AI-driven digital twin hybrid flow production real-time scheduling method and system

Abstract

This application relates to the field of manufacturing process management technology and discloses an AI-driven digital twin real-time scheduling method and system for mixed-flow production. The method collects real-time data streams from the production site, performs layered analysis of equipment operating parameters and material inventory status, and generates equipment fault identification results and inventory status results. Based on these results, it determines disturbances and their impact range; based on the impact range, it identifies affected production tasks and resource shortage requirements; it generates and optimizes reconfigurable paths by combining equipment allocation priority rules, material flow optimization rules, path constraints, and path conflict detection mechanisms; it issues adjustment instructions based on the prioritized execution paths and determines production recovery confirmation results based on execution feedback. This application improves the recovery efficiency and execution stability of mixed-flow manufacturing process management.

Description

Technical Field

[0001] This application relates to the field of manufacturing process management technology, and more specifically, to an AI-driven digital twin mixed-flow production real-time scheduling method and system. Background Technology

[0002] Manufacturing process management is a crucial foundation for ensuring production continuity, resource coordination, and task controllability in modern discrete and process manufacturing. In mixed-flow production scenarios, different product tasks are typically executed in parallel on the same production line, with overlapping occupancy relationships between equipment, material, and transfer resources. The equipment status, material supply status, workstation execution status, and task switching status during the manufacturing process are interconnected. When equipment malfunctions, material supply malfunctions, or changes in the processing capacity of local workstations occur on the production floor, the abnormal status not only affects the currently executing workstation but also continues to propagate along the task connection relationship to subsequent processes, thereby having a ripple effect on task progress, resource replenishment, and cycle time recovery during the manufacturing process.

[0003] While existing manufacturing process management methods can complete routine production scheduling and local adjustments based on preset plans, single-point alarm information, or partial inventory information, they often struggle to uniformly correlate changes in equipment operating status, material availability, and task execution status when sudden disturbances occur in mixed-flow production. On one hand, manufacturing processes involve instantaneous fluctuations, short-term replenishment delays, and single-cycle parameter deviations. Existing methods often fail to distinguish between genuine disturbances and short-term anomalies, easily misusing abnormal signals that shouldn't be included in the manufacturing process adjustment process as scheduling criteria, or failing to promptly identify disturbances that have actually affected production task execution, leading to inaccurate assessment of the impact range in the manufacturing process.

[0004] On the other hand, when affected tasks need to be resumed, existing manufacturing process management methods usually make alternative decisions based only on locally available equipment or local material reserves. They lack a continuous evaluation process for affected production tasks and find it difficult to simultaneously combine the availability of resources, processing capacity, process adaptation, material arrival time, transfer and occupation relationships, and task switching constraints to uniformly screen, compare, and coordinate the execution of manufacturing process recovery plans. Therefore, even if alternative plans are formed, problems such as mismatched resource arrival times, repeated occupation of resources in the same period, inability to directly issue recovery paths for execution, or difficulty in timely recovery of processing rhythm after execution can easily occur, thus affecting the continuity and stability of manufacturing process management.

[0005] Therefore, accurately identifying the actual affected objects when equipment or material disturbances occur during mixed-flow manufacturing, further determining the affected production tasks and resource shortage needs, and forming a manufacturing process recovery plan that can be directly executed and whose recovery results can be verified, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, this application provides the following technical solution: In the first aspect, this application discloses an AI-driven real-time scheduling method for digital twin mixed-flow production, including: Collect real-time data streams from the production site, and perform time-based correction and validity verification on equipment operating parameters and material inventory status to generate equipment fault identification results and inventory status results; Based on the fault persistence status in the equipment fault identification results, determine the equipment abnormal signal; based on the inventory status in the inventory status results, determine the material abnormal signal; perform continuity filtering and consistency filtering on the equipment abnormal signal and the material abnormal signal; and determine the scope of influence based on the equipment abnormal signal and the material abnormal signal retained after filtering. The affected production tasks are determined based on the scope of impact, and the resource shortage demand is determined by combining the current resource availability assessment results and historical resource allocation records; Based on the resource shortage demand results, combined with equipment allocation priority rules, material flow optimization rules, path constraints and path conflict detection mechanisms, a set of reconfigurable paths is generated; Feasibility verification is performed on the set of reconfigurable paths. Reconfigurable paths that have been verified in terms of equipment resources, material resources, and transfer resources are retained. For each verified reconfigurable path, the equipment preparation time, material transfer time, task switching time, and cycle time recovery time are determined. The production recovery time is estimated by combining the paths according to their actual occurrence order. Based on the production recovery time estimate and the current resource availability assessment, the priority execution path is determined. Generate a list of conflict-free operation instructions based on the priority execution path; issue adjustment instructions to the production line control system based on the list of conflict-free operation instructions, and determine the production recovery confirmation result based on the execution feedback.

[0007] Secondly, this application discloses an AI-driven digital twin mixed-flow production real-time scheduling system, including: The data parsing module is used to collect real-time data streams from the production site, and to perform time-based correction and validity verification of equipment operating parameters and material inventory status, generating equipment fault identification results and inventory status results. The disturbance determination module is used to determine the equipment abnormal signal based on the fault persistence status in the equipment fault identification result, determine the material abnormal signal based on the inventory status in the inventory status result, perform continuity filtering and consistency filtering on the equipment abnormal signal and the material abnormal signal, and determine the scope of influence based on the equipment abnormal signal and the material abnormal signal retained after filtering. The shortage determination module is used to identify affected production tasks based on the scope of impact, and to determine the resource shortage demand results by combining the current resource availability assessment results and historical resource allocation records; The path generation module is used to generate a set of reconfigurable paths based on the resource shortage demand results, combined with equipment allocation priority rules, material flow optimization rules, path constraints, and path conflict detection mechanisms. The path selection module is used to verify the feasibility of the set of reconfigurable paths, retain the reconfigurable paths that have been verified in terms of equipment resources, material resources, and transfer resources, and determine the equipment preparation time, material transfer time, task switching time, and cycle time recovery time for the verified reconfigurable paths. It combines these into a production recovery time estimate based on the actual occurrence order, and determines the priority execution path based on the production recovery time estimate and the current resource availability assessment. The recovery execution module is used to generate a list of conflict-free operation instructions based on the priority execution path; issue adjustment instructions to the production line control system based on the list of conflict-free operation instructions; and determine the production recovery confirmation result based on the execution feedback.

[0008] Compared with related technologies, this application has the following advantages: This application addresses the manufacturing process management needs in mixed-flow production scenarios. It tackles the problems of inaccurate identification of real disturbances, unclear definition of affected objects, and difficulty in directly implementing recovery plans. It establishes a continuous processing chain from on-site status analysis, disturbance filtering, determination of impact scope, determination of resource shortage needs, generation and optimization of recovery paths, to execution of adjustment instructions and confirmation of recovery. This integrates equipment-side status, material-side status, task-side impact, and resource-side reorganization into the same manufacturing process management closed loop, improving the process recovery efficiency and execution stability of mixed-flow production under sudden disturbance conditions.

[0009] This application performs hierarchical analysis of equipment operating parameters and material inventory status in the real-time data stream of the production site, and combines abnormal signal filtering to generate the scope of influence. This ensures that the disturbance objects entering the manufacturing process adjustment process are based on continuous state changes and consistency verification, thereby reducing the interference of instantaneous fluctuations, isolated anomalies and invalid inventory changes on the manufacturing process management results, improving the accuracy of real disturbance identification and the reliability of the determination of the affected objects.

[0010] This application determines the affected production tasks based on the scope of impact and determines the resource shortage demand results by combining the current resource availability assessment results and historical resource allocation records. This allows the disturbance state in the manufacturing process to be further transformed into a resource gap description directly corresponding to task recovery. In this way, a stable data connection relationship is established between the anomaly identification results and the manufacturing process recovery scheduling, thereby improving the pertinence of determining the direction, quantity, and timing of resource replenishment.

[0011] This application generates a set of reconfigurable paths by combining equipment allocation priority rules, material flow optimization rules, path constraints, and path conflict detection mechanisms. It then performs feasibility verification and production recovery time estimation on the set of reconfigurable paths, ensuring that candidate recovery paths not only meet resource category and quantity requirements but also equipment availability, material availability, process connection, and resource occupancy requirements. This reduces the number of unexecutable paths that end up in the final manufacturing process management results, thereby improving the feasibility and recovery efficiency of recovery paths.

[0012] This application generates a list of conflict-free operation instructions based on the priority execution path, issues adjustment instructions to the production line control system based on the list of conflict-free operation instructions, and then determines the production recovery confirmation result based on the execution feedback. This enables the manufacturing process management results to continue to be transmitted to the control execution layer and complete the recovery status verification, thereby forming a closed-loop processing process from recovery plan generation to on-site execution and then to recovery result confirmation. This reduces the problem of disconnect between manufacturing process decision-making and on-site execution, and improves the consistency and verifiability of mixed-flow manufacturing process management. Attached Figure Description

[0013] Figure 1 A schematic diagram of the AI-driven digital twin mixed-flow production real-time scheduling method provided in this application; Figure 2 A schematic diagram of the AI-driven digital twin mixed-flow production real-time scheduling system module provided in this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] Example 1

[0016] Please see Figure 1 As shown, this embodiment provides an AI-driven real-time scheduling method for mixed-flow production using digital twins, including the following steps: The system processes real-time data streams from the production site, collects these data streams, and performs time-based correction and validity checks on equipment operating parameters and material inventory status to generate equipment fault identification results and inventory status results.

[0017] In some implementations, the steps for generating equipment fault identification results and inventory status results include: Step 101: Generate the production line digital twin mapping result. Specifically, read the production line static configuration data, which includes at least the correspondence between equipment identifiers and workstation identifiers, the correspondence between material identifiers and feeding workstations, the correspondence between workstation identifiers and process steps, the correspondence between process steps and production task types, the connectivity between transfer nodes, and the addresses of each control interface; then read the production task plan within the current scheduling cycle, which includes at least the task identifier, task type, current process position, planned start time, and planned end time; and then combine the data from the production line static configuration data... Equipment identifiers, material identifiers, workstation identifiers, process steps, and control interface addresses are mapped item by item to the task type and current process position in the production task plan to generate a production line digital twin mapping result. The production line digital twin mapping result includes at least the mapping relationship between equipment and workstation, the mapping relationship between materials and feeding workstation, the mapping relationship between workstation and task, the mapping relationship between transfer nodes and transfer paths, and the mapping relationship between control interfaces and equipment objects. This production line digital twin mapping result is used in subsequent steps to determine the scope of impact, generate a set of reconfigurable paths, calculate the production recovery time estimate, and issue adjustment instructions to the corresponding control interfaces.

[0018] Step 102: Collect real-time data streams from the production site. Specifically, read equipment operating parameters from the equipment control interface and equipment sensor interface, and read material inventory status from the warehouse management interface and logistics acquisition interface. Equipment operating parameters include at least equipment identification, acquisition time, start / stop status, spindle speed, drive current, key component temperature, vibration amplitude, and current processing cycle. Material inventory status includes at least material identification, acquisition time, current location, current quantity, reserved quantity, in-transit quantity, most recent replenishment arrival time, and recent requisition records. Merge the equipment operating parameters and material inventory status collected within the same scheduling processing cycle into the same batch of real-time production site data streams. The scheduling processing cycle is set based on the scheduling refresh frequency, and the setting method is to ensure that it is no greater than the larger of the equipment status refresh cycle and the inventory status refresh cycle, thereby ensuring that the latest status of both the equipment and materials can be obtained simultaneously within one scheduling processing cycle.

[0019] Step 103: Time-normalize the real-time data stream from the production site. Specifically, determine a unified sampling period based on the smaller of the equipment acquisition cycle and the inventory update cycle. Using the unified sampling period as a benchmark, time-align the equipment operating parameters and material inventory status in the same batch of real-time data streams from the production site. For missing records, read the same source records from the previous and next unified sampling cycles. When the change directions of the two same source records are consistent, perform linear interpolation according to their time positions. When the change directions of the two same source records are inconsistent, retain the valid record closest to the current sampling time. For duplicate records, retain the record with the newest acquisition time. For records with reversed time sequence, rearrange them according to the acquisition time. After time normalization, the equipment operating parameters and material inventory status are obtained arranged according to the unified sampling period.

[0020] Step 104: Verify the validity of the equipment operating parameters and generate equipment fault identification results. Specifically, based on the production line digital twin mapping results, extract the spindle speed, drive current, key part temperature, vibration amplitude and current processing cycle of the corresponding equipment in multiple consecutive unified sampling periods according to the equipment identifier. The system reads the rated operating range, stable operating interval, and historical fault samples of the corresponding equipment. The stable operating interval is statistically formed based on historical sampling records of the corresponding equipment during continuous stable production, without alarms or downtime, and with processing cycle time within the standard range. The historical fault samples are extracted from the sampling sequences of confirmed faults in the historical alarm records, downtime records, and maintenance conclusions of the corresponding equipment. The system first compares the current equipment operating parameters with the rated operating range and stable operating interval item by item. When a parameter exceeds the rated operating range and a continuous change trend cannot be formed in the preceding and following unified sampling periods, the record corresponding to that parameter is determined as an invalid record and is removed. When a parameter does not exceed the rated operating range but continuously deviates from the stable operating interval, it is retained for fault identification. The system then calculates the direction of change, the magnitude of change, and the duration of change for the retained parameters. The direction of change is determined based on the increase or decrease relationship of adjacent unified sampling periods, the magnitude of change is determined based on the difference between the current value and the center value of the stable operating interval, and the duration of change is determined based on the number of unified sampling periods that continuously deviate from the stable operating interval.

[0021] If the current parameter combination and a certain type of historical fault sample have the same direction of change in at least two key parameters, and the duration reaches the continuous judgment condition, then the equipment is identified as the corresponding fault category. The continuous judgment condition is set according to the equipment's standard processing cycle, parameter inertia, and allowable false alarm upper limit. The invocation method is to compare whether the current duration has reached the number of consecutive abnormal samplings determined by the aforementioned factors. The above processing is performed on all equipment to generate equipment fault identification results. The equipment fault identification results include at least the equipment identifier, fault category, fault occurrence time, and fault duration status.

[0022] In some implementations, to illustrate the generation process of the equipment fault identification result in step 104, for example, for a processing equipment identified as E03, the spindle speeds read in four consecutive uniform sampling periods were 2980 rpm, 2860 rpm, 2710 rpm, and 2590 rpm, respectively; the drive currents were 41 A, 45 A, 49 A, and 53 A, respectively; the critical component temperatures were 68°C, 72°C, 77°C, and 81°C, respectively; the vibration amplitudes were 0.42 mm, 0.51 mm, 0.63 mm, and 0.76 mm, respectively; and the current processing cycle times were 31 seconds, 33 seconds, 36 seconds, and 39 seconds, respectively. The corresponding center values ​​of the stable operating range of the equipment were spindle speeds of 30 rpm, 2980 rpm, 2860 rpm, 2710 rpm, and 2590 rpm, respectively. The parameters are: 00 rpm, drive current 40 amps, critical component temperature 66 degrees Celsius, vibration amplitude 0.40 mm, and machining cycle time 30 seconds. Based on these data, it can be determined that the spindle speed continuously decreases, while the drive current, critical component temperature, vibration amplitude, and current machining cycle time continuously increase. Furthermore, the two key parameters, spindle speed and drive current, show the same direction of change as historical spindle jamming fault samples, with continuous deviations lasting for four unified sampling cycles. When the continuous judgment condition corresponding to this equipment is invoked, it is determined that the continuous deviation has reached the continuous abnormal sampling number requirement, thereby generating equipment fault identification results, including: equipment identifier E03, fault category as spindle jamming fault, fault occurrence time as the time corresponding to the first abnormal sampling cycle, and fault duration status as ongoing.

[0023] Step 105: Verify the validity of the material inventory status and generate inventory status results. Specifically, based on the production line digital twin mapping results, extract the current location, current quantity, pre-allocated quantity, in-transit quantity, most recent replenishment arrival time, and requisition record of the corresponding material within multiple consecutive unified sampling periods according to the material identifier; read the smallest unit of measurement, normal inventory change direction, and most recent replenishment record of the corresponding material; first, check the current quantity, pre-allocated quantity, and in-transit quantity against the most recent replenishment record and requisition record item by item; when the quantity change direction is inconsistent with both the replenishment record and the requisition record, the quantity change is judged as an invalid change and removed; when the quantity change direction can be explained by the replenishment record or the requisition record, it is retained and included in the inventory status calculation; then, perform calculation on the retained inventory status: determine the available quantity by subtracting the pre-allocated quantity from the current quantity and adding the in-transit quantity that can arrive in the next scheduling cycle; the formula is as follows: ; in, Indicates the available quantity. Indicates the current quantity. Indicates the quantity reserved. This indicates the number of items that are en route and can be delivered within the next scheduling cycle.

[0024] Based on the cumulative usage and corresponding duration of the most recent consecutive unified sampling periods, the usage per unit time is determined; when the usage per unit time is zero, the expected exhaustion time is set as a safe time after the current scheduling processing period; when the usage per unit time is greater than zero, the expected exhaustion time is determined by adding the available quantity to the current time and dividing by the usage per unit time to obtain the remaining duration.

[0025] ; in, Indicates the expected time of exhaustion. Indicates the current moment. Indicates the available quantity. This indicates the amount used per unit of time.

[0026] Next, compare the expected depletion time with the most recent replenishment time; when the expected depletion time is earlier than the most recent replenishment time, the inventory status is determined to be near shortage; when the available quantity is already lower than the expected requisition quantity in the next scheduling cycle, the inventory status is determined to be at risk of supply disruption; in other cases, the inventory status is determined to be continuous supply; perform the above processing on all materials to generate inventory status results; the inventory status results include at least the material identifier, available quantity, expected depletion time, and inventory status.

[0027] Step 106: Output the production line digital twin mapping results, equipment fault identification results, and inventory status results. Specifically, output the production line digital twin mapping results, equipment fault identification results, and inventory status results within the same scheduling processing cycle. Among them, the equipment fault identification results and inventory status results are used for subsequent steps to determine disturbances, and the production line digital twin mapping results are used for subsequent steps to determine the scope of impact.

[0028] For example, to illustrate the process of generating the inventory status result in step 105, for material with material identifier M12, the current quantity is read as 180 units, the pre-reserved quantity as 50 units, and the quantity in transit that can be delivered in the next scheduling cycle as 40 units, then the corresponding available quantity is 170 units; then, the requisition records for the most recent 10 minutes are read. If the cumulative requisition quantity is 100 units, then the requisition quantity per unit time is 10 units / minute; when the current time is 10:00, the remaining time can be determined to be 17 minutes based on the available quantity and the requisition quantity per unit time. Therefore, the expected depletion time is determined to be 10:17. If the most recent replenishment time is 10:25, the expected depletion time is earlier than the replenishment time. Therefore, the material is determined to be in a near-shortage state, and an inventory status result is generated. The inventory status result includes at least: material identifier M12, available quantity 170 units, expected depletion time 10:17, and inventory status is near-shortage state. As another example, if the corresponding available quantity is only 6 units, and the expected requisition quantity in the next scheduling cycle is 12 units, the inventory status can be directly determined to be a supply disruption risk state.

[0029] Based on the fault persistence status in the equipment fault identification results, determine the equipment abnormal signal; based on the inventory status in the inventory status results, determine the material abnormal signal; perform continuity filtering and consistency filtering on the equipment abnormal signal and the material abnormal signal; and determine the scope of influence based on the equipment abnormal signal and the material abnormal signal retained after filtering.

[0030] In some implementations, the steps for determining the scope of influence include: Step 201: Read the equipment fault identification results, inventory status results, and production line digital twin mapping results. Specifically, read the equipment identifier, fault type, fault occurrence time, and fault duration from the output equipment fault identification results; read the material identifier, available quantity, expected depletion time, and inventory status from the output inventory status results; and read the mapping relationship between equipment and workstation, material and feeding workstation, workstation and task, and transfer node and transfer path from the output production line digital twin mapping results.

[0031] Step 202: Generate abnormal signals based on equipment fault identification results and inventory status results. Specifically, equipment fault identification results with a persistent fault status are identified as equipment abnormal signals; inventory status results with a near-shortage status or a supply disruption risk status are identified as material abnormal signals. For equipment abnormal signals, compare the time of fault occurrence with the start and end times of the current scheduling cycle. If the time of fault occurrence is within the current scheduling cycle, or if the time of fault occurrence is earlier than the current scheduling cycle, and the persistent fault status has not been resolved, retain the equipment abnormal signal. For material abnormal signals, compare the time of expected depletion with the end time of the current scheduling cycle. If the time of expected depletion is earlier than the end time of the current scheduling cycle, retain the material abnormal signal. After the above judgment, the abnormal signals participating in the disturbance determination are obtained.

[0032] Step 203: Perform continuous filtering on abnormal signals. Specifically, for each equipment abnormal signal, read the spindle speed, drive current, key component temperature, vibration amplitude, and current processing cycle of the corresponding equipment within multiple consecutive unified sampling periods before and after the fault occurrence. If the above parameters deviate from the stable operating range only within a single unified sampling period, and return to the stable operating range in adjacent unified sampling periods, then remove the equipment abnormal signal. For each material abnormal signal, read the changes in the available quantity and the unit time consumption of the corresponding material within multiple consecutive unified sampling periods before and after the expected depletion time. If the available quantity only experiences a single short-term decrease, and the subsequent unified sampling period returns to a continuous supply state, then remove the material abnormal signal. The number of consecutive unified sampling periods in continuous filtering is set based on the equipment's natural fluctuation cycle and the material consumption refresh cycle. The method of calling it is to select a sampling quantity of not less than one complete fluctuation cycle within the current scheduling processing cycle for judgment.

[0033] This example illustrates the continuous filtering process in step 203. For instance, for an abnormal signal of device E05, the temperatures of key components are 69 degrees Celsius, 83 degrees Celsius, 70 degrees Celsius, 69 degrees Celsius, and 70 degrees Celsius in five consecutive uniform sampling periods, and the drive currents are 39 amps, 52 amps, 40 amps, 39 amps, and 40 amps, respectively. If only the second uniform sampling period deviates from the stable operating range, and the uniform sampling periods before and after it return to the stable operating range, then the abnormal signal of the device is determined to be a single short-term abnormality and removed. For equipment anomaly signals with equipment identification E06, within five consecutive unified sampling periods, the spindle speeds were 3010 rpm, 2890 rpm, 2760 rpm, 2650 rpm, and 2630 rpm, respectively; the drive currents were 40 A, 44 A, 47 A, 50 A, and 51 A, respectively; and the current processing cycle times were 30 seconds, 32 seconds, 35 seconds, 37 seconds, and 38 seconds, respectively. If the equipment deviates from the stable operating range for four consecutive unified sampling periods, the equipment anomaly signal is retained for the next step of consistency filtering. Material anomaly signals can be handled in the same way. If the available quantity drops from 82 pieces to 46 pieces only within a single unified sampling period and then recovers to above 80 pieces, the material anomaly signal is removed. If the available quantity continuously decreases from 82 pieces, 61 pieces, 39 pieces, and 18 pieces within four consecutive unified sampling periods, and the inventory status remains consistently near shortage, the material anomaly signal is retained.

[0034] Step 204: Perform consistency filtering on abnormal signals. Specifically, for each equipment abnormal signal, read the equipment start / stop status, equipment alarm record, and current processing cycle change corresponding to the same scheduling time. When the fault category is consistent with at least one of the start / stop status change, alarm record content, or processing cycle decrease direction, retain the equipment abnormal signal. For each material abnormal signal, read the outbound record, replenishment record, and requisition record corresponding to the same scheduling time. When the inventory status change is consistent with at least one of the outbound direction, replenishment status, or requisition change, retain the material abnormal signal. The abnormal signals retained after continuity filtering and consistency filtering are determined to be the abnormal signals corresponding to the real disturbance.

[0035] In some implementations, inventory status change refers to a change in the status category of the corresponding inventory result for the same material in two adjacent unified sampling periods or two adjacent scheduling processing periods; wherein, the status category includes at least continuous supply status, near-shortage status, and supply disruption risk status. When the corresponding inventory status at one moment is continuous supply status, and the corresponding inventory status at the next moment is near-shortage status or supply disruption risk status, the inventory status change is determined to be a change towards shortage; when the corresponding inventory status at one moment is near-shortage status, and the corresponding inventory status at the next moment is supply disruption risk status, the inventory status change is determined to be a risk-increasing change; when the corresponding inventory status at one moment is supply disruption risk status or near-shortage status, and the corresponding inventory status at the next moment returns to continuous supply status, the inventory status change is determined to be a recovery change.

[0036] Step 205: Determine the scope of impact based on the abnormal signals corresponding to the actual disturbances and the digital twin mapping results of the production line. Specifically, for each equipment abnormal signal, read the corresponding workstation in the digital twin mapping results based on the equipment identifier, and then read the production tasks undertaken by the corresponding workstation that still need to be executed after the impact takes effect. For each material abnormal signal, read the corresponding material supply workstation in the digital twin mapping results based on the material identifier, and then read the production tasks that depend on the corresponding material and still need to be executed after the impact takes effect. Merge the objects determined by the equipment abnormal signals and the objects determined by the material abnormal signals in a union manner to generate the scope of impact. The scope of impact includes at least the affected equipment, affected workstation, affected material, affected production task, and the impact taking effect time.

[0037] Step 206, output the scope of impact; specifically, output the scope of impact as the input object for subsequent steps; among which, the affected production tasks are used for subsequent steps to identify resource gaps, and the affected equipment, affected workstations and affected materials are used to limit the scope of resource availability assessment and the screening scope of historical resource allocation records.

[0038] In some implementations, the steps of determining the affected production tasks based on the scope of impact and the digital twin mapping results of the production line, and combining the current resource availability assessment results with historical resource allocation records to determine the resource shortage demand results, include: Step 301: Determine the affected production tasks based on the scope of influence. Specifically, read the output scope of influence and the current production task plan. The current production task plan includes at least the task identifier, process position, planned start time, planned end time, and current execution status. Match the affected workstations in the scope of influence with the process positions in the current production task plan. When a process position falls into an affected workstation and the planned end time is later than the time when the influence takes effect, the corresponding task is determined as an affected production task. For tasks that are executed continuously across multiple processes, if the current process has fallen into an affected workstation and subsequent processes have not yet been completed, the subsequent processes are also retained as part of the affected production task.

[0039] Step 302: Generate current resource availability assessment results based on the affected production tasks. Specifically, for the affected production tasks, read the current status of alternative equipment, allocable materials, and callable transfer resources based on the production line digital twin mapping results. For equipment resources, read the equipment identifier, current status, start time of deployment, processing capacity per unit time, and process adaptation range, and compare whether it can be deployed before the start of the corresponding affected production task and whether it can cover the process position and minimum processing capacity of the task. The minimum processing capacity is determined based on the processing volume per unit time of the corresponding process under the standard cycle time. For material resources, read the material identifier, current location, allocable quantity, outbound start time, and expected arrival time, and compare whether it can arrive before the start of the corresponding affected production task and whether the quantity can cover the amount used by the task within the preset recovery period. The preset recovery period is determined based on the current scheduling processing cycle and at least one complete task recovery cycle.

[0040] For transfer resources, read the resource identifier, current location, available start time, and carrying capacity, and compare whether they can complete the transfer process of the corresponding equipment or materials; summarize the above comparison results of equipment resources, material resources, and transfer resources according to resource identifier to generate the current resource availability assessment result; the current resource availability assessment result includes at least the resource identifier, resource category, available start time, available capacity, and service range.

[0041] Step 303: Filter historical resource allocation records based on the scope of impact. Specifically, read historical scheduling records and search based on the affected equipment category, affected workstation category, affected material category, and affected production task category within the scope of impact. When a historical scheduling record and the current scope of impact meet at least the same quantity condition in the aforementioned categories, the historical scheduling record is retained as a historical resource allocation record. The consistency quantity condition is set based on the comparability of historical scenarios, and the setting method is to ensure that at least three core constraints—the source of disturbance, the location of impact, and the task type—are covered. For each historical resource allocation record, further read the resource transfer source, resource transfer duration, material transfer duration, task switching duration, and post-recovery processing cycle time to provide historical basis for determining subsequent shortage needs.

[0042] Step 304: Identify resource gaps based on the current resource availability assessment results. Specifically, for each affected production task, read its corresponding equipment category, corresponding material category, minimum processing capacity, expected requisition quantity, and start time. Then, read the available start time, available capacity, and serviceable range of the same type of resources from the current resource availability assessment results. Compare the start time with the available start time, the minimum processing capacity with the available capacity, the process location with the serviceable range, and the expected requisition quantity with the available allocation quantity. When any comparison result does not meet the task execution requirements, it is determined that there is a resource gap for the affected production task, and the resource category, gap quantity, required arrival time, and minimum processing capacity of the corresponding gap are recorded.

[0043] Step 305: Determine the resource shortage demand based on the resource gap and historical resource allocation records. Specifically, for each identified resource gap, prioritize reading the transfer method in the historical resource allocation records that matches the resource category, process location, and task type. When the recovery processing cycle time corresponding to the historical transfer method is within the allowable deviation range of the target processing cycle time, directly use the resource category, resource quantity, transfer source, arrival time, and minimum processing capacity corresponding to that historical transfer method as the current resource shortage demand. The target processing cycle time refers to the standard process cycle time that the corresponding affected production task should achieve under normal production conditions. If the standard process cycle time... If a timeout is found, the average processing time of the most recent stable production period will be used. The allowable deviation range of the target processing time will be set based on the historical fluctuation limit of the standard process timeout or the average processing time of the most recent stable production period. If there is no directly applicable transfer method in the historical resource allocation records, the current resource shortage demand will be directly generated based on the resource category, shortage quantity, required arrival time, and minimum processing capacity recorded in the current resource gap. The above processing will be performed on all resource gaps to generate the resource shortage demand result. The resource shortage demand result will at least include the shortage resource category, shortage resource quantity, required arrival time, minimum processing capacity, and corresponding affected production tasks.

[0044] Furthermore, to illustrate the process of determining the resource shortage demand results in step 305, for example, the affected production task T08 corresponds to assembly station A2, starts at 14:20, has a minimum processing capacity of 60 units per hour, and is expected to require 200 units; step 304 identifies that the task has both equipment resource shortages and material resource shortages. The equipment resource shortage is manifested in the fact that the earliest available start time for the same type of equipment is 14:38, which is 18 minutes later than the task start time, and the available capacity is only 52 units per hour; the material resource shortage is manifested in the fact that the current available allocation quantity is only 120 units, which is less than the expected requirement of 80 units. Then, read the historical resource allocation records. If a historical transfer method consistent with the current impact range is found, in which one piece of equipment of the same type was transferred from standby workstation B1 and 100 pieces of materials were transferred from buffer warehouse C3, the recovery cycle time corresponding to this historical transfer method is 31 seconds, while the target processing cycle time corresponding to the current affected production task is 30 seconds. The two are within the allowable deviation range, so the historical transfer method is directly used to correct the current resource gap and generate resource shortage demand results. The corresponding results include: the shortage resource category is assembly equipment and assembly materials, the shortage resource quantity is 1 piece of equipment and 100 pieces of materials, the required arrival time is 14:20, the minimum processing capacity is 60 pieces per hour, and the corresponding affected production task is T08.

[0045] Step 306: Output the current resource availability assessment result and the resource shortage demand result; specifically, output the current resource availability assessment result and the resource shortage demand result together; the resource shortage demand result is used as the input object for the next step, and the current resource availability assessment result is used as the input object for generating the set of reconfigurable paths.

[0046] In some implementations, the detailed steps for generating a set of reconfigurable paths, based on the resource shortage demand results and the production line digital twin mapping results, and in conjunction with equipment allocation priority rules, material flow optimization rules, path constraints, and path conflict detection mechanisms, include: Step 401: Read the resource shortage demand results and the production line digital twin mapping results; specifically, read the shortage resource category, shortage resource quantity, required arrival time, minimum processing capacity, and corresponding affected production tasks from the output resource shortage demand results; read the mapping relationship between equipment and workstation, the mapping relationship between materials and feeding workstation, the mapping relationship between transfer nodes and transfer paths, and the mapping relationship between control interfaces and equipment objects from the output production line digital twin mapping results.

[0047] Step 402: Select equipment resources according to the equipment allocation priority rules. Specifically, select equipment resources corresponding to the shortage resource category from the currently available equipment, equipment to be released, and standby equipment. The equipment allocation priority rules are executed in the following order: first, compare whether the start time of availability is earlier than the required arrival time; second, compare whether the unit time processing capacity is not lower than the minimum processing capacity; third, compare whether the process adaptation range covers the process position of the corresponding affected production task; and fourth, compare the equipment preparation time. If multiple candidate equipment still exist after the above comparisons, compare the difference between the recovery processing cycle time and the target processing cycle time in the historical resource allocation record, and prioritize the equipment with the smaller difference. The equipment preparation time is determined based on the sum of the tooling preparation time, parameter switching time, and no-load confirmation time required for the equipment to switch from the current state to the target process execution state. The equipment resource selection is completed according to the aforementioned order.

[0048] Step 403: Select material flow direction according to material flow optimization rules; specifically, screen material resources corresponding to the shortage resource category from the current available material locations and standby material locations; the material flow optimization rules are executed in the following order: first, calculate the candidate transfer path from the current material location to the target process location based on the production line digital twin mapping results; then compare the expected transfer time, number of nodes passed through, expected arrival time, and transfer resource occupation of each candidate transfer path; prioritize the material flow direction with shorter expected transfer time, fewer nodes passed through, earlier expected arrival, and no overlap in transfer resource occupation with other reserved transfer tasks; the expected transfer time is calculated based on the path length, transfer resource running speed, and node waiting time.

[0049] Step 404: Generate a reconfiguration path based on equipment resources and material flow direction. Specifically, combine the equipment resources selected in step 402 with the material flow direction selected in the previous step around the corresponding affected production task to form multiple reconfiguration paths. Each reconfiguration path includes at least the incoming equipment identifier, material starting position, material arrival position, corresponding affected production task, and planned start time. The planned start time is determined based on the latest time among the expected equipment preparation completion time, the expected material arrival time, and the earliest recoverable time of the corresponding affected production task.

[0050] Step 405: Select reorganization paths based on path constraints. Specifically, path constraints include at least the following: process sequence constraints, the correspondence between equipment arrival time and planned start time, the correspondence between material arrival time and planned start time, the constraint that a single piece of equipment can only undertake one production task at a time, and the constraint that a single material resource can only correspond to one allocation direction at a time. Perform constraint verification on each reorganization path. If a constraint is not met, remove the reorganization path. If all constraints are met, retain the reorganization path.

[0051] Step 406: Select reorganization paths based on the path conflict detection mechanism. Specifically, for any two retained reorganization paths, compare the incoming equipment identifier, material starting position, transfer resource identifier, and planned start time. When two reorganization paths call the same equipment resource, the same material resource, or the same transfer resource during overlapping time periods, it is determined that there is a path conflict between them. For two reorganization paths with path conflicts, prioritize retaining the reorganization path that ranks higher in the equipment allocation priority rule, ranks higher in the material flow optimization rule, and has an earlier planned start time.

[0052] Step 407: Output the set of reconfigurable paths; specifically, summarize the reconfigurable paths that have been filtered by path constraints and path conflict detection mechanism to generate the set of reconfigurable paths, and use the set of reconfigurable paths as the input object for the production recovery time prediction result.

[0053] In some implementations, the set of reconfigurable paths, the current resource availability assessment results, and the production line digital twin mapping results are used as processing objects. The feasibility of the reconfigurable path set is verified, and reconfigurable paths whose equipment resources, material resources, and transfer resources have all been verified are retained. For each verified reconfigurable path, equipment preparation time, material transfer time, task switching time, and cycle time recovery time are determined. These are then combined according to their actual occurrence sequence to generate a production recovery time estimate. Based on the production recovery time estimate and the current resource availability assessment results, the implementation steps for determining the priority execution path include: Step 501: Read the set of reconfigurable paths, the current resource availability assessment results, and the production line digital twin mapping results. Specifically, read the set of reconfigurable paths, the current resource availability assessment results, and the production line digital twin mapping results. For each reconfigurable path, read the incoming equipment identifier, material starting position, material arrival position, corresponding affected production task, and planned start time. For the current resource availability assessment results, read the resource identifier, resource category, available start time, available capacity, and serviceable range. For the production line digital twin mapping results, read the mapping relationship between transfer nodes and transfer paths, as well as the mapping relationship between control interfaces and equipment objects.

[0054] Step 502: Perform feasibility verification on the set of reconfigurable paths. Specifically, for each reconfigurable path, first verify equipment resources; compare the available start time corresponding to the equipment identifier with the planned start time, compare the available capacity with the minimum processing capacity of the corresponding affected production task, and compare the service range with the process position of the corresponding affected production task; when all three comparisons meet the execution requirements, the equipment resources are verified. Next, verify material resources; compare the estimated transfer time from the material's starting position to its arrival position with the remaining time before the planned start time, and compare the available material quantity with the estimated requisition quantity of the corresponding affected production task; when both comparisons meet the execution requirements, the material resources are verified. Next, verify transfer resources; compare the available start time of transfer resources with the planned transfer start time, and compare the transfer resource carrying capacity with the material transfer quantity; when both comparisons meet the execution requirements, the transfer resources are verified. Only reconfigurable paths where equipment resources, material resources, and transfer resources have all been verified are retained for subsequent estimation processing.

[0055] Step 503: Generate production recovery time estimates based on the verified reorganization paths. Specifically, for each verified reorganization path, determine the equipment preparation time, material transfer time, task switching time, and cycle time recovery time. The equipment preparation time is determined based on the sum of tooling preparation time, parameter switching time, and no-load confirmation time required for the equipment to switch from its current state to the target process execution state. The material transfer time is determined based on the path length from the material's starting position to its destination, the number of nodes traversed, node waiting time, and the speed of the transfer resources. The task switching time is determined based on the scheduling switching time and process connection confirmation time required for the affected production task to switch from its original execution position to its new execution position. The cycle time recovery time is determined based on the average time taken from recovery start to reaching the target processing cycle time in historical resource allocation records for the same type of transfer method. Combine the above four types of times according to their actual occurrence sequence to generate the production recovery time estimates for the corresponding reorganization path.

[0056] Furthermore, in some implementations, for the first For each validated reorganization path, the estimated production recovery time is determined using the following formula:

[0057] in, Indicates the first A validated recombination path; Indicates the first The estimated production recovery time for each validated reorganization path; Indicates the first The device preparation time corresponding to each verified recombination path; Indicates the first The material transfer time corresponding to each validated recombination route; Indicates the first The task switching time corresponding to each verified reorganization path; Indicates the first The clock recovery time corresponding to each validated recombination path; Indicates according to the first The time combination function is calculated by combining the actual sequence of occurrence of equipment preparation, material transfer, task switching and cycle time recovery in the verified reorganization path.

[0058] Equipment preparation time Determine according to the following formula: ; in, Indicates the first The tooling preparation time required for equipment to switch from its current state to the target process execution state in a verified reorganization path; Indicates the first The parameter switching time required for equipment to switch from the current state to the target process execution state in a verified reorganization path; Indicates the first The no-load confirmation time required after the device completes parameter switching in the verified reorganization path.

[0059] Material transfer time Determine according to the following formula:

[0060] in, Indicates the first The length of the transfer path between the material's starting position and its final position in a validated recombination path; Indicates the first The speed of resource transfer in the verified reorganization path; Indicates the first The number of nodes traversed by the material transfer path in a verified reorganization path; Indicates the first One node passed through; Indicates the first The first of the verified recombination pathways The waiting time of each node that passes through it; Indicates the first The sum of the waiting times of all nodes traversed in a verified reorganization path.

[0061] Task switching time Determine according to the following formula:

[0062] in, Indicates the first The scheduling switchover time required for affected production tasks in a verified reorganization path to switch from their original execution location to a new execution location; Indicates the first The time required for process connection confirmation after the affected production task in the verified reorganization path has completed the execution location switch and the process connection confirmation has been performed.

[0063] Beat recovery time Determine according to the following formula:

[0064] in, Indicates the relationship between the historical resource allocation record and the first The number of historical records with the same loading method corresponding to each verified recombination path; Indicates the first A similar historical record; Indicates the first The duration from recovery to reaching the target processing cycle in a similar historical record.

[0065] When the equipment preparation process and the material transfer process can be executed in parallel, and the task switching process is executed after both the equipment preparation process and the material transfer process are completed, and the cycle time recovery process is executed after the task switching process is completed, the time combination function... Determine according to the following formula:

[0066] in, This means that when the equipment preparation process and the material transfer process are executed in parallel, the maximum of the equipment preparation time and the material transfer time is used as the completion time of the parallel phase, and then the task switching time and cycle time recovery time are added together to obtain the first phase. The estimated production recovery time for each validated reorganization path.

[0067] When the equipment preparation process, material transfer process, task switching process, and cycle time recovery process are executed sequentially, the time combination function... Determine according to the following formula:

[0068] Among them, equipment preparation time, material transfer time, task switching time, and cycle time recovery time are added together sequentially to obtain the [number]. The production recovery time estimates for the validated reorganization paths are presented above. These production recovery time estimates characterize the estimated time required for the corresponding reorganization path to reach the target processing cycle from the start of recovery preparation, and serve as the basis for subsequent comparisons of recovery times and path selection among different validated reorganization paths. For example, to illustrate the process of generating the production recovery time estimate in step 503, for the first reconfiguration path that has passed verification, the equipment preparation time is 8 minutes, the material transfer time is 6 minutes, the task switching time is 4 minutes, and the cycle time recovery time is 10 minutes. Equipment preparation and material transfer can be executed in parallel, task switching is performed after both are completed, and cycle time recovery is performed after task switching is completed. Therefore, the estimated production recovery time for this first reconfiguration path is 24 minutes. For the second reconfiguration path that has passed verification, the equipment preparation time is 5 minutes, the material transfer time is 11 minutes, the task switching time is 4 minutes, and the cycle time recovery time is 7 minutes. If equipment preparation and material transfer are also executed in parallel, the estimated production recovery time for this second reconfiguration path is 22 minutes. Therefore, even when the equipment preparation time is short but the material transfer time is long, the estimated production recovery time for the corresponding reconfiguration path must be obtained by combining the actual occurrence order of each time period, rather than prioritizing based solely on a single time period.

[0069] Step 504: Compare the reorganization paths based on the production recovery time estimate and the current resource availability assessment. Specifically, for each reorganization path for which a production recovery time estimate has been obtained, the current resource availability assessment is read again to check whether the equipment, material, and transportation resources involved remain available throughout the entire expected execution period. If any resource loses its availability during the corresponding period, the reorganization path is removed. The remaining reorganization paths are sorted in ascending order of the production recovery time estimate. If the production recovery time estimates of two reorganization paths are the same, the reorganization path with less resource consumption during the entire execution period is retained first.

[0070] Step 505: Output the priority execution path; specifically, when only one reorganization path remains after sorting, the reorganization path is determined as the priority execution path; when multiple reorganization paths exist after sorting, they are checked in the sorting order, and the first reorganization path in which equipment resources, material resources, and transfer resources are continuously available throughout the entire execution period is selected as the priority execution path; the priority execution path is used as the input object for generating a list of conflict-free operation instructions.

[0071] In some implementations, the process involves taking the priority execution path and the production line digital twin mapping result as the processing objects, generating a conflict-free operation instruction list based on the priority execution path, issuing adjustment instructions to the production line control system based on the conflict-free operation instruction list, and then determining the production recovery confirmation result based on the execution feedback. The implementation steps include: Step 601: Read the priority execution path and the production line digital twin mapping results; specifically, read the priority execution path and extract the incoming equipment identifier, material start position, material arrival position, corresponding affected production task and planned start time; at the same time, read the mapping relationship between control interface and equipment object, the mapping relationship between equipment and workstation, and the mapping relationship between transfer node and transfer path in the production line digital twin mapping results.

[0072] Step 602: Generate operation instructions based on the priority execution path. Specifically, the priority execution path is decomposed into five types of actions: resource release, equipment transfer, material transfer, task switching, and equipment startup. For resource release, operation instructions are generated to stop the currently occupied task and release the currently occupied equipment. For equipment transfer, operation instructions are generated for equipment preparation, equipment movement, and equipment access. For material transfer, operation instructions are generated for material outbound, material transfer, and material arrival confirmation. For task switching, operation instructions are generated for task pause, task reassignment, and task resumption. For equipment startup, operation instructions are generated to start the target equipment and enter the target process. Each operation instruction clearly defines the execution object, execution time, and prerequisite completion conditions. The execution time is determined based on the planned start time and the sequential dependencies of each action, and the prerequisite completion conditions are determined based on whether the previous action has been completed.

[0073] Step 603: Generate a conflict-free operation instruction list based on the operation instructions. Specifically, sort all the operation instructions generated in step 602 by execution time and compare them one by one with the control instructions currently being executed in the production line control system. When the same equipment, the same workstation, the same material resource, or the same transfer resource is called by two or more instructions simultaneously during overlapping time periods, the operation instruction with the later order is adjusted to the next executable time after the previous conflicting instruction is completed, or to the next executable time after its prerequisite completion conditions are met. After comparison and adjustment, a conflict-free operation instruction list is generated. Each instruction in the conflict-free operation instruction list is mapped to a specific control interface through the production line digital twin mapping result.

[0074] For example, to illustrate the process of generating the list of conflict-free operation instructions in step 603, if, according to the operation instructions generated in step 602, the execution object corresponding to the equipment loading instruction I1 is equipment E09, and the execution time is from 15:00 to 15:06; the execution object corresponding to the task switching instruction I2 also includes workstation A2, and the execution time is from 15:03 to 15:07; the transfer resource corresponding to the material transfer instruction I3 is R04, and the execution time is from 15:01 to 15:05; another material transfer instruction I4 also corresponds to the transfer resource R04, and the execution time is from 15:04 to 15:08; since instruction I1 and instruction I2 are executed from 15:03 to 15:08... Between 5:06 and 5:06, workstation A2 was occupied in an overlapping manner, and instructions I3 and I4 occupied transfer resource R04 in an overlapping manner between 15:04 and 15:05. Therefore, the instructions that are later in the order will be postponed. If the prerequisite for instruction I2 was met at 15:06, instruction I2 will be executed from 15:06 to 15:10. If the prerequisite for instruction I4 was met at 15:05, instruction I4 will be executed from 15:05 to 15:09. After the above adjustments, a new list of conflict-free operation instructions will be formed, and each instruction will be mapped to the equipment control interface, warehouse control interface, transfer control interface, or scheduling control interface.

[0075] Step 604: Issue adjustment instructions to the production line control system according to the list of conflict-free operation instructions. Specifically, according to the execution order in the list of conflict-free operation instructions, and based on the mapping relationship between control interfaces and equipment objects in the production line digital twin mapping results, send adjustment instructions to the corresponding control interfaces one by one. Equipment-related operation instructions are sent to the equipment control interface, material transfer-related operation instructions are sent to the warehouse control interface or transfer control interface, and task switching-related operation instructions are sent to the scheduling control interface. For each adjustment instruction sent, record the corresponding execution object, the sending time, and the expected completion time, and only after the instruction enters the feedback state will the next dependent adjustment instruction be sent.

[0076] Step 605: Collect execution feedback based on adjustment instructions; specifically, read the execution feedback corresponding to each adjustment instruction from the production line control system; the execution feedback includes at least the instruction reception status, instruction execution status, equipment operation status, material arrival status, task execution status, and current processing cycle; match the execution feedback with the list of conflict-free operation instructions one by one to determine whether each adjustment instruction has been completed, and at which action stage it is stuck if it has not been completed.

[0077] Step 606: Determine the production recovery confirmation result based on the execution feedback; specifically, read all execution feedback, first determine whether the adjustment instructions corresponding to resource release, equipment transfer, material transfer, task switching, and equipment startup have been executed; only when all the above key actions are completed, continue to read the current processing cycle and current material arrival status after equipment startup.

[0078] Next, compare the current processing cycle with the target processing cycle. The target processing cycle is the standard process cycle that the affected production task should achieve under normal production conditions, as determined in the above steps. If the standard process cycle is missing, the average processing cycle of the most recent stable production period is used. The allowable deviation range is set based on the upper limit of historical stable fluctuations.

[0079] If the current processing cycle time is within the allowable deviation range of the target processing cycle time, and the current material availability status can cover the continuous requisition demand within the preset observation period, the production recovery confirmation result will output "Recovery Completed"; otherwise, the production recovery confirmation result will output "Recovery Not Completed".

[0080] The preset observation duration is set based on the time required to complete multiple target processing cycles consecutively. The method of calling it is to continuously check whether the processing cycle and material arrival status meet the conditions within this duration; the production recovery confirmation result is used as the end output of this round of real-time scheduling processing.

[0081] Example 2

[0082] Referring to Figure 2, this embodiment provides an AI-driven digital twin mixed-flow production real-time scheduling system. Since this system uses the AI-driven digital twin mixed-flow production real-time scheduling method from Embodiment 1, it has the same effect, which will not be repeated here. The system includes: The data parsing module is used to collect real-time data streams from the production site and perform layered parsing of equipment operating parameters and material inventory status to generate equipment fault identification results and inventory status results. The disturbance determination module is used to determine disturbances based on equipment fault identification results and inventory status results, filter abnormal signals, and determine the scope of impact. The shortage determination module is used to identify affected production tasks based on the scope of impact, and to determine the resource shortage demand results by combining the current resource availability assessment results and historical resource allocation records; The path generation module is used to generate a set of reconfigurable paths based on the resource shortage demand results, combined with equipment allocation priority rules, material flow optimization rules, path constraints, and path conflict detection mechanisms. The path selection module is used to verify the feasibility of the set of reconfigurable paths, retain the reconfigurable paths that have been verified in terms of equipment resources, material resources, and transfer resources, and determine the equipment preparation time, material transfer time, task switching time, and cycle time recovery time for the verified reconfigurable paths. It combines these into a production recovery time estimate based on the actual occurrence order, and determines the priority execution path based on the production recovery time estimate and the current resource availability assessment. The recovery execution module is used to generate a list of conflict-free operation instructions based on the priority execution path; issue adjustment instructions to the production line control system based on the list of conflict-free operation instructions; and determine the production recovery confirmation result based on the execution feedback.

[0083] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims.

Claims

1. An AI-driven real-time scheduling method for mixed-flow production using digital twins, characterized in that: include: Collect real-time data streams from the production site, and perform time-based correction and validity verification on equipment operating parameters and material inventory status to generate equipment fault identification results and inventory status results; Based on the fault persistence status in the equipment fault identification results, determine the equipment abnormal signal; based on the inventory status in the inventory status results, determine the material abnormal signal; perform continuity filtering and consistency filtering on the equipment abnormal signal and the material abnormal signal; and determine the scope of influence based on the equipment abnormal signal and the material abnormal signal retained after filtering. The affected production tasks are determined based on the scope of impact, and the resource shortage demand is determined by combining the current resource availability assessment results and historical resource allocation records; Based on the resource shortage demand results, combined with equipment allocation priority rules, material flow optimization rules, path constraints and path conflict detection mechanisms, a set of reconfigurable paths is generated; Feasibility verification is performed on the set of reconfigurable paths. Reconfigurable paths that have been verified in terms of equipment resources, material resources, and transfer resources are retained. For each verified reconfigurable path, the equipment preparation time, material transfer time, task switching time, and cycle time recovery time are determined. The production recovery time is estimated by combining the paths according to their actual occurrence order. Based on the production recovery time estimate and the current resource availability assessment, the priority execution path is determined. A list of conflict-free operation instructions is generated based on the priority execution path. Adjustment instructions are then issued to the production line control system based on the list of conflict-free operation instructions, and the production recovery confirmation result is determined based on the execution feedback.

2. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 1, characterized in that, Determining the scope of impact includes: The equipment fault identification result with a continuous fault status is determined as an equipment abnormality signal, and the inventory status result with a near shortage status or a supply disruption risk status is determined as a material abnormality signal. Continuous filtering is performed on equipment and material abnormality signals to remove short-term abnormal signals that appear within a single unified sampling period and recover in adjacent unified sampling periods. Then, consistency filtering is performed on the retained abnormal signals based on the production line digital twin mapping results. The corresponding workstation, corresponding material supply workstation, and corresponding production task are read and merged in a union manner to determine the scope of impact.

3. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 1, characterized in that, Methods for determining resource shortage demand outcomes include: Based on the equipment category, material category, minimum processing capacity, expected requisition quantity and start time of the affected production task, compare them with the available start time, available capacity, available service range and available allocation quantity in the current resource availability assessment results to identify resource gaps; Then, based on historical resource allocation records consistent with the current impact range, the resource categories, shortage quantities, start times, and minimum processing capacities corresponding to the resource gaps are adjusted to determine the shortage resource categories, shortage resource quantities, required arrival times, and minimum processing capacities, generating resource shortage demand results.

4. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 3, characterized in that, Generating a set of recombinable paths includes: Based on the equipment allocation priority rules, select equipment resources from available equipment, equipment to be released, and standby equipment that can be put into use with a start time no later than the required arrival time and a processing capacity per unit time no less than the minimum processing capacity. Based on the material flow optimization rules, calculate the estimated transfer time and estimated arrival time from each available material location and spare material location to the corresponding affected production task execution location, and select the material flow direction with the shortest estimated transfer time from the material flow directions whose estimated arrival time is not later than the required arrival time. The selected equipment resources and material flows are combined around the corresponding affected production tasks to generate a set of reconfigurable paths.

5. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 4, characterized in that, For each reorganization path, perform path constraint verification. Path constraints include process sequence constraints, the correspondence between equipment arrival time and plan start time, the correspondence between material arrival time and plan start time, the constraint that a single piece of equipment can only undertake one production task at the same time, and the constraint that a single material resource can only correspond to one allocation direction at the same time. If a reorganization path does not meet any of the path constraints, the corresponding reorganization path is removed; if all path constraints are met, the corresponding reorganization path is retained.

6. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 5, characterized in that, For any two retained reorganization paths, compare the incoming equipment identifier, material starting position, transfer resource identifier, and planned start time; When two reorganization paths call the same equipment resources, the same material resources, or the same transfer resources during overlapping periods, it is determined that there is a path conflict between the two reorganization paths. For two reorganization paths that have path conflicts, retain the reorganization path that is prioritized by the equipment allocation priority rule, the material flow optimization rule, and the planned start time, and summarize the finally retained reorganization paths to generate a set of reorganizable paths.

7. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 1, characterized in that, Determining the preferred execution path includes: For each reconfigurable path in the set of reconfigurable paths, verify the equipment resources, material resources, and transfer resources respectively; When equipment resources meet the process adaptation range and unit time processing capacity, material resources meet the available quantity and expected arrival time, and transfer resources meet the carrying capacity and start time of input, the corresponding reorganization path will be determined as the verified reorganization path. For the verified reorganization paths, determine the equipment preparation time, material transfer time, task switching time, and cycle time recovery time respectively, and combine them according to the actual sequence of occurrence to generate the production recovery time estimate; Based on the estimated production recovery time and the current resource availability assessment, the priority execution path is determined from the validated reorganization paths.

8. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 7, characterized in that, The generated production recovery time estimate results include: The equipment preparation time is determined based on the tooling preparation time, parameter switching time, and no-load confirmation time required for the equipment in the priority execution path to switch from the current state to the target process execution state. The material transfer time is determined based on the path length from the material's starting position to its destination, the number of nodes it passes through, the node waiting time, and the speed of the transfer resources. The task switching time is determined based on the scheduling switching time and process connection confirmation time required for the affected production task to switch from its original execution position to its new execution position. The cycle recovery time is determined based on the average time taken from recovery to reaching the target processing cycle in similar historical resource allocation records; The production recovery time estimate is generated by combining the parallel or sequential relationships between equipment preparation time, material transfer time, task switching time, and cycle time recovery time.

9. The AI-driven digital twin mixed-flow production real-time scheduling method according to claim 1, characterized in that, Methods for determining the confirmation of production resumption include: Based on the priority execution path, generate operation instructions corresponding to resource release, equipment transfer, material transfer, task switching, and equipment startup. Adjust the execution time of operation instructions that call the same equipment resource, the same material resource, or the same transfer resource and whose execution time overlaps, and generate a list of conflict-free operation instructions. Based on the list of conflict-free operation instructions, adjustment instructions are issued to the production line control system. The production recovery confirmation result is determined based on the completion status of each operation instruction in the execution feedback, the comparison results between the current processing cycle and the target processing cycle, and the current material arrival status.

10. An AI-driven real-time scheduling system for mixed-flow production of digital twins, used to implement the AI-driven real-time scheduling method for mixed-flow production of digital twins as described in any one of claims 1-9, characterized in that, The system includes: The data parsing module is used to collect real-time data streams from the production site, and to perform time-based correction and validity verification of equipment operating parameters and material inventory status, generating equipment fault identification results and inventory status results. The disturbance determination module is used to determine the equipment abnormal signal based on the fault persistence status in the equipment fault identification result, determine the material abnormal signal based on the inventory status in the inventory status result, perform continuity filtering and consistency filtering on the equipment abnormal signal and the material abnormal signal, and determine the scope of influence based on the equipment abnormal signal and the material abnormal signal retained after filtering. The shortage determination module is used to identify affected production tasks based on the scope of impact, and to determine the resource shortage demand results by combining the current resource availability assessment results and historical resource allocation records; The path generation module is used to generate a set of reconfigurable paths based on the resource shortage demand results, combined with equipment allocation priority rules, material flow optimization rules, path constraints, and path conflict detection mechanisms. The path selection module is used to verify the feasibility of the set of reconfigurable paths, retain the reconfigurable paths that have been verified in terms of equipment resources, material resources, and transfer resources, and determine the equipment preparation time, material transfer time, task switching time, and cycle time recovery time for the verified reconfigurable paths. It combines these into a production recovery time estimate based on the actual occurrence order, and determines the priority execution path based on the production recovery time estimate and the current resource availability assessment. The recovery execution module is used to generate a list of conflict-free operation instructions based on the priority execution path; issue adjustment instructions to the production line control system based on the list of conflict-free operation instructions; and determine the production recovery confirmation result based on the execution feedback.

Images