Intelligent material scheduling method and system for bench assembly line

Abstract

The application discloses a material intelligent scheduling method and system for a desktop assembly pipeline, belongs to the technical field of material scheduling, and comprises a data acquisition module, a scheduling decision module, an execution control module and a monitoring and early warning module, further comprises a digital twin platform and an edge computing data fusion module; the data acquisition module is used for synchronously collecting real-time operation data of production line each station production rhythm, material position and state, AGV running track and equipment operation parameter; the edge computing data fusion module is in communication connection with the data acquisition module, the scheduling decision module comprises the digital twin platform and an intelligent scheduling engine; the execution control module is in communication connection with the scheduling decision module, the monitoring and early warning module is in communication connection with the digital twin platform and the intelligent scheduling engine respectively, the monitoring and early warning module is linked with the digital twin platform, the global operation state of the production line is displayed through a three-dimensional visual interface, production abnormalities are automatically detected and a closed-loop rescheduling mechanism is triggered.

Description

Technical Field

[0001] This invention discloses a method and system for intelligent material scheduling in desktop computer assembly lines, belonging to the field of material scheduling technology. Background Technology

[0002] Desktop PC assembly is a typical example of a multi-variety, small-batch discrete manufacturing model. A single production line typically produces multiple models with different configurations simultaneously. Each machine contains multiple core materials, resulting in a complex BOM hierarchy and significant differences in material specifications. With the explosive growth of personalized demands in the consumer electronics market, the trend of order fragmentation has intensified, and production plans have changed frequently. Traditional static batch material preparation and scheduling methods based on manual experience have shortcomings such as inability to respond to real-time fluctuations in the production line, severe information silos, delayed response to anomalies, and frequent path conflicts among multiple AGVs operating in parallel. This leads to both production line shutdowns due to material shortages and material accumulation, resulting in low material management efficiency.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a method and system for intelligent material scheduling in desktop computer assembly lines, which at least solves the technical problems of traditional desktop computer assembly lines relying on manual experience and adopting a static batch material preparation mode, which cannot adapt to dynamic production scenarios with multiple varieties, small batches, and fragmented orders, resulting in material supply and demand mismatch and low production line utilization.

[0005] According to one aspect of the present invention, in order to achieve the above-mentioned objective, a material intelligent scheduling system for a desktop computer assembly line is provided, including a data acquisition module, a scheduling decision module, an execution control module, and a monitoring and early warning module, and further including a digital twin platform and an edge computing data fusion module;

[0006] The data acquisition module includes an RFID tag reader, a UWB positioning device, a visual recognition camera, and a sensor array, which are used to synchronously collect real-time data on production cycle time, material status, AGV running trajectory, and equipment operating parameters at production line workstations.

[0007] The edge computing data fusion module communicates with the data acquisition module, performs noise filtering and feature extraction preprocessing on the raw data, and transmits it to the data platform through the 5G industrial private network. It connects with ERP, MES, and WMS systems to obtain business data and completes privacy-preserving fusion of multi-source heterogeneous data based on federated learning technology.

[0008] The scheduling decision module includes a digital twin platform and an intelligent scheduling engine. The digital twin platform constructs a three-dimensional model of all elements of the production line and materials, binds a unique UHF RFID tag to each material and creates a digital twin, and establishes a digital archive of the entire life cycle of materials and a millisecond-level virtual-real bidirectional mapping between the physical and digital worlds. The intelligent scheduling engine has built-in a material demand prediction model based on deep Q-networks and a multi-AGV collaborative path planning algorithm based on time windows.

[0009] The execution control module is connected to the scheduling decision module, including the AGV scheduling system, electronic picking system and workstation terminal, and is used to convert scheduling decisions into equipment control commands;

[0010] The monitoring and early warning module communicates with the digital twin platform and the intelligent scheduling engine respectively. It works in conjunction with the digital twin platform to display the production line operation status, automatically detects production anomalies, and triggers a closed-loop rescheduling mechanism.

[0011] Furthermore, the material demand forecasting model of the intelligent scheduling engine adopts a training method that combines offline pre-training and online incremental learning. The comprehensive optimization objectives are to maximize production line utilization, minimize material inventory costs, and minimize the total AGV delivery distance. The material demand forecasting accuracy of the model is greater than or equal to 95%.

[0012] Furthermore, the execution steps of the time window-based multi-AGV cooperative path planning algorithm are as follows: First, an initial optimal path is generated for each AGV. Then, based on the travel speed and task priority of each AGV, a unique time window is assigned to each segment on the path. Only one AGV is allowed to pass through the same segment at the same time. When an AGV failure, task change, or path blockage occurs, the time window-based multi-AGV cooperative path planning algorithm recalculates the paths and time windows of all affected AGVs.

[0013] Furthermore, the digital twin platform has simulation and deduction capabilities. It can simulate the production line operation status under different production plans, different material supply rhythms, and different abnormal scenarios, and output optimized production plans and scheduling schemes.

[0014] Furthermore, the monitoring and early warning module automatically identifies five types of anomalies: material shortage, incorrect or missing materials, AGV malfunction, equipment downtime, and order changes. The monitoring and early warning module classifies and grades these anomalies into three levels based on their type and severity. For minor anomalies with delays of ≤5 minutes, the monitoring and early warning module automatically adjusts the timing of subsequent delivery tasks and reallocates AGV transport resources. For serious anomalies such as shortages of critical materials, the monitoring and early warning module immediately issues warnings through multiple channels, including sound and light, SMS, and the APP, and generates at least three alternative scheduling plans.

[0015] Furthermore, the AGV scheduling system monitors the battery level, location, operating status, and task execution progress of all AGVs in real time. When the remaining battery level of an AGV is less than 20%, the AGV scheduling system automatically dispatches the AGV to the nearest charging area for charging. After charging is completed, the AGV automatically returns to the task queue and takes on high-priority scheduling tasks.

[0016] According to one embodiment of the present invention, a method for intelligent material scheduling in a desktop computer assembly line is also provided, for controlling the above-mentioned intelligent material scheduling system, the method comprising the following steps:

[0017] Step S1: Construct a three-dimensional digital twin model of all elements of the production line and materials, mapping the geometry, positional relationships and operating logic of the assembly line workstations, AGV travel paths, material temporary storage areas, automated warehouses and automatic loading and unloading mechanisms, binding a unique UHF RFID electronic tag to each material unit and creating a corresponding digital twin, establishing a digital archive of the entire material lifecycle, and establishing a millisecond-level virtual-physical bidirectional mapping relationship between the physical production line and the three-dimensional digital twin model;

[0018] Step S2: Collect real-time status data of the production line through multi-source sensing devices, perform noise filtering and feature extraction preprocessing on the collected raw data, and integrate the business data of the ERP system, MES system and WMS system to form a unified global status view of the production line.

[0019] Step S3: Construct a material demand forecasting model based on a deep Q-network, using historical production data, current production plan, material inventory level at each workstation, equipment failure rate, order change probability, and material delivery time as input features, and output the material demand for each workstation in the next 15 minutes, 30 minutes, and 60 minutes.

[0020] Step S4: Generate material delivery tasks based on the material demand forecasting results output by the material demand forecasting model. Use a priority scheduling mechanism to allocate higher scheduling priority to urgent orders, material shortage workstations, and abnormal materials. Use a time window-based AGV path planning algorithm to assign a unique path time window to each AGV.

[0021] Step S5: The scheduling instructions are sent to the execution control module for collaborative execution. Data on the scheduling execution process is collected in real time and fed back to the three-dimensional digital twin model and intelligent scheduling engine. The scheduling execution status is evaluated in real time. If an anomaly is detected, the rescheduling mechanism is triggered to generate the optimal scheduling scheme.

[0022] Furthermore, in step S3, the offline pre-training of the material demand forecasting model uses historical production data of more than or equal to 6 months, and the online incremental learning of the material demand forecasting model is performed every 15 minutes to iteratively optimize the network parameters of the material demand forecasting model based on real-time production line operation data.

[0023] Furthermore, in step S4, a dynamic batch delivery strategy is adopted, which dynamically adjusts the single delivery batch and delivery cycle according to the real-time material consumption rate of each workstation, and the material quantity of a single delivery is less than or equal to the material consumption of the corresponding workstation in 2 hours.

[0024] Furthermore, in step S5, an anomaly is detected by comparing the real-time collected data with the preset normal operation threshold. The response time of the rescheduling mechanism is less than or equal to 1 minute. After the rescheduling scheme is generated, it is immediately sent to the execution control module for execution.

[0025] In this embodiment of the invention, by organically combining digital twin full-element virtual-real mapping, edge computing federated data fusion, deep Q-network multi-time-dimensional material demand prediction, and time window-based multi-AGV collaborative scheduling, the transformation of material scheduling from passive response to proactive prediction is realized. This significantly improves production line utilization, greatly reduces material management error rate, significantly improves AGV equipment utilization, significantly shortens production anomaly response time, effectively reduces material management labor costs, and comprehensively improves the automation level of material management and production operation stability of desktop assembly lines. This solves the technical problem that traditional desktop assembly lines rely on manual experience and adopt a static batch material preparation mode for material scheduling, which cannot adapt to dynamic production scenarios with multiple varieties, small batches, and fragmented orders, resulting in material supply and demand mismatch and low production line utilization. Attached Figure Description

[0026] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0027] Figure 1 This is a structural block diagram of a material intelligent scheduling system for a desktop computer assembly line according to one embodiment of the present invention.

[0028] Figure 2 This is a flowchart of a material intelligent scheduling method for a desktop computer assembly line according to one embodiment of the present invention. Detailed Implementation

[0029] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] According to an embodiment of the present invention, a method for intelligent material scheduling for a desktop computer assembly line is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0032] This method embodiment can be executed in an electronic device or similar computing device that includes a memory and a processor. Taking operation on a vehicle terminal as an example, the vehicle terminal may include one or more processors (processors may include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), digital signal processing (DSP) chips, microcontroller units (MCUs), field-programmable gate arrays (FPGAs), neural network processors (NPUs), tensor processors (TPUs), artificial intelligence (AI) type processors, etc.) and a memory for storing data. Optionally, the vehicle terminal may also include transmission devices, input / output devices, and display devices for communication functions. Those skilled in the art will understand that the above structural description is merely illustrative and does not limit the structure of the vehicle terminal. For example, the vehicle terminal may include more or fewer components than described above, or have a different configuration than described above.

[0033] The memory can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the intelligent material scheduling method for desktop computer assembly lines in this embodiment of the invention. The processor executes various functional applications and data processing by running the computer program stored in the memory, thereby realizing the aforementioned intelligent material scheduling method for desktop computer assembly lines. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to mobile terminals via a network. Examples of such networks include, but are not limited to, the Internet, enterprise intranets, local area networks, mobile communication networks, and combinations thereof.

[0034] The transmission device is used to receive or send data via a network. Specific examples of the network mentioned above may include a wireless network provided by the mobile terminal's communication provider. In one example, the transmission device includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0035] Display devices can be, for example, touchscreen liquid crystal displays (LCDs) and touch displays (also referred to as "touchscreens" or "touch displays"). The LCD allows users to interact with the user interface of the mobile terminal. In some embodiments, the mobile terminal has a graphical user interface (GUI), which allows users to interact with the GUI through finger contact and / or gestures on a touch-sensitive surface. Optional human-computer interaction functions include: creating web pages, drawing, word processing, creating electronic documents, playing games, video conferencing, instant messaging, sending and receiving emails, call interfaces, playing digital video, playing digital music, and / or web browsing, etc. Executable instructions for performing the above human-computer interaction functions are configured / stored in one or more processor-executable computer program products or readable storage media.

[0036] Figure 1 This is a structural block diagram of a material intelligent scheduling system for a desktop computer assembly line according to one embodiment of the present invention, such as... Figure 1As shown, the overall design adopts a modular distributed approach. Each module achieves data exchange and collaborative work through a standardized industrial communication interface. Through the organic combination of full-element perception, intelligent decision-making, and closed-loop execution, it realizes intelligent management and control of the entire lifecycle of materials in the desktop computer assembly line.

[0037] The data acquisition module provides fundamental data support for the system and is deployed at key nodes throughout the production line. It consists of RFID tag readers, UWB positioning devices, visual recognition cameras, and sensor arrays. The RFID tag readers use ultra-high frequency (UHF) reading and writing equipment to read information from electronic tags bound to material units, obtaining basic attribute data such as material code, batch, quantity, and supplier. The UWB positioning devices communicate with tags mounted on AGVs and material handling vehicles via base stations deployed on the production line ceiling, acquiring real-time location and movement status data of mobile devices. The visual recognition cameras use industrial-grade high-definition equipment to collect real-time data on material inventory, material placement, and equipment operation at each workstation, automatically identifying anomalies such as material accumulation and misplacement through image recognition algorithms. The sensor array, including temperature sensors, vibration sensors, and current sensors, is installed on key production line equipment to collect physical parameters during equipment operation for early prediction of equipment failures. The data acquisition module synchronously collects all of the above data at a millisecond-level sampling frequency, ensuring the real-time nature and completeness of production line status data.

[0038] The edge computing data fusion module is deployed on the production line and connects to the data acquisition module via industrial Ethernet, undertaking the functions of data preprocessing and multi-source data fusion. This module first preprocesses the raw data uploaded by the data acquisition module, including noise filtering, outlier removal, key feature extraction, and data standardization, significantly reducing the amount of invalid data transmitted and lowering the cloud computing load. The preprocessed data is then transmitted at high speed to the data platform via a 5G industrial private network for unified storage and management. Simultaneously, the edge computing data fusion module connects to the enterprise's existing ERP, MES, and WMS systems through standardized API interfaces to obtain business data such as production plans, material inventory, and order information. To address the issue of data privacy protection across multiple systems, this module employs federated learning technology, training data features locally on each business system and only transmitting the trained model parameters to the data platform for fusion. This creates a unified global view of the production line status without leaking the original business data, completely breaking down information silos in traditional scheduling.

[0039] The scheduling decision module is the core control unit of the system, consisting of a digital twin platform and an intelligent scheduling engine, and is responsible for generating globally optimal material scheduling decisions.

[0040] Digital Twin Platform: The digital twin platform first constructs a three-dimensional digital model of all elements of the production line and materials, accurately mapping the geometry, spatial relationships, and operational logic of the assembly line workstations, AGV travel paths, material storage areas, automated warehouses, and automatic loading / unloading mechanisms. A unique UHF RFID electronic tag is attached to each material unit, and a digital twin corresponding to the physical material is created in the digital twin model. This establishes a digital archive covering the entire process from material warehousing, storage, picking, distribution, and online delivery to finished product output, recording the status changes and flow information of materials at each stage. By synchronizing the operational data of the physical production line in real time, the digital twin platform establishes a millisecond-level bidirectional mapping relationship between the physical and digital worlds. Any status change of the physical production line is reflected in the digital twin model in real time, and scheduling instructions generated by the digital twin model are also sent to the physical production line for execution in real time.

[0041] Intelligent Scheduling Engine: The intelligent scheduling engine incorporates a material demand forecasting model based on deep Q-networks and a multi-AGV collaborative path planning algorithm based on time windows. The material demand forecasting model takes the global status data of the production line as input and outputs the material demand for each workstation at different time dimensions in the future; the multi-AGV collaborative path planning algorithm generates conflict-free AGV travel paths and task execution sequences based on material delivery tasks and the real-time status of AGVs.

[0042] The execution control module communicates with the scheduling decision module, responsible for translating scheduling decisions into specific equipment control commands and executing them. It consists of an AGV scheduling system, an electronic picking system, and workstation terminals. The AGV scheduling system receives material delivery tasks from the intelligent scheduling engine and controls the AGVs to load, transport, and unload materials according to the planned path. The electronic picking system guides material handlers to accurately pick materials according to scheduling instructions through light indicators and voice guidance. After picking, it automatically prints a delivery slip containing material information and the target workstation. The workstation terminals are installed at each assembly workstation, displaying real-time information on materials to be delivered, current production tasks, and process requirements, guiding workers in assembly operations.

[0043] The monitoring and early warning module communicates with both the digital twin platform and the intelligent scheduling engine to achieve real-time monitoring and closed-loop control of the scheduling process. This module works in conjunction with the digital twin platform, providing a visual 3D interface that intuitively displays the overall operating status of the production line, the real-time location of materials, the AGV's trajectory, and abnormal information. Simultaneously, the monitoring and early warning module automatically detects various production anomalies by comparing real-time collected data with preset normal operating thresholds, triggering a closed-loop rescheduling mechanism to minimize the impact of anomalies on production.

[0044] The intelligent scheduling engine's material demand forecasting model employs a training method combining offline pre-training and online incremental learning. In the offline pre-training phase, historical production data from the enterprise is used to train the model, enabling it to grasp the basic material demand patterns of the production line. In the online incremental learning phase, the model iterates and optimizes its parameters at fixed intervals based on real-time production line data, continuously adapting to the dynamic changes in the production line. The model's comprehensive optimization objectives are maximizing production line utilization, minimizing material inventory costs, and minimizing the total AGV delivery distance. A multi-objective optimization algorithm is used to find the optimal forecast result, ensuring the accuracy of material demand forecasting.

[0045] The time-window-based multi-AGV cooperative path planning algorithm executes as follows: First, it uses the most classic heuristic search algorithm in path planning to generate an initial optimal path for each AGV performing a task. Then, based on the travel speed and task priority of each AGV, a unique time window is assigned to each segment of the path, ensuring that only one AGV is allowed to pass through the same segment at the same time, fundamentally avoiding path conflicts. When unexpected situations such as AGV failure, task changes, or path congestion occur, the algorithm immediately suspends the task execution of the affected AGVs, recalculates the paths and time windows of all affected AGVs, and updates the scheduling instructions to ensure the continuity and safety of AGV operations.

[0046] The digital twin platform has simulation and deduction capabilities, which can import different production plans, material supply rhythms and abnormal scenario parameters to simulate the production line operation status under corresponding conditions. Through data analysis, it can identify potential scheduling bottlenecks and production risks in advance, and output optimized production plans and scheduling schemes to help managers adjust production strategies in advance and reduce losses caused by production uncertainties.

[0047] The monitoring and early warning module can automatically identify five common production anomalies: material shortages, incorrect or missing materials, AGV malfunctions, equipment downtime, and order changes. It then classifies these anomalies into three levels based on their type and impact on production. For minor anomalies with short delays, the system automatically adjusts the timing of subsequent delivery tasks and reallocates AGV resources, restoring normal production without manual intervention. For serious anomalies, such as shortages of critical materials that could lead to production line shutdowns, the system immediately issues warnings to relevant managers through multiple channels, including sound and light alerts, SMS messages, and an app, and generates multiple alternative scheduling plans for managers to choose from, minimizing anomaly handling time.

[0048] The AGV scheduling system monitors the battery level, location, operating status, and task execution progress of all AGVs in real time. When the system detects that an AGV's remaining battery level is below a set threshold, it automatically dispatches the AGV to the nearest charging area for charging. After charging is complete, the AGV automatically returns to the task queue and takes on higher-priority tasks. This mechanism effectively avoids task interruptions caused by insufficient battery power, improving the overall utilization rate of AGV equipment.

[0049] This embodiment achieves a transformation in material scheduling from passive response to proactive prediction by organically combining digital twin full-element virtual-real mapping, edge computing federated data fusion, deep Q-network multi-time-dimensional material demand prediction, and time-window-based multi-AGV collaborative scheduling. This significantly improves production line utilization, greatly reduces material management error rate, significantly increases AGV equipment utilization, significantly shortens production anomaly response time, effectively reduces material management labor costs, and comprehensively improves the automation level of material management and production operation stability of desktop assembly lines. It also solves the technical problems of traditional desktop assembly lines relying on manual experience and adopting a static batch material preparation mode, which cannot adapt to dynamic production scenarios with multiple varieties, small batches, and fragmented orders, resulting in material supply and demand mismatch and low production line utilization.

[0050] Figure 2 This is a method for intelligent material scheduling in a desktop computer assembly line according to one embodiment of the present invention, such as... Figure 2 As shown, the method includes the following steps:

[0051] Step S1: Full-element digital twin modeling of production line and materials, the specific steps are as follows:

[0052] This step establishes a digital foundation for the entire scheduling system, creating a digital mirror of the physical production line. First, a 3D modeling tool is used to construct a full-element 3D digital twin model of the production line and materials. This model accurately maps the spatial location of each workstation on the assembly line, the topology of the AGV travel network, the storage location division of the material temporary storage area, the shelf layout of the automated warehouse, and the motion logic of the automatic loading and unloading mechanism. The geometric accuracy of the model reaches the centimeter level, and the operating logic is completely consistent with the physical equipment.

[0053] Each material unit is bound to a unique 920MHz UHF RFID electronic tag, which pre-stores basic attribute information such as material code, batch number, quantity, supplier, warehousing time, and expiration date. A one-to-one digital twin is created for each physical material in a 3D digital twin model, which synchronizes the physical material's location, status, and flow records in real time. A digital archive covering the entire material lifecycle—from warehousing, storage, picking, distribution, and production line to finished product output—is established, automatically recording information such as time nodes, operators, and equipment numbers at each stage, supporting full-process material traceability.

[0054] By using a 5G industrial private network, millisecond-level data synchronization between the physical production line and the 3D digital twin model is achieved, establishing a two-way mapping relationship between the virtual and the real: any change in the state of the physical production line (such as material movement, equipment start-up and shutdown, and changes in workstation cycle time) will be synchronized to the digital twin model within 100ms; the scheduling instructions generated by the digital twin model will also be sent to the corresponding execution equipment of the physical production line in real time, ensuring that the virtual and real states are completely consistent.

[0055] Step S2: Multi-source data acquisition and global state fusion, the specific steps are as follows:

[0056] This step provides a complete and accurate data foundation for scheduling decisions. Real-time production line status data is collected synchronously through multi-source sensing devices deployed throughout the entire production line. Specifically: RFID tag readers collect material identification and location data; UWB positioning devices collect real-time location and movement trajectory data of AGVs and material handling vehicles; visual recognition cameras collect data on material inventory and equipment operation at workstations; and sensor arrays collect operating parameter data such as equipment temperature, vibration, and current.

[0057] The collected raw data undergoes preprocessing, including: filtering sensor noise data using a moving average method, removing outlier data using the 3σ criterion, extracting key feature data reflecting the production line's operating status, and standardizing all data to unify data format and units. The preprocessed data is then transmitted via a 5G industrial private network to edge computing nodes for preliminary processing before being uploaded to a data platform for unified storage.

[0058] Simultaneously, standardized API interfaces are used to connect with existing enterprise ERP, MES, and WMS systems to obtain business data such as production plans, material inventory, order information, and process parameters. Federated learning technology is employed to perform privacy-preserving fusion processing on multi-source heterogeneous data. Each business system completes data feature training locally, and only the trained model parameters are uploaded to the data platform for global fusion. Without disclosing the original business data, a unified global view of the production line status, including production line operation status, material status, equipment status, and order status, is formed.

[0059] Step S3: Material demand forecasting based on deep Q-networks, the specific steps are as follows:

[0060] This step enables accurate prediction of material demand, providing a basis for advance delivery. A material demand forecasting model based on a deep Q-network (DQN) is constructed. The model adopts a three-layer fully connected neural network structure. The input layer contains six types of input features: historical production data, current production plan, material inventory level at each workstation, equipment failure rate, order change probability, and material delivery time. The hidden layer contains 128 neurons. The output layer outputs the material demand of each workstation in the next 15 minutes, 30 minutes, and 60 minutes.

[0061] The material demand forecasting model employs a training method combining offline pre-training and online incremental learning. In the offline pre-training phase, at least six months of historical production data from the enterprise are used to train the model, enabling it to grasp the basic material demand patterns of the production line. In the online incremental learning phase, the model updates its parameters every 15 minutes, iteratively optimizing network parameters based on real-time production line data to continuously adapt to dynamic changes in the production line. The model aims to maximize production line utilization, minimize material inventory costs, and minimize the total AGV delivery distance as comprehensive optimization objectives. A weighted summation method is used to transform the multi-objective problem into a single-objective optimization problem, from which the optimal forecast result is obtained.

[0062] Step S4: Dynamic material scheduling and AGV collaborative path planning, the specific steps are as follows:

[0063] This step generates the optimal scheduling plan based on the material demand forecast results. First, based on the material demand at each workstation across different time dimensions output by the material demand forecast model, material delivery tasks are automatically generated, specifying the type and quantity of materials to be delivered, the target workstation, and the delivery deadline. A priority scheduling mechanism is adopted, dividing the scheduling tasks into three priorities: urgent orders, tasks corresponding to workstations with material shortages, and tasks involving abnormal materials have the highest priority; regular production tasks have medium priority; and replenishment tasks have the lowest priority, ensuring that critical tasks are executed first.

[0064] A dynamic batch delivery strategy is adopted, which dynamically adjusts the batch size and delivery cycle of each delivery based on the real-time material consumption rate of each workstation. The amount of material delivered in a single delivery does not exceed the material consumption of the corresponding workstation in 2 hours, thereby avoiding material accumulation at workstations and improving material turnover rate.

[0065] To address the path conflict problem in parallel operations of multiple AGVs, a time window-based AGV path planning algorithm is adopted: First, the A* algorithm is used to generate an initial optimal path for each AGV with a task to be executed; then, based on the travel speed and task priority of each AGV, a unique time window is allocated to each segment of the path to ensure that only one AGV is allowed to pass through the same segment at the same time; when an AGV failure, task change, or path blockage occurs, the algorithm immediately recalculates the paths and time windows of all affected AGVs, updates the scheduling instructions, and avoids path crossings and collisions.

[0066] Step S5: Scheduling execution and closed-loop feedback control, the specific steps are as follows:

[0067] This step enables the implementation and closed-loop optimization of scheduling instructions. The scheduling instructions generated by the scheduling decision module are sent to the execution control module via industrial Ethernet. The AGV scheduling system, electronic picking system, and workstation terminals collaborate to complete the scheduling task: the electronic picking system guides material handlers to pick materials accurately using lights and voice prompts; the AGV scheduling system controls AGVs to load materials in the picking area and deliver them to the target workstations according to the planned path; the workstation terminals display material delivery information and production tasks in real time, guiding workers in assembly operations.

[0068] All data during the scheduling process is collected in real time, including material delivery time, AGV driving status, and workstation production progress. This data is fed back to the 3D digital twin model and intelligent scheduling engine for real-time evaluation of the scheduling execution status. Anomaly detection is achieved by comparing the real-time collected data with preset normal operation thresholds. When anomalies such as material shortages, incorrect or missing materials, AGV malfunctions, equipment downtime, or order changes are detected, a rescheduling mechanism is immediately triggered. The response time of the rescheduling mechanism is no more than 1 minute. The optimal scheduling plan is regenerated and immediately sent to the execution control module for execution, achieving closed-loop control of the scheduling process.

[0069] According to one embodiment of the present invention, an electronic device is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the above-described intelligent material scheduling method for a desktop computer assembly line.

[0070] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:

[0071] Step S1: Construct a three-dimensional digital twin model of all elements of the production line and materials, mapping the geometry, positional relationships and operating logic of the assembly line workstations, AGV travel paths, material temporary storage areas, automated warehouses and automatic loading and unloading mechanisms, binding a unique UHF RFID electronic tag to each material unit and creating a corresponding digital twin, establishing a digital archive of the entire material lifecycle, and establishing a millisecond-level virtual-real bidirectional mapping relationship between the physical production line and the three-dimensional digital twin model;

[0072] Step S2: Collect real-time status data of the production line through multi-source sensing devices, perform noise filtering and feature extraction preprocessing on the collected raw data, and integrate the business data of the ERP system, MES system and WMS system to form a unified global status view of the production line.

[0073] Step S3: Construct a material demand forecasting model based on a deep Q-network. The model takes historical production data, current production plan, material inventory level of each workstation, equipment failure rate, order change probability and material delivery time as input features, and outputs the material demand of each workstation in the next 15 minutes, 30 minutes and 60 minutes.

[0074] Step S4: Generate material delivery tasks based on the material demand prediction results output by the material demand prediction model, and use a priority scheduling mechanism to allocate higher scheduling priority to urgent orders, material shortage workstations and abnormal materials. Use a time window-based AGV path planning algorithm to allocate a unique path time window to each AGV.

[0075] Step S5: The scheduling instruction is sent to the execution control module for collaborative execution. Data of the scheduling execution process is collected in real time and fed back to the three-dimensional digital twin model and the intelligent scheduling engine. The scheduling execution status is evaluated in real time. If an anomaly is detected, a rescheduling mechanism is triggered to generate the optimal scheduling scheme.

[0076] According to one embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the storage medium is located to execute the above-described intelligent material scheduling method for desktop computer assembly lines.

[0077] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0078] Step S1: Construct a three-dimensional digital twin model of all elements of the production line and materials, mapping the geometry, positional relationships and operating logic of the assembly line workstations, AGV travel paths, material temporary storage areas, automated warehouses and automatic loading and unloading mechanisms, binding a unique UHF RFID electronic tag to each material unit and creating a corresponding digital twin, establishing a digital archive of the entire material lifecycle, and establishing a millisecond-level virtual-real bidirectional mapping relationship between the physical production line and the three-dimensional digital twin model;

[0079] Step S2: Collect real-time status data of the production line through multi-source sensing devices, perform noise filtering and feature extraction preprocessing on the collected raw data, and integrate the business data of the ERP system, MES system and WMS system to form a unified global status view of the production line.

[0080] Step S3: Construct a material demand forecasting model based on a deep Q-network. The model takes historical production data, current production plan, material inventory level of each workstation, equipment failure rate, order change probability and material delivery time as input features, and outputs the material demand of each workstation in the next 15 minutes, 30 minutes and 60 minutes.

[0081] Step S4: Generate material delivery tasks based on the material demand prediction results output by the material demand prediction model, and use a priority scheduling mechanism to allocate higher scheduling priority to urgent orders, material shortage workstations and abnormal materials. Use a time window-based AGV path planning algorithm to allocate a unique path time window to each AGV.

[0082] Step S5: The scheduling instruction is sent to the execution control module for collaborative execution. Data of the scheduling execution process is collected in real time and fed back to the three-dimensional digital twin model and the intelligent scheduling engine. The scheduling execution status is evaluated in real time. If an anomaly is detected, a rescheduling mechanism is triggered to generate the optimal scheduling scheme.

[0083] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0084] According to one embodiment of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the above-described intelligent material scheduling method for desktop computer assembly lines.

[0085] Optionally, in this embodiment, the above-mentioned computer program product can be configured as a computer program that performs the following steps:

[0086] Step S1: Construct a three-dimensional digital twin model of all elements of the production line and materials, mapping the geometry, positional relationships and operating logic of the assembly line workstations, AGV travel paths, material temporary storage areas, automated warehouses and automatic loading and unloading mechanisms, binding a unique UHF RFID electronic tag to each material unit and creating a corresponding digital twin, establishing a digital archive of the entire material lifecycle, and establishing a millisecond-level virtual-real bidirectional mapping relationship between the physical production line and the three-dimensional digital twin model;

[0087] Step S2: Collect real-time status data of the production line through multi-source sensing devices, perform noise filtering and feature extraction preprocessing on the collected raw data, and integrate the business data of the ERP system, MES system and WMS system to form a unified global status view of the production line.

[0088] Step S3: Construct a material demand forecasting model based on a deep Q-network. The model takes historical production data, current production plan, material inventory level of each workstation, equipment failure rate, order change probability and material delivery time as input features, and outputs the material demand of each workstation in the next 15 minutes, 30 minutes and 60 minutes.

[0089] Step S4: Generate material delivery tasks based on the material demand prediction results output by the material demand prediction model, and use a priority scheduling mechanism to allocate higher scheduling priority to urgent orders, material shortage workstations and abnormal materials. Use a time window-based AGV path planning algorithm to allocate a unique path time window to each AGV.

[0090] Step S5: The scheduling instruction is sent to the execution control module for collaborative execution. Data of the scheduling execution process is collected in real time and fed back to the three-dimensional digital twin model and the intelligent scheduling engine. The scheduling execution status is evaluated in real time. If an anomaly is detected, a rescheduling mechanism is triggered to generate the optimal scheduling scheme.

[0091] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0092] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0093] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0094] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0095] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0096] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0097] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A material intelligent scheduling system for desktop computer assembly lines, comprising a data acquisition module, a scheduling decision module, an execution control module, and a monitoring and early warning module, characterized in that, It also includes a digital twin platform and an edge computing data fusion module; The data acquisition module includes an RFID tag reader, a UWB positioning device, a visual recognition camera, and a sensor array, used to synchronously collect real-time data on production cycle time, material status, AGV running trajectory, and equipment operating parameters at production line workstations. The edge computing data fusion module is communicatively connected to the data acquisition module. It performs noise filtering and feature extraction preprocessing on the raw data, transmits it to the data platform through the 5G industrial private network, connects with ERP, MES, and WMS systems to obtain business data, and completes privacy-preserving fusion of multi-source heterogeneous data based on federated learning technology. The scheduling decision module includes the digital twin platform and the intelligent scheduling engine; The digital twin platform constructs a three-dimensional model of all elements of the production line and materials, binds a unique UHF RFID tag to each material and creates a digital twin, establishes a digital archive of the entire life cycle of materials and a millisecond-level virtual-real bidirectional mapping between the physical and digital worlds; the intelligent scheduling engine has built-in a material demand prediction model based on deep Q network and a multi-AGV collaborative path planning algorithm based on time window. The execution control module is communicatively connected to the scheduling decision module and includes an AGV scheduling system, an electronic picking system, and a workstation terminal, used to convert scheduling decisions into equipment control commands. The monitoring and early warning module is communicatively connected to the digital twin platform and the intelligent scheduling engine, respectively. It works in conjunction with the digital twin platform to display the production line operation status, automatically detects production anomalies, and triggers a closed-loop rescheduling mechanism.

2. The intelligent material scheduling system according to claim 1, characterized in that, The intelligent scheduling engine's material demand prediction model adopts a training method that combines offline pre-training and online incremental learning. The comprehensive optimization objectives are to maximize production line utilization, minimize material inventory costs, and minimize the total AGV delivery distance. The material demand prediction accuracy of the material demand prediction model is greater than or equal to 95%.

3. The intelligent material scheduling system according to claim 1, characterized in that, The execution steps of the time window-based multi-AGV cooperative path planning algorithm are as follows: First, an initial optimal path is generated for each AGV. Then, based on the travel speed and task priority of each AGV, a unique time window is assigned to each segment of the path. Only one AGV is allowed to pass through the same segment at the same time. When an AGV malfunctions, the task changes, or the path is blocked, the time window-based multi-AGV cooperative path planning algorithm recalculates the paths and time windows of all affected AGVs.

4. The intelligent material scheduling system according to claim 1, characterized in that, The digital twin platform has simulation and deduction capabilities. It simulates the production line operation status under different production plans, different material supply rhythms, and different abnormal scenarios, and outputs optimized production plans and scheduling schemes.

5. The intelligent material scheduling system according to claim 1, characterized in that, The monitoring and early warning module automatically identifies five types of anomalies: material shortage, incorrect or missing materials, AGV malfunction, equipment downtime, and order changes. The monitoring and early warning module classifies and grades the anomalies into three levels based on their type and severity. For minor anomalies with delays of ≤5 minutes, the monitoring and early warning module automatically adjusts the timing of subsequent delivery tasks and reallocates AGV transport resources. For serious anomalies such as shortages of critical materials, the monitoring and early warning module immediately issues warnings through multiple channels, including sound and light, SMS, and the APP, and generates at least three alternative scheduling plans.

6. The intelligent material scheduling system according to claim 1, characterized in that, The AGV scheduling system monitors the battery level, location, operating status, and task execution progress of all AGVs in real time. When the remaining battery level of an AGV is less than 20%, the AGV scheduling system automatically dispatches the AGV to the nearest charging area for charging. After charging is completed, the AGV automatically returns to the task queue and takes on high-priority scheduling tasks.

7. A method for intelligent material scheduling in a desktop computer assembly line, characterized in that, The method for controlling the intelligent material scheduling system according to any one of claims 1-6 includes the following steps: Step S1: Construct a three-dimensional digital twin model of all elements of the production line and materials, mapping the geometry, positional relationships and operating logic of the assembly line workstations, AGV travel paths, material temporary storage areas, automated warehouses and automatic loading and unloading mechanisms, binding a unique UHF RFID electronic tag to each material unit and creating a corresponding digital twin, establishing a digital archive of the entire material lifecycle, and establishing a millisecond-level virtual-real bidirectional mapping relationship between the physical production line and the three-dimensional digital twin model; Step S2: Collect real-time status data of the production line through multi-source sensing devices, perform noise filtering and feature extraction preprocessing on the collected raw data, and integrate the business data of the ERP system, MES system and WMS system to form a unified global status view of the production line. Step S3: Construct a material demand forecasting model based on a deep Q-network, using historical production data, current production plan, material inventory level at each workstation, equipment failure rate, order change probability, and material delivery time as input features, and output the material demand for each workstation in the next 15 minutes, 30 minutes, and 60 minutes. Step S4: Generate material delivery tasks based on the material demand forecast results output by the material demand forecasting model, and use a priority scheduling mechanism to allocate higher scheduling priority to urgent orders, material shortage workstations and abnormal materials. Use a time window-based AGV path planning algorithm to allocate a unique path time window to each AGV. Step S5: The scheduling instruction is sent to the execution control module for collaborative execution. Data of the scheduling execution process is collected in real time and fed back to the three-dimensional digital twin model and the intelligent scheduling engine. The scheduling execution status is evaluated in real time. If an anomaly is detected, the rescheduling mechanism is triggered to generate the optimal scheduling scheme.

8. The intelligent material scheduling method according to claim 7, characterized in that, In step S3, the offline pre-training of the material demand forecasting model uses historical production data of 6 months or more, and the online incremental learning of the material demand forecasting model is performed every 15 minutes to iteratively optimize the network parameters of the material demand forecasting model based on real-time production line operation data.

9. The intelligent material scheduling method according to claim 7, characterized in that, In step S4, a dynamic batch delivery strategy is adopted, which dynamically adjusts the batch size and delivery cycle of a single delivery based on the real-time material consumption rate of each workstation. The amount of material delivered in a single delivery is less than or equal to the material consumption of the corresponding workstation in 2 hours.

10. The intelligent material scheduling method according to claim 7, characterized in that, In step S5, an anomaly is detected by comparing the real-time collected data with the preset normal operation threshold. The response time of the rescheduling mechanism is less than or equal to 1 minute. After the rescheduling scheme is generated, it is immediately sent to the execution control module for execution.

Images