Intelligent Manufacturing Full-Process Virtual-Real Linkage Training System Based on Digital Twin

Abstract

This invention belongs to the field of intelligent manufacturing simulation technology. It discloses a full-process virtual-real linkage training system for intelligent manufacturing based on digital twins, comprising: acquiring training task instructions and constructing a process twin sequence containing a set of target equipment, process connection relationships, and workpiece flow identifiers; collecting the execution status, response status, and pose status of each target equipment, recording the status fragments into a unified time-series ledger to form a field status mirror; verifying and determining the execution status and connection of each process node, generating linkage freeze or linkage release markers for process nodes with cross-equipment connection relationships; upon receiving an operation request, calling the field status mirror and linkage identifiers to perform virtual-real consistency verification, issuing linkage execution instructions to process nodes that pass the verification, and constructing a full-process virtual-real linkage training result; thus improving the security of the digital twin training system in complex multi-unit collaborative operation scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing simulation technology, and more specifically, to an intelligent manufacturing full-process virtual-real linkage training system based on digital twins. Background Technology

[0002] Existing digital twin training systems generally face the problem of state synchronization distortion between the cloud-based virtual environment and on-site physical equipment. Specifically, when the cloud-based digital twin model interacts with multiple physical devices on-site in real time, factors such as network transmission fluctuations, data processing time, and inconsistent response speeds of various device controllers lead to time differences and information discrepancies between the device status displayed in the virtual environment and the actual status of the physical devices. For example, in the collaborative operation of robot assembly and warehousing logistics, the cloud interface may show that the robot has completed the grasping action and issued a transfer request, but the actual physical robot on-site is still in motion due to execution delays. If the AGV (Automated Guided Vehicle) arrives in place ahead of schedule according to the instructions of the virtual environment, it will cause collisions at the workstation or a wasted trip. When the next operation is made based on the distorted virtual image, incorrect instructions will be transmitted to the physical system, which may result in equipment collisions, workpieces falling, or even safety accidents. Current solutions mainly rely on high-frequency and continuous querying of device status and setting up buffer queues. However, this passive data synchronization method not only significantly increases the pressure on cloud servers and network bandwidth, but also fails to fundamentally eliminate the continuous accumulation of state deviations caused by the asynchronous response of multiple devices, making it difficult to guarantee the reliability of virtual-physical linkage.

[0003] In view of this, the present invention proposes a virtual-real linkage training system for the entire intelligent manufacturing process based on digital twins to solve the above problems. Summary of the Invention

[0004] To overcome the aforementioned shortcomings of existing technologies and achieve the above objectives, this invention provides the following technical solution: a digital twin-based intelligent manufacturing full-process virtual-real linkage training system, comprising:

[0005] Sequence Construction Module: Obtains training task instructions and constructs corresponding process twin sequences based on the training task instructions. The process twin sequence includes the set of target equipment participating in this training, the process connection relationship, and the workpiece flow identifier.

[0006] Image building module: Based on the target device set, collect the on-site execution status, control response status and pose confirmation status of each target device, generate corresponding status fragment records according to device category, write the status fragment records into a unified time-series ledger, and form the on-site status image of the current training cycle;

[0007] The marking module performs on-site status mirroring to verify and determine the connection of each process node in the process twin sequence. It generates linkage markers for process nodes with cross-equipment connection relationships. The linkage markers include linkage freeze markers or linkage release markers.

[0008] Execution module: Upon receiving a subsequent operation request, it calls the on-site status image and linkage identifier to perform virtual-real consistency verification, issues linkage execution instructions to the process nodes that pass the verification, and constructs the full-process virtual-real linkage training results.

[0009] The technical effects and advantages of this invention, a fully virtual and real-world integrated training system for intelligent manufacturing based on digital twins, are as follows:

[0010] This invention pre-constructs a complete process network containing a set of target devices, process connection relationships, and workpiece flow identifiers through a process twin sequence. A unified time-series ledger records the state fragments of each device sequentially to form a real-world state mirror, providing a unified time reference for comparing the virtual environment and the physical devices. This avoids the pressure on cloud servers and network bandwidth caused by traditional high-frequency polling methods. The core manifestation of virtual-real distortion is that the virtual interface displays the process as completed in advance while the actual equipment is still executing. By verifying the execution status and connection judgment of each process node, it accurately identifies three process connection states: waiting, reverse access, and normal connection. For process nodes with cross-device connection relationships, it generates linkage freeze or linkage release flags, thereby preventing subsequent equipment from following instructions from the distorted virtual environment. Premature access can cause workstation collisions or idle waiting. Upon receiving subsequent operation requests, the system compares the correspondence between the actual situation on-site and the interface mapping at the workstation, action stage, and handover stage. Combined with the currently held linkage identifier, a virtual-real consistency check is performed. Requests with workstation misalignment, advanced action stage, missing items in the handover stage, or holding linkage freeze markers are blocked and transferred to the confirmation area. Independent verification is performed through the supplementary confirmation mechanism of adjacent verification chains to prevent erroneous operations based on distorted virtual images from being transmitted to the physical system, causing equipment collisions, workpiece drops, or safety accidents. This fundamentally eliminates the accumulation of state deviations caused by asynchronous responses from multiple devices, ensuring the reliability of virtual-real linkage and improving the operational accuracy and safety of the digital twin training system in complex multi-unit collaborative operation scenarios. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the intelligent manufacturing full-process virtual-physical linkage training system based on digital twins of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0013] Example 1

[0014] Please see Figure 1 As shown, this embodiment of the intelligent manufacturing full-process virtual-physical linkage training system based on digital twins includes:

[0015] Sequence construction module: Obtains training task instructions and constructs corresponding process twin sequences based on the training task instructions;

[0016] In digital twin training systems, the problem of state synchronization distortion between the cloud-based virtual environment and on-site physical equipment severely impacts the safety and effectiveness of training. Traditional solutions rely on frequent queries of device status, which not only increases server load but also fails to fundamentally solve the problem of accumulating state deviations caused by asynchronous responses from multiple devices. Therefore, this solution constructs a process twin sequence, establishing a unified linkage basis before the training task begins, enabling the virtual environment and physical equipment to operate collaboratively within the same process framework.

[0017] Specifically, the process task field, equipment call field, and workpiece batch field in the training task instruction are analyzed. The process task field describes the type of processing technology to be completed in this training, such as: parts assembly and inspection, CNC machining and transfer, etc.; the equipment call field specifies the name and quantity of the equipment unit to be involved in this training; the workpiece batch field records the batch number, transfer quantity, and transfer order of the workpiece.

[0018] Based on the requirements of the process task field, the corresponding equipment unit is selected from the digital vocational comprehensive training platform. In this embodiment, the digital vocational comprehensive training platform includes seven types of equipment units: robot unit, warehousing unit, CNC machining unit, embodied intelligent robot unit, robot assembly unit, intelligent vision inspection unit, and digital intelligent control unit. For example, when the process task field is parts assembly and inspection, the system automatically selects the robot assembly unit, intelligent vision inspection unit, and warehousing unit as components of the target equipment set; when the process task field is flexible processing and transfer, the system selects the CNC machining unit, robot unit, AGV execution device, and vision inspection device. By parsing the equipment call field, the specific number of each type of equipment unit is determined, generating a target equipment set containing all participating equipment. It should be noted that each target device in the target equipment set has a unique equipment identifier.

[0019] Based on the workpiece flow sequence in the workpiece batch field, establish a process access position, a process execution position, and a process handover position for each target device in the target device set. The process access position indicates the workpiece's ready state to enter the device, including whether the workstation is idle, whether the preceding workpiece has left, and whether the device is in a receptive state. The process execution position indicates the status of the device performing operations such as processing, assembly, inspection, or transfer on the workpiece, including the start time of the action, the current execution progress, and the estimated completion time. The process handover position indicates the handover status of the workpiece about to leave the device and move to the next device, including whether the workpiece has completed its current process, whether the subsequent device is ready to receive it, and whether the handover channel is unobstructed.

[0020] Based on the equipment call field, cross-unit process connection relationships are formed. The process connection relationship describes the flow logic of the workpiece between different devices. For example, when the workpiece needs to be transferred to the intelligent vision inspection unit for quality inspection after the workpiece is assembled from the robot assembly unit, a connection relationship is established in the process twin sequence from the process handover position of the robot assembly unit to the process access position of the intelligent vision inspection unit. When the workpiece needs to be transferred to the storage unit by the AGV execution device after the workpiece is processed from the CNC machining unit, a multi-segment connection relationship is established from the process handover position of the CNC machining unit to the process access position of the AGV execution device, and finally to the process access position of the storage unit.

[0021] For each process node in the process connection relationship, node entry conditions, node holding conditions, and node release conditions are written to form a verifiable process twin sequence. The node entry conditions specify the prerequisites for the process node to be activated, such as: the preceding equipment has completed and issued a handover confirmation signal, the current equipment station is idle and there is no fault alarm, and the workpiece identification is consistent with the batch requirements. The node holding conditions specify the constraints that the process node must continuously meet during execution, such as: the equipment remains in the execution state, the workpiece remains within the equipment station, and there is no safety boundary intrusion. The node release conditions specify the standards for the completion of the process node and allow it to flow, such as: the equipment return action completion mark, the subsequent equipment has issued a receiving preparation signal, and the workpiece quality parameters meet the requirements.

[0022] Each process node in the process twin sequence is associated with a visual object in the digital twin interactive interface and a control object in the field control link, ensuring that the same process node uses the same node identifier in both the interface display and on-site execution. The digital twin interactive interface is a display platform for the cloud-based virtual environment, presenting the operating status of each device, the flow trajectory of the workpiece, and the execution progress of the process through 3D visualization technology. The field control link is a communication network connecting cloud commands and field device controllers, responsible for accurately transmitting control commands issued from the cloud to the corresponding devices and collecting real-time status feedback from the devices. By using a unified node identifier, it is ensured that when the cloud interface displays that a process node is completed, the corresponding process node in the field device controller must also be in a completed state, thereby establishing a strong correlation between the virtual and physical worlds.

[0023] The process twin sequence is loaded with a training scenario. This includes writing the workpiece flow identifier, target equipment set, and process connection relationships. The loaded process twin sequence is then set as the sole linkage basis for this training. The workpiece flow identifier uses a coding format of batch number + workpiece number + flow stage. For example, BATCH20240615-WP01-STAGE02 indicates that workpiece number 1 from batch June 15, 2024, is currently in the second flow stage. Setting the process twin sequence as the sole linkage basis means that all subsequent equipment scheduling, status verification, and instruction issuance will be based on this process twin sequence. Any operation request not in this sequence will be considered invalid and rejected by the system, thus avoiding confusion in the virtual-physical linkage due to external interference or misoperation.

[0024] It should be noted that the workpiece flow order in this embodiment is determined according to the processing route of the technological task. For example, under the process route of parts assembly - quality inspection - warehousing, the workpiece flow order is: robot assembly unit → intelligent vision inspection unit → AGV execution device → warehousing unit; under the process route of CNC machining - deburring - surface inspection - assembly, the workpiece flow order is: CNC machining unit → embodied intelligent robot unit → intelligent vision inspection unit → robot assembly unit. Implementers can set the workpiece flow order according to the specific technological task requirements.

[0025] Image building module: Based on the target device set, collect the on-site execution status, control response status and pose confirmation status of each target device, generate corresponding status fragment records according to device category, write the status fragment records into a unified time-series ledger, and form the on-site status image of the current training cycle;

[0026] Traditional high-frequency query methods only obtain the instantaneous status of the device each time, which cannot fully record the process of device status changes. As a result, when the virtual environment and the physical device are out of sync, the system cannot trace the time and cause of the status deviation. This solution establishes a unified time-series ledger to record the status changes of all devices in chronological order, forming a field status mirror, and providing a reliable data foundation for subsequent virtual-physical consistency verification.

[0027] Specifically, status acquisition triggers are initiated for each target device in the target device set. Status acquisition triggering is a status query command sent by the system to the device controller. This embodiment uses a combination of periodic triggering and event-driven triggering. The time interval for periodic triggering is set to 200 milliseconds. This value is based on the fact that the minimum action cycle of a robot motion controller is typically 100 milliseconds, the position update cycle of an AGV actuator is approximately 150 milliseconds, and the status refresh cycle of a CNC machining controller is approximately 180 milliseconds. Therefore, 200 milliseconds is chosen as the periodic triggering interval to cover the status update cycles of various devices while avoiding excessively high frequencies that could burden communication. Event-driven triggering responds immediately when a device experiences a sudden change in status. For example, when events such as device alarms, workpiece arrival, or action completion occur, the device controller actively pushes status updates to the system.

[0028] The status acquisition triggers correspond to the local status output terminals of the robot motion controller, warehouse handling controller, CNC machining controller, vision inspection device, AGV execution device, and assembly execution device, respectively. Each type of device controller has a standardized status output interface. In this embodiment, the OPCUA communication protocol is used to establish a data connection between the cloud system and the field controller. The OPCUA protocol has the characteristics of cross-platform compatibility, high security, and good real-time performance, making it suitable for equipment status acquisition in industrial settings.

[0029] The system reads the on-site execution status from each target device. The on-site execution status reflects the current stage of the action being performed by the device, including action start flag, action progress flag, action completion flag, and action exception flag. The action start flag records the moment when the device receives the execution command and begins execution. The action progress flag records the real-time progress of the action being performed by the device, for example: current execution progress 65%, the robotic arm has reached the middle position. The action completion flag records the moment and result when the device completes this action. The action exception flag records any abnormal situations that occur during the execution process, for example: a workpiece position offset of 3.2mm is detected, exceeding the allowable range of 2.0mm. The read on-site execution status is encapsulated into an execution status fragment, which is stored in JSON format and includes fields such as device identifier, status type, status value, and timestamp.

[0030] The control response status is read from each target device. This status reflects the device's reception and processing of commands sent from the cloud, including command reception flag, command lock flag, command execution flag, and command release flag. The command reception flag indicates that the device controller has successfully received the control command sent from the cloud and completed command parsing. For example: Received command ID: CMD20240615102315, command content: Move to workstation P03. The command lock flag indicates that the device controller has loaded the command into its local execution queue and locked it, preventing further reception of commands for the same device. Other instructions for the process, such as: instruction CMD20240615102315 is locked, the device enters dedicated execution mode; instruction execution flag indicates that the device has started executing the instruction, for example: start executing instruction CMD20240615102315, estimated time is 12 seconds; instruction release flag indicates that the device has completed instruction execution and released the locked state, and can receive the next instruction, for example: instruction CMD20240615102315 has been completed and released, the device returns to idle state; the read control response status is encapsulated into a response status fragment.

[0031] The pose confirmation status is read from each target device. The pose confirmation status reflects the device's current spatial position, attitude direction, and workpiece attachment status, including the current position segment identifier, attitude landing point identifier, workpiece attachment identifier, and workstation occupancy identifier. The current position segment identifier records the spatial area where the device or its actuator is located, for example: the robotic arm is currently located in the working area A3, with coordinates (X: 1250mm, Y: 780mm, Z: 450mm). The attitude landing point identifier records the attitude angle of the device's actuator, for example: the robotic arm's end effector attitude: pitch angle 15 degrees, yaw angle -8 degrees, roll angle 2 degrees. The workpiece attachment identifier records whether the device carries a workpiece and the workpiece's identification information, for example: attached workpiece: BATCH20240615-WP01, attachment method: vacuum adsorption, adhesion force 78%. The workstation occupancy identifier records the workstation resources occupied by the device. The read pose confirmation status is encapsulated into a pose status fragment.

[0032] Based on the target device's device category, execution state segments, response state segments, and pose state segments are merged and registered, forming a set of state segment records for the same target device within the same training cycle. A training cycle refers to the time period from the start of execution at a process node to its completion and release. In this embodiment, a typical training cycle duration is 5 to 60 seconds, depending on the complexity of the specific process. The merging and registration process involves arranging all execution state segments, response state segments, and pose state segments collected by the same device within the same training cycle in chronological order according to their timestamps, forming a complete state change sequence for the device within that cycle.

[0033] Status fragment records are written to a unified time-series ledger in the order of entry. This ledger is a distributed ledger structure that records all equipment status changes chronologically, employing a chain-like storage method from blockchain technology to ensure the immutability and traceability of status records. Each record in the ledger includes fields such as record number, timestamp, equipment identifier, status fragment type, status fragment content, and the hash value of the previous record. The integrity of the records is ensured through a hash chain. For each group of status fragment records, corresponding equipment identifiers, process node identifiers, and workpiece flow identifiers are written into the unified time-series ledger, establishing a link between status records and a process twin sequence, forming a mirror image of the on-site status.

[0034] The field status mirror is a snapshot of the unified time-series ledger at the current moment, containing the latest status of all devices and the complete history of status changes from the start of the training to the current moment. Through the field status mirror, the system can accurately grasp the real-time status of each device and also trace back the device status at any point in time, providing a reliable basis for verifying the consistency between the virtual and real systems.

[0035] It should be noted that this embodiment uses a time synchronization protocol to ensure that all device controllers are synchronized with the clock of the cloud system. The clock synchronization accuracy reaches the microsecond level, avoiding the time sequence disorder of status records due to clock deviation.

[0036] The marking module verifies and determines the execution status of each process node in the process twin sequence based on the on-site status mirror, and generates linkage identifiers for process nodes with cross-equipment connection relationships.

[0037] The root cause of the virtual-real synchronization distortion is that the virtual environment displays and schedules based on the expected state, while the actual state of the physical equipment may deviate from the expectation due to factors such as execution delay and network fluctuation. This solution confirms whether the physical equipment has truly reached the expected state through on-site verification and confirms whether the process handover between cross-equipment has the conditions for execution through connection judgment. In this way, the problem of virtual-real inconsistency can be detected before the instruction is issued, avoiding equipment collisions and workpiece drops caused by erroneous instructions.

[0038] Specifically, the system retrieves the status fragment records corresponding to the current process node and the status fragment records of subsequent nodes with the same workpiece flow identifier from the unified time-series ledger corresponding to the on-site status mirror. The current process node refers to the process node currently being executed or about to be executed in the process twin sequence, for example: Process node N03: The robot assembly unit performs assembly actions; subsequent nodes refer to process nodes that are located after the current process node in the workpiece flow sequence, for example: Process node N04: The intelligent vision inspection unit performs quality inspection. By retrieving the status fragment records of subsequent nodes, it is possible to detect in advance whether the subsequent equipment is ready to receive the workpiece, avoiding workpiece delays or equipment waiting due to the inability of subsequent equipment to be ready after the current equipment has completed its work.

[0039] The state fragment record includes execution state fragments, response state fragments, and pose state fragments. These three types of fragments describe the state of the device from different dimensions. The execution state fragment describes the device's action execution, the response state fragment describes the device's processing of instructions, and the pose state fragment describes the device's spatial position and workpiece attachment. The combination of the three can comprehensively determine whether the device is truly in position.

[0040] The status of the current target device and the status of the subsequent target device are verified based on the execution status segments, response status segments, and pose status segments of the current process node and the subsequent node, generating process node verification results. The specific verification process is as follows: First, check whether there is an action completion flag in the execution status segment of the current target device. If so, it means that the current device has completed the execution of the action of this process. Second, check whether there is an instruction release flag in the response status segment of the current target device. If so, it means that the current device has completed instruction processing and is ready to release resources. Third, check whether the workpiece attachment flag in the pose status segment of the current target device shows that the workpiece has been separated or transferred to the handover position. If so, it means that the workpiece is ready to leave the current device. For the subsequent target device, check whether there is an action start flag or a receive preparation flag in its execution status segment, check whether the station occupancy flag in its pose status segment shows that the station is idle, and check whether there is an instruction receive flag or an instruction lock flag in its response status segment.

[0041] When the verification result of a process node indicates that the current target device has completed its process node but the subsequent target device has not yet met the entry conditions, a waiting-to-be-occupied flag is established for that process node. At this time, the virtual environment displays that the current device is complete, but in reality, the workpiece cannot be transferred to the subsequent device; the workpiece remains on the current device, and the current device's workstation resources are occupied, preventing it from receiving the next workpiece. The waiting-to-be-occupied flag is used to mark this state where the current device is completed but the subsequent device is not ready, preventing the virtual environment from mistakenly marking the current device as idle and preventing the system from issuing new workpiece receiving instructions to the current device. For example, the robot assembly unit has completed assembly and placed the workpiece at the handover position, but the intelligent vision inspection unit's workstation is occupied because the inspection of the previous workpiece has not yet been completed. In this case, although the robot assembly unit has completed its action, the workpiece is still at its handover position, and a waiting-to-be-occupied flag is established.

[0042] When the verification result of a process node indicates that the current target device has not yet completed its process node, but the subsequent target device has already initiated an access action, a reverse access flag is established for that process node. This situation is considered abnormal, indicating that the subsequent device has started prematurely before the current device has completed its task. This could be due to a display error in the virtual environment causing the subsequent device to receive an early instruction, or it could be due to a malfunction in the device controller causing a false action. The reverse access flag is used to mark this dangerous state, and the system needs to immediately issue a freeze command to both the current device and the subsequent device to prevent collisions or workpiece drops. For example, if the robot assembly unit has only completed 60% of its assembly action and the workpiece is still held by the robotic arm, but the intelligent vision inspection unit's conveyor belt has already started and extended into the robot's working area to receive the workpiece, a reverse access flag is established, and both devices are immediately frozen.

[0043] When the verification result of a process node indicates that the current target device has completed its process node and the subsequent target device has initiated an access action, a normal connection marker is established for that process node. This situation indicates that the virtual environment and the physical equipment are in the same state, the process handover is ready for execution, and workpiece transfer can proceed safely. The normal connection marker is used to mark this ideal state, allowing the system to confidently issue subsequent linkage execution commands. For example, if the robot assembly unit has completed assembly and stably placed the workpiece at the handover position, the intelligent vision inspection unit's workstation is idle, and the conveyor belt has started and is in place, a normal connection marker is established, and workpiece transfer is permitted.

[0044] This analysis clarifies the preceding, current, and succeeding target devices for the current process node within the process connection relationship. The preceding target device is the device that precedes the current process node in the workpiece flow sequence; the current target device is the device that executes the current process node; and the succeeding target device is the device that follows the current process node in the workpiece flow sequence. When the preceding, current, and succeeding target devices belong to different equipment categories or equipment units, the current process node is determined to be a process node with cross-equipment connection relationships.

[0045] Cross-device handover is a high-risk area for virtual-real synchronization distortion because different devices have different controller response speeds, execution accuracies, and communication delays, which can easily lead to inconsistencies in state at the handover point. For example, the handover between a robot assembly unit (robot category) and an intelligent vision inspection unit (vision inspection category) is a cross-device handover. The robot assembly unit has a motion completion accuracy of ±0.1mm and a response time of about 80ms, while the intelligent vision inspection unit has a workpiece positioning accuracy of ±0.5mm and a conveyor belt start time of about 200ms. The performance difference between the two may cause the workpiece position to shift at the handover point. As another example, the handover between a CNC machining unit (CNC category) and an AGV execution device (transfer category) is also a cross-device handover. After the CNC machining unit completes processing, the workpiece position is fixed, but the positioning accuracy of the AGV execution device may be affected by ground flatness and navigation errors, resulting in a deviation of ±5mm, which requires special verification.

[0046] The waiting position identifier, reverse access identifier, normal connection identifier, and cross-device connection relationship are written into the connection record area of ​​the corresponding process node. The connection record area is an additional information area for each process node in the process twin sequence, used to record the connection status and verification results of the process node, providing a basis for subsequent linkage identifier generation and virtual-real consistency verification.

[0047] Linkage identifiers are generated for process nodes with cross-device connection relationships. Specifically, process nodes with cross-device connection relationships are selected from the process node connection record area. In this embodiment, cross-device connection relationships include the handover relationship between the robot assembly unit and the warehousing unit, the handover relationship between the robot unit and the AGV execution device, the handover relationship between the CNC machining unit and the intelligent vision inspection unit, and the handover relationship between any other different equipment categories or different equipment units.

[0048] For each process node with cross-device connection relationships, the waiting position identifier, reverse access identifier, or normal connection identifier in its connection record area is read. When a waiting position identifier is read, a linkage freeze mark for the subsequent target device is generated. The linkage freeze mark is a control signal used to prevent the subsequent target device from starting prematurely or receiving new instructions, ensuring that the workpiece is allowed to flow only after the current device has completed workpiece release and the subsequent device is ready. The linkage freeze mark contains the freeze object identifier (i.e., the device identifier of the subsequent target device), the freeze start station identifier (i.e., the station currently occupied by the subsequent device), and the freeze duration condition (i.e., the current device's workpiece has not been released or the subsequent device's station is not free). For example: when the robot assembly unit has completed assembly but the intelligent vision inspection unit's station is occupied, a linkage freeze mark for the intelligent vision inspection unit is generated, the freeze object is VIS-DET-01, the freeze start station is inspection station D01, and the freeze duration condition is that the previous workpiece inspection has not been completed and the station is not free.

[0049] When a reverse access identifier is read, a bidirectional linkage freeze flag is generated for both the preceding and following target devices. This situation is considered an abnormal state, requiring both the preceding and following devices to be frozen simultaneously to prevent further deterioration. The bidirectional linkage freeze flag contains a retreat confirmation bit (requiring the following device to retreat to a safe position), a stop confirmation bit (requiring both the preceding and following devices to pause their actions), and a re-access bit (the access point to re-establish the connection relationship after the state returns to normal). For example, if the robot assembly unit has not yet completed but the intelligent vision inspection unit has started prematurely, a bidirectional linkage freeze flag is generated, requiring the intelligent vision inspection unit's conveyor belt to immediately retreat and stop, and the robot assembly unit to pause its current action and maintain workpiece clamping. The process resumes according to the re-access bit after manual confirmation of safety.

[0050] When a normal connection identifier is read and both the corresponding preceding and following target devices have completion confirmation records, a linkage release flag is generated. A completion confirmation record indicates that the preceding device's execution status segment contains an action completion flag and its response status segment contains an instruction release flag, while the following device's pose status segment shows the workstation is idle and its response status segment contains an instruction reception flag or a ready flag. The linkage release flag is a permission signal indicating that the virtual and physical states are consistent and the handover conditions are met, allowing the system to issue a linkage execution command to execute the workpiece flow. The linkage release flag contains the release object identifier (i.e., the combination of the preceding and following device identifiers), the access workstation identifier (i.e., the workstation number where the following device receives the workpiece), and the access duration (i.e., the time window during which workpiece flow is allowed; after this time window, the release flag becomes invalid and needs to be re-verified). In this embodiment, the access duration is set to 3 seconds. The basis for this setting is that the physical transfer of workpieces across devices usually takes 0.5 to 2 seconds. Considering the network communication delay of about 100ms to 300ms, setting a time window of 3 seconds can meet the normal flow requirements and avoid the risk caused by the release mark being valid for a long time due to sudden equipment failure.

[0051] The linkage freeze flag or linkage release flag is written into the node display area of ​​the digital twin interactive interface and the command gating area of ​​the field control link, respectively. The node display area is a visual area in the virtual environment that displays the status of process nodes. When a process node is marked with a linkage freeze flag, it is displayed as a red frozen state on the interface, indicating that it is waiting for the preceding sequence to complete or the following sequence to be ready. Operators cannot perform any operations on this process node on the interface. When a process node is marked with a linkage release flag, it is displayed as a green released state on the interface, indicating that it is allowed to proceed. Operators can perform subsequent operations on this process node on the interface. The command gating area is a command filtering module in the field control link. When a device is marked with a linkage freeze flag, the command gating area intercepts all execution commands sent to that device, allowing only safety-related shutdown commands and status query commands to pass through. When a device is marked with a linkage release flag, the command gating area allows linkage execution commands sent to that device, allowing commands to be normally issued to the device controller. By synchronously writing linkage flags at both the interface display and command control levels, the operability of the equipment on both the virtual and physical sides remains consistent.

[0052] It should be noted that the linkage identifier has a dynamic updating characteristic. When the equipment status changes, the system re-executes the position verification and connection determination, and updates the linkage identifier. For example, a certain process node is initially marked as a linkage freeze. When the subsequent equipment completes the processing of the previous workpiece and becomes idle, the system detects a change in the status of the subsequent equipment, re-verifies and finds that the connection conditions are met, then updates the linkage freeze identifier to a linkage release identifier, thereby achieving dynamic synchronization between the virtual and real states.

[0053] Execution module: Upon receiving a subsequent operation request, it calls the on-site status mirror and linkage identifier to perform a virtual-real consistency check, issues linkage execution instructions to the process nodes that pass the check, and constructs the full-process virtual-real linkage training results;

[0054] The core mechanism of this solution is the consistency verification between the virtual environment and the physical equipment. By comparing the interface display of the virtual environment with the on-site status of the physical equipment, control commands are only issued when the two are consistent, thus fundamentally eliminating erroneous operations caused by the asynchrony between the virtual and physical environments.

[0055] Specifically, the target process node identifier, target equipment identifier, and workpiece transfer identifier are extracted from the subsequent operation request. The subsequent operation request refers to the operation command initiated by the operator on the digital twin interactive interface, such as: initiating the detection action of process node N04, or transferring workpiece BATCH20240615-WP01 to the storage unit. The target process node identifier indicates which process node this operation targets, the target equipment identifier indicates which equipment(s) are involved in this operation, and the workpiece transfer identifier indicates which batch and which workpiece is being processed in this operation.

[0056] Retrieve the corresponding state fragment records from the field state mirror. Based on the target equipment identifier, search the unified time-series ledger for the latest execution state fragment, response state fragment, and pose state fragment of the equipment to obtain the actual state of the equipment in the field. Search the unified time-series ledger for the most recent completion confirmation record, the most recent pose landing point record, and the most recent workstation occupancy record corresponding to the target process node identifier to form a node reality group. The most recent completion confirmation record refers to the time and content of the most recent action completion identifier for the process node, used to determine whether the process node has been completed; the most recent pose landing point record refers to the most recently updated position and attitude information of the equipment corresponding to the process node, used to determine whether the equipment is in the expected position; the most recent workstation occupancy record refers to the most recently updated occupancy status of the workstation corresponding to the process node, used to determine whether the workstation is available. The node reality group represents the actual state of the physical equipment.

[0057] The interface presentation record corresponding to the target process node identifier is read from the digital twin interactive interface. The interface presentation record includes the workstation display status, action display status, and handover display status. The workstation display status refers to the workstation occupancy status displayed on the interface; for example, workstation P03 is displayed as occupied, with equipment RBT-ASM-01 occupying for 8 seconds. The action display status refers to the execution status of equipment actions displayed on the interface; for example, the robot assembly unit is displayed as executing, with a current progress of 85%. The handover display status refers to the workpiece handover status displayed on the interface; for example, workpiece BATCH20240615-WP01 is displayed as having been moved from the assembly unit and awaiting receipt by the inspection unit. This forms the interface mapping group required for this verification, and the interface mapping group represents the display status of the virtual environment.

[0058] The correspondence between the node real-time group and the interface mapping group in terms of workstation display status, action display status, and handover display status is compared to generate a consistency comparison result. The comparison process includes three dimensions:

[0059] The first dimension is workstation consistency comparison: compare the most recent workstation occupancy record in the node real-time group with the workstation display status in the interface mapping group to determine whether the actually occupied workstation is consistent with the workstation displayed on the interface. If the actually occupied workstation is P03 and the workstation displayed on the interface is also P03, then the workstation is consistent; if the actually occupied workstation is P03 but the workstation displayed on the interface is P04 or is displayed as idle, then there is a workstation misalignment.

[0060] The second dimension is the consistency comparison of action stages: comparing the most recent completion confirmation record in the node real-time group with the action display status in the interface mapping group to determine whether the actual action stage is consistent with the interface display stage. If the most recent completion confirmation record shows that the assembly action has been completed, but the interface action display status still shows that it is in progress at 85%, it means that the interface is lagging behind the actual situation, which is an inconsistency in action stages; if the most recent completion confirmation record shows that the assembly action is in progress, but the interface action display status shows that it has been completed, it means that the interface is ahead of the actual situation, which is an advance in action stages, a dangerous situation of inconsistency between reality and virtuality.

[0061] The third dimension is the consistency comparison during the handover phase: comparing the most recent pose landing point record in the node real-time group with the handover display status in the interface mapping group to determine whether the actual workpiece position is consistent with the workpiece flow status displayed on the interface. If the actual pose landing point record shows that the workpiece is still held in the assembly unit's robotic arm, but the interface handover display status shows that the workpiece has been moved out of the assembly unit, then there is a missing item in the handover phase, indicating that the handover flow displayed on the interface has not actually occurred.

[0062] The system reads the linkage freeze or linkage release flag currently held by the target process node. When the consistency comparison result corresponds to the same workstation, the same action stage, and the same handover stage, and the current linkage release flag is held, a verification record is generated. The verification record indicates that the virtual environment and the physical equipment are completely consistent, and the cross-equipment connection conditions have been met, allowing the issuance of control commands corresponding to subsequent operation requests.

[0063] When the consistency comparison result shows any of the following: workstation misalignment, premature action phase, or missing handover phase, or if a linkage freeze flag is currently in effect, a blocking verification record is generated. A blocking verification record indicates that the virtual environment and physical equipment status are inconsistent, or that cross-device connection conditions are not met. It prohibits issuing control commands corresponding to subsequent operation requests to prevent erroneous operations from causing equipment collisions or workpiece drops. For example: the interface displays that the intelligent vision inspection unit workstation is idle and can receive workpieces, but the actual status mirror shows that the inspection unit workstation is occupied by a previous workpiece for 18 seconds, with an inspection progress of 72%. The consistency comparison result shows a workstation misalignment (interface displays idle, but actual workstation is occupied), and a blocking verification record is generated.

[0064] Subsequent operation requests are moved to the confirmation area for independent verification. The confirmation area is a cached region in the system specifically for handling operation requests that fail the virtual-to-real consistency verification. These requests undergo deep verification and state synchronization to prevent normal operations from being erroneously blocked due to temporary network latency or lag in state updates.

[0065] In the pending confirmation area, independent confirmation entries are created for subsequent operation requests. These entries include the target process node identifier, target equipment identifier, workpiece flow identifier, and a blocking verification record, recording the reason and time when the operation request was blocked. Based on the workstation misalignment, advanced action phase, or missing handover phase in the blocking verification record, adjacent process node records from the unified time-series ledger are retrieved to form an adjacent verification chain. Adjacent process nodes refer to the preceding and succeeding nodes adjacent to the current process node in the process twin sequence. The adjacent verification chain is a status record chain containing the preceding, current, and succeeding nodes, used to comprehensively verify the status of each link in the process connection relationship. For example, when the blocking verification record of process node N04 (intelligent vision inspection unit performing inspection) shows workstation misalignment, the status records of process nodes N03 (robot assembly unit performing assembly) and N05 (AGV execution device performing transfer) are retrieved to form an adjacent verification chain of N03-N04-N05.

[0066] Based on the sequential relationship of adjacent verification chains, a handover re-interrogation operation is performed on the preceding target equipment, an execution position re-interrogation operation is performed on the current target equipment, and an access position re-interrogation operation is performed on the subsequent target equipment, generating supplementary confirmation records. The handover re-interrogation operation queries the controller of the preceding equipment to re-inquire whether the workpiece has been handed over, whether the handover position has been released, and whether the workpiece has left the preceding equipment's workstation. The execution position re-interrogation operation queries the controller of the current equipment to re-inquire about the execution status, progress, estimated completion time, and workstation occupancy of the current action. The access position re-interrogation operation queries the controller of the subsequent equipment to re-inquire whether the workstation is idle, ready to receive the workpiece, and whether the receiving mechanism is in position. The re-interrogation operation uses direct communication with the equipment controller, bypassing the potentially delayed status acquisition triggering mechanism, to obtain the real-time status of the equipment. The supplementary confirmation record is the latest status information obtained from the re-interrogation operation, used to update the field status image.

[0067] Write the supplementary confirmation record back to the confirmation entry, and update the connection record area of ​​the target process node based on the supplementary confirmation record. The update process compares the equipment status in the supplementary confirmation record with the status in the original connection record area. If the supplementary confirmation record shows that the equipment status has changed (e.g., the previously occupied workstation is now free, or the previously in-process action has been completed), then update the corresponding fields in the connection record area, and re-execute the position verification and connection judgment to generate a new connection identifier.

[0068] When the updated connection record area corresponds to a normal connection identifier, the confirmed entry is marked as a pending release entry. A pending release entry indicates that although the initial verification failed, after independent verification, it was found that the virtual and real states have been restored to consistency and the connection conditions have been met, so the operation request can be released for execution. The system removes the pending release entry from the pending confirmation area, generates a linkage release mark for it, and allows the issuance of the corresponding linkage execution command.

[0069] When the updated connection record area still corresponds to a waiting placeholder or reverse access identifier, the confirmed entry is marked as a continue-hold entry, and the linkage freeze mark is retained. A continue-hold entry indicates that after independent confirmation, the situation where the virtual and real are inconsistent or the connection conditions are not met still exists, and the operation request still needs to be blocked. The system will continue to hold the entry in the pending confirmation area, and re-perform independent confirmation every certain period of time (set to 500 milliseconds in this embodiment) until the connection conditions are met or the operator actively cancels the operation request.

[0070] It should be noted that the independent confirmation mechanism can both prevent normal operations from being mistakenly blocked due to network latency and ensure that dangerous operations with discrepancies between the real and virtual systems are continuously blocked, thus achieving a balance between security and availability.

[0071] Extract the target process node identifier from the verification record, and read the target equipment subset corresponding to the target process node based on the process twin sequence. The target equipment subset refers to all the equipment that needs to be linked to execute the process node. For example, the target equipment subset corresponding to process node N04, which transfers the workpiece from the assembly unit to the inspection unit, includes three devices: the robot assembly unit (responsible for releasing the workpiece), the AGV execution device (responsible for transferring the workpiece), and the intelligent vision inspection unit (responsible for receiving the workpiece).

[0072] For each target device in the subset of the target equipment set, generate an execution segment, a handover execution segment, and a occupancy release segment. The execution segment indicates the content of the current action, including action type, action parameters, and execution sequence. For example, for a robot assembly unit, generate an execution segment to release the gripper, release the workpiece to handover position P03, and return to a safe position. The handover execution segment indicates the continuation of workpiece flow, including workpiece handover sequence, handover confirmation method, and handover safety checks. For example, for an AGV actuator, generate a handover execution segment to receive the workpiece at handover position P03 in the assembly unit, confirm that the workpiece is stably placed on the AGV platform, and start the transfer to the detection unit. The occupancy release segment indicates the release content of the workstation resources, including workstation release time, workstation status update, and permission for the next workpiece to enter. For example, for a robot assembly unit, generate an occupancy release segment to release the occupancy status of workstation P03 after the workpiece leaves, allowing the reception of the next workpiece.

[0073] The equipment execution segment, handover execution segment, and occupancy release segment are combined into a coordinated execution instruction according to the sequence of the target process nodes. The coordinated execution instruction is a composite instruction involving the coordinated actions of multiple devices. The actions of each device are arranged according to their timing to ensure the continuity and safety of workpiece flow. Corresponding workstation occupancy identifiers and safety boundary identifiers are written into the coordinated execution instruction. The workstation occupancy identifier is used to mark all workstations involved in this coordinated execution, preventing other operation requests from occupying these workstations and causing conflicts during coordinated execution. The safety boundary identifier is used to define the safe operating range of each device, preventing equipment movement beyond the safety boundary and causing collisions.

[0074] The corresponding instruction segments of the linkage execution command are sent sequentially to each target device in the field control link, and after each instruction segment is sent, the corresponding target device is waited for a command lock flag to be returned. Sequential sending means that the commands are sent to each device in the order of workpiece flow. For example, the device execution segment and the position release segment are sent to the robot assembly unit first. After the robot assembly unit returns the command lock flag, the handover execution segment is sent to the AGV execution device. Finally, the receiving preparation command is sent to the intelligent vision inspection unit. This sequential sending method ensures that the preceding device is ready to release the workpiece before the subsequent device starts receiving, avoiding the confusion of the device action timing caused by sending commands simultaneously.

[0075] In the field control link, a corresponding arrival acknowledgment bit, lock bit, and execution start bit are established for each instruction segment. The arrival acknowledgment bit is used by the device controller to confirm that the instruction has been received, the lock bit is used by the device controller to confirm that the instruction has been loaded into the local execution queue and locked, and the execution start bit is used by the device controller to confirm that the instruction has started execution.

[0076] When the target device receives the corresponding command segment, it writes an arrival confirmation record to the arrival receipt bit. The arrival confirmation record includes the command reception time and the command content verification result (checking whether the command format is correct and whether the parameters are within the allowed range).

[0077] When the target device completes local takeover of the corresponding instruction segment, an instruction lock identifier is written to the lock bit, and a lock entry record is registered in the unified time-series ledger. Local takeover means that the device controller converts the instructions sent from the cloud into locally executable control code and loads it into the execution queue. At this time, the device enters a dedicated execution mode and no longer receives other instructions for the same process. The instruction lock identifier includes the lock time, lock instruction number, and estimated execution duration. The lock entry record is a state segment record for the corresponding device in the unified time-series ledger, recording the time and reason when the device enters the lock state.

[0078] When the target device starts executing the corresponding instruction segment, an execution entry record is written to the execution start position and associated with the target process node identifier. The execution entry record includes the execution start time, execution content description, and execution status monitoring. By associating with the target process node identifier, a mapping relationship between instruction execution and process twin sequence is established, which facilitates subsequent traceability and result recording.

[0079] If an arrival confirmation record exists at the arrival acknowledgment position but the lock position has not been written with the command lock flag, the corresponding command segment is marked as a pending command segment. A pending command segment indicates that the device has received the command but has not yet completed local takeover. This could be because the device controller is processing a previous command that has not yet been released, or it could be due to a processing delay in the device controller. For the pending command segment, the latest response status fragment in the field status mirror is invoked to check whether the corresponding target device has entered the command reception or command execution flag. If the response status fragment shows that the device has entered the command reception flag but not yet the command lock flag, it means that the device is processing the command, which is a normal delay. If the response status fragment shows that the device has entered the command execution flag, it means that the device has actually started execution but failed to update the lock position in time due to communication delay, which is a network delay. If the response status fragment shows that the device is still in an idle state or executing other commands, it means that the command segment was not successfully delivered or was rejected by the device, which is an abnormal situation.

[0080] Based on the inspection results, the pending command segment is rewritten as either a continue waiting segment or a return confirmation segment. If the delay is normal or due to network latency, the pending command segment is rewritten as a continue waiting segment, and the system continues to wait for the device to return a command lock flag. The waiting time is set to 2 seconds. This value is based on the fact that the device controller's command processing cycle is typically 200 milliseconds to 1 second, plus a network round-trip latency of approximately 100 milliseconds to 500 milliseconds. Setting a waiting time of 2 seconds can cover most normal delay situations. If the delay is abnormal, the pending command segment is rewritten as a return confirmation segment. The system stops waiting and withdraws the command segment from the field control link, records the reason and time of the command issuance failure, and issues an alarm to the operator.

[0081] The "Continue Waiting" segment remains in the field control link, while the "Return Acknowledgment" segment is written to the "Pending Acknowledgment" area. If the "Continue Waiting" segment receives a command lock flag within the waiting time, it transitions to normal execution; if no command lock flag is received after the waiting time expires, it also transitions to a "Return Acknowledgment" segment and is written to the "Pending Acknowledgment" area. Returned acknowledgment segments in the "Pending Acknowledgment" area require manual intervention to check the equipment status, troubleshoot, and reissue the command.

[0082] When all target devices return the command lock flag, the coordinated execution command is registered as an accessed command. An accessed command indicates that all command segments in the coordinated execution command have been successfully delivered to the corresponding devices and locked by the devices. The devices are about to begin or have already begun coordinated execution, and the workpiece flow enters the actual execution stage. The system displays the status of accessed commands in real time on the digital twin interactive interface, allowing operators to see the coordinated execution progress of each device.

[0083] When any target device fails to return a command lock flag, the linkage execution command is split into a device execution segment, a handover execution segment, and a placeholder release segment, and the unconnected command segment is written to the pending confirmation area. Because linkage execution requires coordinated action from all participating devices, if any device fails to connect, the entire linkage process cannot execute normally. The sent command segments need to be withdrawn to prevent incomplete linkage caused by some devices starting while others do not. After writing the unconnected command segments to the pending confirmation area, the system prompts the operator that some devices in the linkage execution command have not connected, and the linkage process has been paused. The operator should check the device [device identifier] status. The operator can view the reason for the non-connection in the pending confirmation area and handle it.

[0084] It should be noted that the issuance of the coordinated execution command adopts the principle of full locking before execution, ensuring the atomicity of multi-device collaboration and avoiding workpiece flow interruptions or idle equipment runs caused by partial execution by some devices and partial non-execution by others. In this embodiment, the maximum waiting time for the coordinated execution command is set to 5 seconds, which is the maximum allowed time from the issuance of the first command segment to the receipt of the locking flag by all command segments. If this time is exceeded, the coordinated execution command is considered to have failed to be issued. The basis for setting this value is: the waiting time for a single command segment is 2 seconds, and coordinated execution commands usually involve 2 to 3 devices. Considering the time interval between sequential sending, setting a maximum waiting time of 5 seconds can meet the coordinated execution requirements under normal circumstances.

[0085] Finally, a complete virtual-real integrated training result is constructed. The construction of the training result is a record and archive of the entire training process, providing data support for subsequent training evaluation, problem tracing and process optimization.

[0086] Specifically, the records document the pass verification records, block verification records, connected instructions, released pending items, continued holding items, and returned confirmation segments for each target process node during this training. These records comprehensively reflect the execution status of the virtual-real consistency verification, recording which process nodes successfully passed the verification and were executed, which process nodes were blocked due to virtual-real inconsistencies, which process nodes were released after independent confirmation, which process nodes were continuously held due to unmet conditions, and which instruction segments were returned due to equipment malfunctions.

[0087] Based on the sequential relationships of processes in the process twin sequence, a node result chain table is established for each target process node. The node result chain table arranges each process node according to the order of workpiece flow, with each process node corresponding to a single record. The record includes the process node identifier, corresponding equipment identifier, execution start time, execution end time, execution result (success / failure / interruption), verification result (pass / blocked / released / continued), and any exception information. The node result chain table clearly presents the execution trajectory of the entire training process.

[0088] Each node result in the node result chain table corresponds to at least one interface presentation record, one field status mirror record, and one control link record. The interface presentation record is the history of the display status changes of the process node on the digital twin interactive interface, including when the node is displayed as waiting, executing, or completed. The field status mirror record is a sequence of state fragment records for the corresponding equipment in the unified time-series ledger, including the complete change process of execution state fragments, response state fragments, and pose state fragments. The control link record is a complete record of the issuance, receipt, locking, and execution of the linkage execution command corresponding to the process node in the field control link. Through the correspondence and association of these three types of records, the synchronization between the virtual environment and the physical equipment at the process node can be compared and analyzed, revealing potential virtual-real discrepancies.

[0089] Each node result in the node result chain table is written to the training results page. The training results page is a comprehensive report page that presents the training results in a visual manner, including the workpiece flow trajectory area, equipment linkage trajectory area, workstation occupancy trajectory area, and instruction access trajectory area.

[0090] The workpiece flow trajectory area displays the flow process of the workpiece between various devices in a timeline manner, marking the entry time, processing / inspection time, and exit time of the workpiece in each device, as well as whether there are any abnormal situations such as waiting or interruption during the flow process. If waiting occurs at a certain stage during the flow process, the waiting time and the reason for waiting are marked on the timeline.

[0091] The equipment linkage trajectory area displays the linkage execution status of each device using a Gantt chart. The horizontal axis represents time, and the vertical axis represents the device list. The execution period of each device is represented by a color block, with different colors indicating different execution states (green for normal execution, yellow for waiting, and red for abnormal interruption). Overlapping areas of color blocks represent the collaborative linkage period of multiple devices. The Gantt chart provides a clear view of the load status of each device, the tightness of linkage, and whether there are any idle or overloaded devices.

[0092] The workstation occupancy trajectory area records the occupancy sequence of each workstation in a table format, listing the start time of occupancy, occupied equipment, occupied workpiece, occupancy duration, release time, and workstation occupancy rate (the proportion of occupancy duration to total training time) for each workstation during the training process. The workstation occupancy trajectory helps analyze the utilization efficiency of workstation resources and identify workstation bottlenecks.

[0093] The instruction access trajectory area displays the issuance and access process of coordinated execution instructions in a sequence diagram format. It lists the issuance time, target device, instruction segment content, arrival confirmation time, locking time, execution start time, and the time taken from issuance to locking, and from locking to execution for each coordinated execution instruction. The instruction access trajectory helps analyze network communication quality and device response speed, and identify problems such as communication latency or slow device response.

[0094] The training results page is linked and archived with a unified time-series ledger to form a traceable, interconnected virtual and physical training archive. Linking and archiving refers to establishing an index link between each record in the training results page and the corresponding record in the unified time-series ledger using key fields such as process node identifiers, equipment identifiers, and timestamps. This allows for quick location of detailed original data in the unified time-series ledger when viewing any record on the training results page, achieving complete traceability from result to process.

[0095] When the next training task instruction is issued, the naming rules for node identifiers, workstation occupancy identifiers, and workpiece flow identifiers in the virtual-physical linkage training archive are invoked as the registration basis for the next process twin sequence. This reuse of naming rules ensures that the identifier system of different batches of training tasks remains consistent, facilitating data comparison and analysis across batches. For example, if the node identifiers of this training use the naming rule of "N + two-digit sequence number" (N01, N02, N03...), and the workpiece flow identifiers use the naming rule of "BATCH + date + WP + two-digit workpiece number + STAGE + two-digit stage number", then the next training will also use this rule, only updating the date and sequence number, making the training data of different batches comparable.

[0096] This embodiment establishes a unified scheduling basis for virtual-physical linkage by constructing a process twin sequence. It records the entire process of physical equipment status changes through a unified time-series ledger, generates linkage identifiers to control cross-device handover through arrival verification and connection judgment, and ensures that control commands are issued only when the virtual and physical states are consistent through virtual-physical consistency verification. This fundamentally solves the problem of state synchronization distortion between the cloud virtual environment and the on-site physical equipment, avoiding safety accidents such as equipment collisions and workpiece drops caused by virtual-physical asynchrony. At the same time, compared with the traditional high-frequency query method, this solution adopts a state acquisition mechanism that combines event-driven and periodic acquisition, as well as a gating interception mechanism based on linkage identifiers, which effectively reduces the pressure on cloud servers and network bandwidth, and improves the security, reliability, and real-time performance of the digital twin training system.

[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fully virtual-real integrated training system for intelligent manufacturing based on digital twins, characterized in that: include: Sequence Construction Module: Obtains training task instructions and constructs corresponding process twin sequences based on the training task instructions. The process twin sequence includes the set of target equipment participating in this training, the process connection relationship, and the workpiece flow identifier. Image building module: Based on the target device set, collect the on-site execution status, control response status and pose confirmation status of each target device, generate corresponding status fragment records according to device category, write the status fragment records into a unified time-series ledger, and form the on-site status image of the current training cycle; The marking module performs on-site status mirroring to verify and determine the connection of each process node in the process twin sequence. It generates linkage markers for process nodes with cross-equipment connection relationships. The linkage markers include linkage freeze markers or linkage release markers. Execution module: Upon receiving a subsequent operation request, it calls the on-site status image and linkage identifier to perform virtual-real consistency verification, issues linkage execution instructions to the process nodes that pass the verification, and constructs the full-process virtual-real linkage training results.

2. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 1, characterized in that, Obtain the training task instructions and construct the corresponding process twin sequence based on the training task instructions, including: The process task field, equipment call field, and workpiece batch field in the training task instruction are analyzed. Robot units, warehousing units, CNC machining units, embodied intelligent robot units, robot assembly units, intelligent vision inspection units, and digital intelligent control units corresponding to the process task fields are selected from the digital vocational comprehensive training platform to generate a target equipment set. According to the workpiece flow sequence in the workpiece batch field, establish a process access position, process execution position and process handover position for each target device in the target device set, and form a cross-unit process connection relationship based on the device call field; For each process node in the process connection relationship, write the node entry condition, node retention condition and node release condition to form a verifiable process twin sequence; Each process node in the process twin sequence is associated with a visual object in the digital twin interactive interface and a control object in the field control link, so that the same process node uses the same node identifier in both interface display and field execution. The process twin sequence is loaded with training scenarios. The loading of training scenarios includes writing workpiece flow identifiers, target equipment sets and process connection relationships, and setting the loaded process twin sequence as the sole linkage basis for this training.

3. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 2, characterized in that, The methods for obtaining the on-site status image of the current training cycle include: Initiate status acquisition triggers to each target device in the target device set. The status acquisition triggers correspond to the local status output terminals of the robot motion controller, warehouse handling controller, CNC machining controller, vision inspection device, AGV execution device, and assembly execution device, respectively. The system reads the field execution status from each target device, including the action start flag, action in progress flag, action completion flag, and action exception flag, and encapsulates the reading results into an execution status fragment. It also reads the control response status from each target device, including the command receive flag, command lock flag, command execute flag, and command release flag, and encapsulates the reading results into a response status fragment. Finally, it reads the pose confirmation status from each target device, including the current position segment flag, pose landing point flag, workpiece attachment flag, and workstation occupancy flag, and encapsulates the reading results into a pose status fragment. Based on the equipment category of the target equipment, the execution state segments, response state segments, and pose state segments are merged and registered, so that the same target equipment forms a set of state segment records within the same training cycle; the state segment records are written into a unified time-series ledger in the order of entry, and the corresponding equipment identifier, process node identifier, and workpiece flow identifier are written into each set of state segment records in the unified time-series ledger to form a field state mirror.

4. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 3, characterized in that, Based on the on-site status mirror, the execution status verification and connection determination of each process node in the process twin sequence are performed, including: Retrieve the state fragment record corresponding to the current process node and the state fragment record of the subsequent node with the same workpiece flow identifier from the unified time-series ledger corresponding to the field state mirror. The state fragment record includes: execution state fragment, response state fragment and pose state fragment. Based on the execution state fragments, response state fragments, and pose state fragments of the current process node and subsequent nodes, the status of the current target device and the status of the subsequent target device are verified, and the process node verification results are generated. When the verification result of a process node indicates that the current target device has completed the process node and the subsequent target device has not yet met the entry conditions, a waiting position flag is established for the process node; when the verification result of a process node indicates that the current target device has not completed the process node but the subsequent target device has already initiated an access action, a reverse access flag is established for the process node; when the verification result of a process node indicates that the current target device has completed the process node and the subsequent target device has already initiated an access action, a normal connection flag is established for the process node. The process node is analyzed to determine the preceding target equipment, current target equipment, and subsequent target equipment in the process connection relationship. When the preceding target equipment, current target equipment, and subsequent target equipment belong to different equipment categories or equipment units, the current process node is determined to be a process node with cross-equipment connection relationship. Write the waiting position identifier, reverse access identifier, normal connection identifier, and cross-device connection relationship into the connection record area of ​​the corresponding process node.

5. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 4, characterized in that, Generate linkage markers for process nodes that have cross-device connection relationships, including: Select process nodes with cross-equipment connection relationships from the process node connection record area. Cross-equipment connection relationships include the handover relationship between robot assembly unit and warehousing unit, the handover relationship between robot unit and AGV execution device, and the handover relationship between CNC machining unit and intelligent vision inspection unit. For each process node with cross-device connection relationships, read the waiting position identifier, reverse access identifier, or normal connection identifier from its connection record area; when a waiting position identifier is read, generate a linkage freeze mark for the subsequent target device, and write the freeze object identifier, freeze start station identifier, and freeze duration condition into the linkage freeze mark; when a reverse access identifier is read, generate a two-way linkage freeze mark for both the preceding and subsequent target devices, and write the return confirmation bit, stop confirmation bit, and reconnection bit into the two-way linkage freeze mark; when a normal connection identifier is read and both the corresponding preceding and subsequent target devices have completion confirmation records, generate a linkage release mark, and write the release object identifier, access station identifier, and access duration into the linkage release mark; Write the linkage freeze mark or linkage release mark into the node display area of ​​the digital twin interactive interface and the command gating area of ​​the field control link, respectively.

6. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 5, characterized in that, Upon receiving a subsequent operation request, the system invokes the on-site status mirror and linkage identifier to perform a virtual-to-real consistency check, including: Extract the target process node identifier, target equipment identifier, and workpiece flow identifier from the subsequent operation request, and retrieve the corresponding status fragment records from the on-site status mirror; Retrieve the most recent completion confirmation record, the most recent pose landing record, and the most recent workstation occupancy record corresponding to the target process node identifier in the unified time-series ledger to form a node status group; Read the interface presentation record corresponding to the target process node identifier in the digital twin interactive interface. The interface presentation record includes the workstation display status, action display status and handover display status, forming the interface mapping group required for this verification. Compare the correspondence between the node real-time group and the interface mapping group in terms of workstation display status, action display status and handover display status, and generate consistency comparison results; Read the linkage freeze mark or linkage release mark currently held by the target process node. When the consistency comparison result corresponds to the same workstation, the same action stage and the same handover stage, and the linkage release mark is currently held, generate a verification record. When the consistency comparison result shows any of the following situations: workstation misalignment, action stage ahead of schedule, or missing items in the handover stage, or when a linkage freeze mark is currently held, a blocking verification record is generated, and subsequent operation requests are transferred to the confirmation area for independent confirmation.

7. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 6, characterized in that, Subsequent operation requests will be transferred to the confirmation area for independent confirmation, including: In the pending confirmation area, create independent confirmation entries for subsequent operation requests. Confirmation entries include target process node identifier, target equipment identifier, workpiece flow identifier, and blockage verification record. Based on the workstation misalignment, action phase advancement, or missing items in the handover phase in the blocking verification record, the records of adjacent process nodes in the unified time sequence ledger are retrieved to form an adjacent verification chain. According to the order of adjacent verification chains, a handover re-interrogation operation is performed on the preceding target device, a bit re-interrogation operation is performed on the current target device, and an access bit re-interrogation operation is performed on the following target device to generate a supplementary confirmation record. Write the supplementary confirmation record back to the confirmation entry, and update the connection record area of ​​the target process node based on the supplementary confirmation record; when the updated connection record area corresponds to the normal connection mark, mark the confirmation entry as a release pending entry; when the updated connection record area still corresponds to the waiting placeholder mark or the reverse access mark, mark the confirmation entry as a continue holding entry, and retain the linkage freeze mark.

8. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 7, characterized in that, For the process nodes that have passed the verification, issue linkage execution instructions, including: Extract the target process node identifier from the verification record, and read the target equipment set subset corresponding to the target process node based on the process twin sequence; For each target device in the subset of the target device set, generate a device execution segment, a handover execution segment, and a space release segment. The device execution segment is used to indicate the execution content of the current action, the handover execution segment is used to indicate the continuation content of the workpiece flow, and the space release segment is used to indicate the release content of the workstation resources. The equipment execution segment, handover execution segment, and occupancy release segment are combined into a linkage execution command according to the sequence of the target process nodes, and the corresponding workstation occupancy identifier and safety boundary identifier are written into the linkage execution command; the corresponding instruction segment of the linkage execution command is sent to each target device in the field control link in sequence, and after each instruction segment is sent, the corresponding target device is waited for the return of the instruction lock identifier; When all target devices return the command lock flag, the linkage execution command is registered as an access command; when any target device does not return the command lock flag, the linkage execution command is split back into the device execution segment, handover execution segment, and placeholder release segment, and the unaccessed command segment is written into the pending confirmation area.

9. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 8, characterized in that, After each command segment is issued, wait for the corresponding target device to return a command lock flag, including: In the field control link, establish corresponding arrival acknowledgment bit, lock bit, and execution start bit for each instruction segment; When the target device receives the corresponding instruction segment, it writes an arrival confirmation record in the arrival receipt position; when the target device completes local takeover of the corresponding instruction segment, it writes an instruction lock identifier in the lock position and registers a lock entry record in the unified time sequence ledger; when the target device starts executing the corresponding instruction segment, it writes an execution entry record in the execution start position and associates the execution entry record with the target process node identifier. If an arrival confirmation record has been created at the arrival acknowledgment position but no command lock identifier has been written at the lock position, the corresponding command segment is marked as a pending command segment. For the pending command segment, the latest response status segment in the field status mirror is called to check whether the corresponding target device has entered the command reception or command execution stage. Based on the check result, the pending command segment is rewritten as a continue waiting segment or a return confirmation segment. The continue waiting segment is retained in the field control link, and the return confirmation segment is written to the pending confirmation area.

10. The intelligent manufacturing full-process virtual-real linkage training system based on digital twins according to claim 9, characterized in that, Construct a fully integrated virtual and real-world training outcome, including: Record the pass verification records, block verification records, connected instructions, released pending items, continued holding items, and return confirmation sections for each target process node in this training. Based on the sequential relationship of the processes in the process twin sequence, a node result chain table is established for each target process node. Each node result in the node result chain table corresponds to at least one interface presentation record, one field status mirror record, and one control link record. Write the result of each node in the node result chain table into the training result page. The training result page includes the workpiece flow trajectory area, the equipment linkage trajectory area, the workstation occupation trajectory area, and the instruction access trajectory area. The training results page is linked and sealed with the unified time-series ledger to form a traceable virtual-physical linkage training archive; and when the next training task instruction is entered, the node identifier naming rules, workstation occupancy identifier naming rules, and workpiece flow identifier naming rules in the virtual-physical linkage training archive are called as the registration basis for the next process twin sequence.

Images