the analysis of node advantage factors, it can effectively identify the optimal path nodes, improve the intelligence level of path selection, and ensure that the decision of each step in the transportation process is optimal. Based on the feasibility value assessment of the material simulation unit, it can achieve accurate task scheduling, ensure that each material can be transported on time, safely and with priority, and meet the material supply needs in the production process. The comprehensively optimized transportation path greatly improves the completion efficiency of transportation tasks, reduces the waste of transportation time and transportation resources, and provides a more efficient material distribution service for the production line.

[0099] Preferably, the steps of driving each automatic transport unit to transport materials according to the expected transport path, synchronously collecting multimodal perception data of each automatic transport unit, and performing interactive constraint analysis to calculate potential risk factors in real time to obtain the optimal adjustment strategy include: S51: driving the automatic transport unit to go to designated locations of the material storage site in sequence according to the expected transport path, automatically loading the materials at the designated locations, and stacking them on the automatic transport unit in a designated order; S52: collecting data on the automatic transport unit's own state and external environment through a sensor group pre-installed on the automatic transport unit to obtain multimodal perception data; S53: Analyzing the transport stability risk and environmental collision risk of the automatic transport units based on the multimodal sensing data to generate potential risk factors for the transport stability risk and environmental collision risk brought about by the current moving speed of each automatic transport unit; S54: performing a path conflict risk assessment on each automated transport unit based on the expected transport path and the current moving speed of each automated transport unit to generate a potential risk factor for path conflict caused by the current moving speed of each automated transport unit; S55: Based on the potential risk factors of the current moving speed of each automated transport unit in terms of transport stability risk, environmental collision risk, and path conflict, a four-dimensional tensor analysis of the spatiotemporal dimension, the spatial dimension, the transport unit dimension, and the risk type dimension is performed to construct a risk factor tensor matrix; S56: Perform risk cost constraints and efficiency cost constraints on the speed adjustment of each automatic transport unit according to the risk factor tensor matrix, so as to map the speed adjustment trial strategy of each automatic transport unit to the collaborative strategy space, and perform collaborative analysis on the speed adjustment trial strategy of each automatic transport unit based on the collaborative strategy space to obtain the optimal adjustment strategy of each automatic transport unit.

[0100] Specifically, according to the expected transportation route, the automatic transport unit goes to various designated locations in the storage site in turn. These locations are automatically assigned by the system according to the preset route and task requirements. The automatic transport unit arrives at each location in the specified order and automatically loads the materials. When it arrives at the designated location, the automatic transport unit will automatically load the materials and stack them on it in the specified order.

[0101] More specifically, the automatic transport unit is equipped with multiple sensors (such as temperature, humidity, acceleration, camera, radar, LIDAR, etc.) to perceive its own status and changes in the surrounding environment in real time. The data collected by these sensors form multimodal perception data, including dynamic and static data from different sources, such as motion data, environmental data, etc.

[0102] More specifically, based on the collected multimodal perception data, the automatic transport unit will conduct its own transport stability risk analysis. For example, the data may include factors such as vibration, tilt, and center of gravity changes of materials during transportation, which affect the stability of transportation. At the same time, environmental collision risks will also be analyzed. Sensor data can monitor potential collision risks such as surrounding obstacles, other automatic transport units, and pedestrians in real time, and evaluate the risk factors brought by the current moving speed.

[0103] More specifically, during the transportation process, a path conflict risk assessment is performed based on the expected path and current moving speed of the automatic transport unit. Multiple automatic transport units may travel on different paths, and it is assessed whether the current path will conflict with other paths. The path conflict analysis not only considers path intersections, but also factors such as transportation time and route occupancy, to generate a potential risk factor for path conflict for each automatic transport unit.

[0104] More specifically, all relevant risk factors (such as transportation stability risk, environmental collision risk, and path conflict risk) are analyzed in multiple dimensions and decomposed from the time-space dimension, spatial dimension, transportation unit dimension, and risk type dimension using a four-dimensional tensor representation. Through four-dimensional tensor analysis, all risk factors are combined together to form a complete risk factor tensor matrix, which facilitates subsequent risk assessment and adjustment strategy optimization.

[0105] More specifically, based on the risk factor tensor matrix, risk cost constraints and efficiency cost constraints are imposed on the speed adjustment of each automatic transport unit. The risk cost constraint ensures that the automatic transport unit avoids high-risk behaviors (such as collisions or instability caused by moving too fast), while the efficiency cost constraint ensures that speed adjustment does not significantly affect transportation efficiency. These constraints help balance risk and efficiency and avoid increased risk caused by overly aggressive speed adjustments.

[0106] More specifically, by mapping the speed adjustment trial strategy of each automatic transport unit to the collaborative strategy space, the collaborative relationship between multiple transport units can be comprehensively considered. The collaborative strategy space represents the collaborative effect when multiple transport units perform tasks together. The collaborative analysis method is used to coordinate and optimize the speed adjustment strategies between each transport unit to ensure that the automatic transport units in the system can effectively cooperate and avoid collisions, conflicts and other problems.

[0107] More specifically, based on the analysis of the collaborative strategy space described above, an optimal adjustment strategy is ultimately generated for each automated transport unit. This strategy takes into account each unit's risk factors, mission requirements, and collaborative goals, and dynamically adjusts the speed of each transport unit to achieve an efficient and safe transportation process.

[0108] It is understood that this method can collect multimodal perception data in real time and dynamically assess potential risks during transportation (such as stability risk, environmental collision risk, and path conflict risk). Through the input and analysis of real-time data, the system can make timely adjustments to reduce the likelihood of risks. Through a comprehensive analysis of transport stability risk and environmental collision risk, the system can optimize the driving path and speed of the automated transport units, reducing transportation accidents caused by instability or collisions and improving transportation safety and stability. Through path conflict risk assessment and speed adjustment, the system can effectively avoid path conflicts between multiple automated transport units, ensuring that each transport unit can coordinate its operations and avoiding efficiency losses or equipment damage caused by conflicts.

[0109] More specifically, by constructing a collaborative strategy space and conducting collaborative analysis of speed adjustment strategies based on this space, the system can ensure more efficient cooperation between multiple automated transport units, avoid unnecessary delays or ineffective routes, and improve overall transportation efficiency. By combining risk and efficiency cost constraints, the system can maximize transportation efficiency while ensuring transportation safety. This strategy allows speed adjustments to avoid high risks without excessively reducing transportation efficiency. Ultimately, through optimized speed adjustment strategies, each automated transport unit can automatically adjust its speed and driving strategy based on the current mission, environment, and route conditions, achieving intelligent collaborative decision-making. This optimized decision-making can better adapt to complex transportation environments and enhance the flexibility and adaptability of automated transportation systems.

[0110] Reference Figure 2 As shown, in a second aspect, the present invention provides a collaborative scheduling device for a digital manufacturing platform, which is used to implement the collaborative scheduling method for a digital manufacturing platform described in any one of the first aspects, including: A real-time perception module, which simulates the factory's operating conditions in real time through a digital manufacturing platform to generate a real-time perception matrix; A demand forecasting module is used to synchronously obtain real-time production work orders and perform material demand forecasting on the real-time production work orders based on the real-time sensing matrix to obtain material shortage information; A target analysis module is used to obtain the positioning data and status data of each automatic transport unit, and to construct a transport target map and a corresponding material transport characteristic matrix by combining the material shortage information with the real-time perception matrix; A path analysis module is used to simulate the transportation path of the transportation target map through a graph neural network model, and iteratively optimize the transportation path using the material transportation characteristic matrix as a supervision condition to obtain the expected transportation path of each automatic transportation unit; The strategy adjustment module is used to drive each automatic transport unit to transport materials according to the expected transportation path, and simultaneously collect multimodal perception data of each automatic transport unit to calculate potential risk factors in real time for interactive constraint analysis to obtain the optimal adjustment strategy.

[0111] In this embodiment, for the specific implementation of each module in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0112] In a third aspect, the present invention provides a collaborative scheduling device for a digital manufacturing platform, comprising a memory and a processor, wherein the memory stores a collaborative scheduling program for the digital manufacturing platform that can run on the processor, and when the processor executes the collaborative scheduling program for the digital manufacturing platform, it implements a collaborative scheduling method for a digital manufacturing platform as described in any one of the first aspects.

[0113] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A collaborative scheduling method for a digital manufacturing platform, characterized in that: include: Use the digital manufacturing platform to simulate the factory's operating conditions in real time to generate a real-time perception matrix; Synchronously obtain real-time production work orders, and perform material demand forecasting on the real-time production work orders based on the real-time sensing matrix to obtain material shortage information; Obtaining the positioning data and status data of each automatic transport unit, and combining the material shortage information with the real-time perception matrix to construct a transport target map and a corresponding material transport characteristic matrix; The transport path of the transport target map is simulated by a graph neural network model, and the transport path is iteratively optimized using the material transport characteristic matrix as a supervision condition to obtain the expected transport path of each automatic transport unit; According to the expected transportation path, each automatic transport unit is driven to transport materials, and multimodal perception data of each automatic transport unit is simultaneously collected to calculate potential risk factors in real time for interactive constraint analysis to obtain the optimal adjustment strategy.

2. The collaborative scheduling method for a digital manufacturing platform according to claim 1, characterized in that: The steps for generating a real-time perception matrix by simulating factory operations in real time through a digital manufacturing platform include: Based on the factory layout information collected in advance, the factory layout is digitally simulated on the digital manufacturing platform to generate a digital twin that reflects the factory layout. The factory collects sensor data through sensor modules pre-deployed at the production site and material storage site to obtain the factory's original monitoring data; Performing information parsing on the digital twin based on the raw monitoring data to generate dense feature vectors of key areas and sparse feature vectors of edge areas based on the digital twin; wherein the key areas correspond to production equipment in the production site and material storage structures in the material storage site, and the edge areas correspond to passages in the production site and the material storage site; Generate timestamps for the dense feature vectors and sparse feature vectors at each moment, and perform temporal correlation analysis on the dense feature vectors and sparse feature vectors within the specified time window based on the timestamps to generate potential feature vectors, thereby jointly constructing the real-time perception matrix.

3. The collaborative scheduling method for a digital manufacturing platform according to claim 1, wherein: The steps of synchronously acquiring a real-time production work order and performing material demand forecasting on the real-time production work order according to the real-time sensing matrix to obtain material shortage information include: Obtain the latest real-time production work orders through the work order registration device of the digital manufacturing platform; Performing a production order processing performance analysis on each piece of production equipment in the factory production site based on the real-time perception matrix to obtain the production performance adaptability of each piece of production equipment to the real-time production order; Allocating production tasks to the real-time production work order based on the production performance adaptability of each production equipment to the real-time production work order; Splitting and constructing production events for the assigned production tasks based on the current allocated materials of each production equipment fed back by the real-time perception matrix, so as to obtain the production event information flow expected to be executed by each production equipment in the future time period; Analyze the consumption trend of allocated materials for each production device's production event information flow to obtain the expected time when the allocated materials of each production device will be insufficient to meet the real-time production work order; Based on the expected time of material shortage of each production equipment, the material shortage events of the production event information flow are sequentially constructed to obtain the material shortage event flow of each production equipment, and the material shortage event flow of each production equipment is collectively used as the material shortage information.

4. The collaborative scheduling method for a digital manufacturing platform according to claim 1, wherein: The steps of obtaining the positioning data and status data of each automatic transport unit and combining the material shortage information with the real-time perception matrix to construct a transport target map and a corresponding material transport characteristic matrix include: Acquiring positioning data and status data of each automatic transport unit through the positioning function and self-test function of the automatic transport unit; wherein the status data includes power status data and transport status data; Based on the real-time perception matrix, the road conditions of the factory's production site and storage site are digitally reproduced to generate a basic site map, and corresponding transport simulation units are constructed on the basic site map according to the positioning data and status data of each automatic transport unit; Using the real-time sensing matrix as information guidance, constructing material simulation units corresponding to transportable materials on the basic site map according to the material shortage information, so as to construct a transportation target map; Analyzing and labeling the transport stability requirements, transport weight burden, and transport volume burden of each material simulation unit on the transport target map based on a pre-built material transport knowledge base, so as to configure material transport characteristics for each material simulation unit on the transport target map; An interactive analysis is performed on the material transport characteristics of each material simulation unit on the transport target map to analyze the feasibility of collaborative transport and the priority of the transport sequence of each material simulation unit, so as to generate corresponding collaborative transport characteristic vectors and transport priority characteristic vectors between each material simulation unit according to the analysis results, thereby obtaining a material transport characteristic matrix.

5. The collaborative scheduling method for a digital manufacturing platform according to claim 4, characterized in that: The steps of simulating the transport path of the transport target map using a graph neural network model, iteratively optimizing the transport path using the material transport characteristic matrix as a supervision condition, and obtaining the expected transport path of each automatic transport unit include: Substituting the transport target map into a pre-trained graph neural network model, and allowing the graph neural network model to allocate transport tasks to the transport target map, so as to assign corresponding material simulation units to each transport simulation unit as objects of the transport task; Based on the transportation task allocation of each transport simulation unit, the transport route of each transport simulation unit is simulated through the graph neural network model. The simulation of the transport route is conditionally constrained according to the route proximity condition and the route non-conflict condition to generate a simulated transport path in which each transport simulation unit moves to the designated location in sequence to perform the material loading operation; Evaluate the feasibility of collaborative transportation and the rationality of material transportation sequence on the simulated transportation path of each transportation simulation unit according to the material transportation characteristic matrix; The evaluation results are used as supervision conditions to extract the path nodes of the simulated transport path of each carrier simulation unit, and the node advantage factors of the extracted path nodes are analyzed to generate the node advantage factors of each path node; Constructing a dominant node topology network according to the node dominance factor of each path node in each of the simulated transportation paths, and performing a feasibility value assessment on each material simulation unit in the transportation target map as a transportation task object of the same transport simulation unit based on the dominant node topology network; Based on the feasibility value assessment results, the transport task objects of each transport simulation unit are adjusted, and the transport routes are adjusted accordingly to achieve the optimization of the simulated transport path. The evaluation and optimization of the simulated transport path is repeated until the expected transport path of each automatic transport unit that meets the preset standards is obtained.

6. The collaborative scheduling method for a digital manufacturing platform according to claim 1, wherein: The steps of driving each automatic transport unit to transport materials according to the expected transport path, synchronously collecting multimodal perception data of each automatic transport unit, and calculating potential risk factors in real time to perform interactive constraint analysis to obtain the optimal adjustment strategy include: Driving the automatic transport unit to the designated locations of the storage site in sequence according to the expected transport path, automatically loading the materials at the designated locations, and stacking them on the automatic transport unit in the designated order; The sensor group pre-installed on the automatic transport unit collects data on its own status and external environment to obtain multimodal perception data; Analyzing the transport stability risk and environmental collision risk of the automatic transport units based on the multimodal sensing data to generate potential risk factors for the transport stability risk and environmental collision risk brought about by the current moving speed of each automatic transport unit; Performing a path conflict risk assessment on each automated transport unit based on its expected transport path and current moving speed to generate a potential risk factor for path conflict caused by the current moving speed of each automated transport unit; Based on the potential risk factors of each automated transport unit's current moving speed in terms of transport stability risk, environmental collision risk, and path conflict, a four-dimensional tensor analysis of the spatiotemporal dimension, spatial dimension, transport unit dimension, and risk type dimension is performed to construct a risk factor tensor matrix. According to the risk factor tensor matrix, the risk cost constraints and efficiency cost constraints of the speed adjustment of each automatic transport unit are imposed, so as to map the speed adjustment trial strategy of each automatic transport unit to the collaborative strategy space. Based on the collaborative strategy space, the speed adjustment trial strategy of each automatic transport unit is collaboratively analyzed to obtain the optimal adjustment strategy of each automatic transport unit.

7. A collaborative scheduling device for a digital manufacturing platform, characterized in that: A collaborative scheduling method for a digital manufacturing platform according to any one of claims 1 to 6, comprising: A real-time perception module, which simulates the factory's operating conditions in real time through a digital manufacturing platform to generate a real-time perception matrix; A demand forecasting module is used to synchronously obtain real-time production work orders and perform material demand forecasting on the real-time production work orders based on the real-time sensing matrix to obtain material shortage information; A target analysis module is used to obtain the positioning data and status data of each automatic transport unit, and to construct a transport target map and a corresponding material transport characteristic matrix by combining the material shortage information with the real-time perception matrix; A path analysis module is used to simulate the transportation path of the transportation target map through a graph neural network model, and iteratively optimize the transportation path using the material transportation characteristic matrix as a supervision condition to obtain the expected transportation path of each automatic transportation unit; The strategy adjustment module is used to drive each automatic transport unit to transport materials according to the expected transportation path, and simultaneously collect multimodal perception data of each automatic transport unit to calculate potential risk factors in real time for interactive constraint analysis to obtain the optimal adjustment strategy.

8. A collaborative scheduling device for a digital manufacturing platform, comprising a memory and a processor, wherein the memory stores a collaborative scheduling program for the digital manufacturing platform that can be run on the processor, characterized in that: When the processor executes the collaborative scheduling program of the digital manufacturing platform, the collaborative scheduling method of the digital manufacturing platform according to any one of claims 1 to 6 is implemented.

Images