A resin well lid injection molding production line process collaborative control system

Abstract

The present application relates to the technical field of industrial process control and manufacturing execution system, in particular to a resin well cover injection molding production line process collaborative control system. The system establishes a process state vector for each production task which is updated with process events, constructs a process constraint propagation graph based on process maturity constraints, equipment capacity constraints and upstream and downstream bearing constraints, calculates the process state debt of the actual state transition time relative to the target transition time, and determines the deviation transmission range along the constraint propagation graph. A rolling optimization window is formed with the affected tasks and associated processes, and task release, device reservation, sequence conversion and inspection insertion instructions are jointly generated. The state vector and process state debt are updated according to the execution feedback with version identification. Thus, the product sequence conversion meets the actual maturity condition and the downstream acceptance condition at the same time, inhibits the step-by-step accumulation of process deviation, reduces the large-scale rescheduling caused by early demolding, work-in-process congestion and local abnormalities.

Description

Technical Field

[0001] This invention relates to the field of industrial process control and manufacturing execution systems, specifically to a collaborative management and control system for resin manhole cover injection molding production line processes. Background Technology

[0002] Resin manhole cover injection molding production lines typically include continuous processes such as raw material preparation, injection molding, pressure holding, cooling or curing, demolding, finishing, and quality inspection. Existing manufacturing execution systems (MES) generally establish production task queues based on preset process routes, planned cycle times, and equipment status, and issue material feeding, injection, transfer, and inspection instructions according to equipment idle signals, planned completion times, or manual confirmation. The injection molding equipment controller is responsible for executing set parameters such as pressure, speed, temperature, holding time, and cooling time. The MES collects events such as equipment start, completion, shutdown, and alarms, and records task progress, work-in-process quantity, and equipment utilization status. When a process is delayed, conventional systems typically handle it by postponing subsequent tasks, adjusting equipment queue order, or regenerating a local plan; when downstream stations are temporarily congested, they limit the inflow of work-in-process by pausing transfers or adding waiting queues. Such schemes use whether the equipment has completed its actions and whether the task has reached the planned time as the main criteria for judgment. They can achieve basic production tracking and process connection, but they do not uniformly incorporate the actual maturity status of the product during the pressure holding, cooling or curing process, the downstream acceptability, and the transmission relationship of the same deviation in subsequent processes into the control criteria.

[0003] Batch variations in resin materials, fluctuations in the injection molding process, and changes in equipment load can cause the actual injection cycle, holding pressure completion time, and cooling or curing progress to deviate from the preset cycle time. Even if the product has received the equipment completion signal, it may not yet have reached the process conditions for demolding or transfer to another process. When existing manufacturing execution systems determine the transfer process based on fixed scheduled times or the completion status of a single piece of equipment, they cannot distinguish between local delays that can be absorbed and state deviations that will continue to accumulate along subsequent processes, nor can they determine whether the remaining processing capacity of downstream processes is sufficient to accept the current product. As a result, the system may pre-schedule demolding, trimming, or inspection tasks, allowing immature products to enter the next process; it may also continue to release new injection molding tasks when downstream is congested, causing work-in-process inventory and equipment waiting time to increase step by step. Local equipment parameter control can only maintain the operation of the current process, and routine rescheduling does not consider the status of processes that have already been executed and cannot be undone, which can easily expand the scope of adjustment. The main problem with existing technologies is the lack of a collaborative control mechanism that can characterize the actual process status of the product, quantify state deviations and propagate them along process dependencies, and then jointly adjust task release, equipment reservation, and transfer time based on this. Summary of the Invention

[0004] The purpose of this invention is to provide a collaborative control system for resin manhole cover injection molding production line processes, which can solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A collaborative control system for resin manhole cover injection molding production line processes includes a processor, a memory, and a data interface for communicating with the injection molding equipment controller and process terminals. The memory stores instructions that can be executed by the processor, which instruct the processor to: establish a process state vector for each resin manhole cover production task that is updated with process events;

[0007] A process constraint propagation diagram is constructed based on process maturity constraints, equipment capacity constraints, and upstream and downstream capacity constraints.

[0008] Calculate the process state debt of each task's actual state transition time relative to the target transition time and propagate it along the process constraint propagation graph;

[0009] A rolling optimization window is formed using the affected tasks and related processes to generate task release, sequence transfer, and equipment execution instructions, and the process state vector and process state debt are updated based on the execution feedback.

[0010] Preferably, the process status vector is associated with both the product identifier and the production task, and includes the raw material batch identifier, process version, current process identifier, completed process stage, actual injection molding end time, pressure holding completion status, remaining cooling or curing time, allowable demolding time interval, expected entry time of subsequent processes, and quality lock status.

[0011] The processor updates the irreversible state field and the variable state field in the process state vector according to the timestamped process event returned by the device controller, and retains the state version mapping relationship before and after the update.

[0012] Preferably, the process constraint propagation graph takes the process state node corresponding to the production task as the vertex, the state dependency relationship between the preceding and following processes as the directed edge, and encodes the process maturity condition, equipment availability, process continuity in the same batch, and downstream work-in-process capacity as edge constraints respectively.

[0013] The processor determines the activation state of each edge based on the constraint satisfaction interval corresponding to each directed edge, and only allows the process state debt to propagate through directed edges that are in the active state and point to processes that have not yet been completed, and truncates the propagation path corresponding to completed irreversible processes.

[0014] Preferably, for each production task, the processor jointly compares the actual state transition time, the earliest transition time that meets the process maturity conditions, and the acceptable time that meets the downstream capacity conditions to form a process state debt with direction marking.

[0015] For multiple production tasks within the same rolling optimization window, a debt sequence is established according to the arrival order and accumulation level of the process status debt in the process constraint propagation diagram, and the debt sequence is used to limit the release order, equipment reservation time period, transfer time and inspection task insertion position of the tasks to be adjusted.

[0016] Preferably, the processor maps injection end event, pressure holding completion event, cooling or curing stage event, demolding confirmation event and quality release event to a unified process time scale axis, and reconstructs missing or out-of-order events based on the duration between adjacent valid events;

[0017] During the reconstruction process, the set of historical tasks that can participate in the reconstruction is limited by the process version and raw material batch identifier in the process state vector. The stage duration distribution of the historical task set is used to determine the candidate time intervals of missing events, and the candidate time intervals that have a time sequence conflict with the subsequently confirmed events are eliminated.

[0018] Preferably, for each directed edge in an active state, the processor generates a propagation allocation coefficient that changes with the process event based on the remaining processable capacity of the processes at both ends of the edge, the current work-in-process inventory, the allowed time interval for the target task to be transferred, and the reserved time period of the associated equipment.

[0019] When there are multiple propagable paths for the same process state debt, the process state debt is allocated according to the continuous product of the propagation allocation coefficients on each propagation path, and the allocated debt amount is written into the corresponding subsequent process state node until a frozen node, a quality locked node, or an inactive directed edge is reached.

[0020] Preferably, when the processor establishes the rolling optimization window, it sets the process tasks that have already started execution and are irreversible as frozen tasks, and uses their equipment occupancy range and the earliest subsequent release time as fixed constraints.

[0021] For non-frozen tasks, a candidate adjustment set is constructed based on the aforementioned debt sequence. Only the release time of the task to be fed, the order of injection molding tasks, the start time of the next production cycle, the time of product transfer, and the insertion position of the inspection task are adjusted. Candidate adjustment schemes that violate the allowable demolding time interval, the continuity of the same batch of processes, or the capacity constraints of downstream work-in-process are eliminated.

[0022] Preferably, the processor generates a task control sequence with version identifier for the scrolling optimization window, and sends the task release instruction, equipment reservation instruction and sequence transfer instruction to the corresponding injection molding equipment controller and process terminal respectively;

[0023] The processor receives execution start events, execution completion events, and execution rejection events for each instruction. It verifies the task control sequence according to the task identifier, process identifier, and version identifier carried by the event. It isolates and stores feedback with inconsistent versions and updates the process state vector, the process constraint propagation graph, and the debt sequence only based on execution feedback with consistent versions.

[0024] Preferably, when multiple candidate adjustment schemes meet the constraints, the processor calculates the residual amount of process state debt, the number of debt propagation layers across processes, and the subsequent release offset of frozen tasks after the execution of each candidate adjustment scheme, and selects the target adjustment scheme in a hierarchical order of priority: the residual amount of process state debt, the second priority of the number of debt propagation layers across processes, and the subsequent release offset of frozen tasks.

[0025] When different candidate adjustment schemes have the same value at the same level, the candidate adjustment scheme with the smallest change in the original task order is used as the basis for generating task release, equipment reservation and sequence transfer instructions in the rolling optimization window.

[0026] Preferably, after completing a version-consistent execution feedback update, the processor extracts the debt difference and directed edge activation state changes of each process state node before and after the update, and determines the actual debt resolution path corresponding to the current task control sequence.

[0027] When the actual debt resolution path is inconsistent with the predicted resolution path used when generating the task control sequence, only the production tasks and related processes involved in the difference path are re-formed into local rolling optimization windows, and the original task control sequence version, equipment reservation time period and transition time of the production tasks outside the difference path are retained.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] 1. Establish a process state vector for each resin manhole cover production task, updating with process events. This vector incorporates raw material batches, current process, injection end time, holding pressure status, remaining cooling or curing time, allowable demolding time interval, expected entry time of subsequent processes, and quality lock-in status into the same state record. A process constraint propagation graph is constructed based on process maturity constraints, equipment capacity constraints, and downstream load constraints. The system generates directional process state debt by comparing actual state transition times with target transition times and propagates it along process paths that still have actual dependencies, enabling local cycle offsets to be identified before they become downstream congestion. The rolling optimization window only adjusts material release, injection sequence, equipment reservation, and transition times around the affected tasks, ensuring that product transitions conform to actual maturity conditions and downstream acceptance conditions, avoiding premature transitions and the gradual accumulation of deviations caused by fixed cycle time control.

[0030] 2. The process state vector saves state versions according to the product identifier. Timestamped events returned by the equipment are used to distinguish between irreversible and variable states, reducing interference from out-of-order or missing events on task judgment. The process constraint propagation graph performs propagation truncation on completed irreversible processes, quality-locked nodes, and inactive constraint edges, limiting the calculation range of state deviations. During rolling optimization, tasks that have already started and are irreversible are frozen; only non-frozen tasks are adjusted, and candidate solutions are verified for allowable demolding time intervals, process continuity within the same batch, and downstream work-in-process capacity. The task control sequence with version identifiers is also used to verify equipment execution feedback and isolate feedback from older versions. When the actual resolution path differs from the predicted path, only the differing path is recalculated, while the equipment reservation and reordering arrangements for the remaining tasks remain unchanged. Attached Figure Description

[0031] Figure 1 A flowchart illustrating the overall process control of the resin manhole cover injection molding production line.

[0032] Figure 2 Intensity map of process status vector field;

[0033] Figure 3 This is a graph showing how the status of a process changes with events.

[0034] Figure 4 A flowchart for process event correction, status update, and constraint propagation;

[0035] Figure 5 A cost graph for candidate missing events;

[0036] Figure 6 This is a flowchart of rolling optimization and version feedback driven by process status debt. Detailed Implementation

[0037] In one embodiment, a manufacturing execution system (MAS) for collaborative management of resin manhole cover injection molding production line processes includes a processor, a memory, and a data interface for communicating with the injection molding equipment controller and process terminals. The memory stores instructions executable by the processor. These instructions enable the processor to establish a process state vector for each resin manhole cover production task, which is updated with process events. Based on process maturity constraints, equipment capacity constraints, and upstream and downstream carrying capacity constraints, a process constraint propagation graph is constructed. The process state debt of each task's actual state transition time relative to the target transition time is calculated and propagated along the process constraint propagation graph. A rolling optimization window is formed with affected tasks and related processes. Task release, sequence transition, and equipment execution instructions are generated, and the process state vector and process state debt are updated based on execution feedback.

[0038] In this embodiment, the resin manhole cover injection molding production line organizes raw material preparation, injection molding, pressure holding, cooling or curing, demolding, finishing, and quality inspection in sequence according to the production process. The manufacturing execution system does not use a preset cycle time as the sole basis for sequence transition. Instead, it binds the actual state of each product in each process, the process conditions that allow it to enter the next process, and the acceptability status of the next process to the same production task. The production task generates a unique product identifier from the moment it receives the production plan. The product identifier remains consistent between the injection molding equipment controller, the process terminal, and the manufacturing execution system. Any process event carries the product identifier, process identifier, event type, occurrence time, and execution version. This allows the equipment to complete its actions, the product's maturity status, and the downstream acceptance status to be associated and processed under the same time reference.

[0039] Please refer to Figure 1 After receiving the production plan, the Manufacturing Execution System (MES) establishes the initial state of the task based on the process route corresponding to the resin manhole cover. The initial state records at least the raw material batch, process version, process to be executed, planned release time, candidate set of injection molding equipment, calculation rules for the allowable demolding time interval, and quality release requirements. The processor writes the initial state into the task status area in the memory and locates the corresponding task according to the product identifier pointed to by the process event when the process event arrives. If the event belongs to equipment start, injection end, pressure holding completion, cooling or curing progress, demolding confirmation, trimming completion, or quality release, the corresponding status field is updated. If the event cannot be matched with the existing task, process version, or current process, it is written to the isolation area without directly changing the executable status of the production task.

[0040] In this embodiment, the process state vector is not a simple copy of the equipment state, but rather a cross-process state expression centered on the product. The processor converts the action completion status fed back by the equipment controller into the product state, converts the manual confirmation or inspection conclusion fed back by the process terminal into the quality state, and converts the downstream process queue and equipment reservation results into the acceptable state. At any given time, the process state vector can indicate which irreversible actions the product has completed, which process stages it still needs to go through, when demolding or transfer is allowed at the earliest, which subsequent process it is scheduled to enter, and whether it is currently under quality lock. This avoids the equipment completion signal being directly equated with the product having the conditions for transfer.

[0041] The process state vector can be represented as:

[0042] ;

[0043] Where i represents the production task corresponding to the product, and t represents the time of status recording. Indicates the batch number of raw materials. Indicates the process version. Indicates the current process. This indicates a set of completed process stages. Indicates the actual end time of injection molding. This indicates that the pressure holding process is complete. Indicates the remaining cooling or curing time. and These represent the lower and upper bounds of the allowable demolding time interval, respectively. Indicates the expected start time of subsequent processes. Indicates a quality lock status. This indicates the status version. When a product completes injection molding at time 120 and requires 30 more time units to cool, the processor will... Write it as 120 and It is written as 30, and continues to decrease after the cooldown progress event is reached. Only when The state vector is allowed to transition to the state to be demolded only when it returns to zero and the current time falls within the allowed demolding time interval.

[0044] Please refer to Figure 2 , Figure 2 The horizontal direction represents the sampling window that progresses continuously according to the production rhythm. It is expressed in Chinese segments: initial segment, front segment, mid-front segment, middle segment, mid-rear segment, rear segment, and final segment, without using numerical sequence numbers. The vertical direction represents the relative intensity of each process status field after normalization. The higher the value, the more significant the contribution of the field to the process status vector within the corresponding time window. Figure 2The inner surfaces correspond to fields such as equipment temperature, mold pressure, mold opening and closing status, material batch disturbance, conveying position offset, operation task load, and waiting queue length in the injection molding production line. The curves are calculated by the process state vector formula in the text.

[0045] from Figure 2 It can be seen that the equipment temperature and mold pressure remain at a high level from the initial stage to the later stages, reflecting the continuous impact of heating, melting, mold filling, and pressure holding stages on the molding quality during the resin manhole cover injection molding process. A significant step change in the mold opening and closing state occurs near the middle stage, indicating a clear state transition boundary between mold closing completion, pressure holding end, and demolding start. The material batch disturbance curve is generally flat, but slightly rises in the later stages, indicating that changes in different resin batches or filler ratios are gradually transmitted to subsequent processes through differences in viscosity, flowability, and cooling shrinkage.

[0046] The waiting queue length and the workload of the operation tasks increased synchronously in the later and final stages, indicating that local queues are prone to form in the trimming, inspection, transfer or buffering stages when the production line cycle time is close to saturation.

[0047] The Manufacturing Execution System organizes the process status nodes of each production task into a directed process constraint propagation graph based on the process route. Figure 2 The vertices in the abstract node correspond to the state of a specific product in a specific process. Multiple products are not merged into the same abstract node. Figure 2 The directed edges in the process correspond to the constraint dependency formed by the state of the previous process on the state of the next process. The raw material preparation node points to the injection molding node, the injection molding node points to the holding pressure node, the holding pressure node points to the cooling or curing node, the cooling or curing node points to the demolding node, and the demolding node continues to point to the finishing and quality inspection node. At the same time, cross-task constraint edges are established based on the queuing relationship of the same equipment, the continuous execution relationship of the same batch, and the capacity relationship of the downstream station.

[0048] Each constraint edge stores the constraint type, constraint source, constraint satisfaction range, associated equipment, occupancy relationship, and current activation state. Process maturity constraints are used to limit the earliest time when the product transitions from the holding pressure, cooling, or curing state to the demolding state. Equipment capability constraints are used to limit whether candidate equipment can execute the corresponding process version. Upstream and downstream carrying capacity constraints are used to limit whether subsequent processes can accept new work-in-process within a predetermined time. Process continuity constraints within the same batch are used to prevent the arbitrary insertion of incompatible tasks in production intervals that need to maintain continuous execution. The processor sets the constraint edge to the active state only when both ends of the constraint edge have not ended and the constraint condition still has an impact on the current plan.

[0049] Process state debt is used to represent the time deviation of the current state of a production task relative to the target state that meets the process maturity conditions and downstream acceptance conditions. Positive state debt indicates that the actual transition will be later than the target feasible time, negative state debt indicates that the planned transition is earlier than the earliest feasible time, and zero value indicates that the current arrangement is consistent with the target feasible time. After each injection molding end, pressure holding completion, cooling or curing progress, equipment shutdown, downstream queue change or quality lock event, the processor recalculates the state debt of the relevant node and determines whether the deviation needs to be propagated to subsequent processes based on the activated constraint edges.

[0050] The processor determines the actual state transition time, the earliest process maturity time, and the downstream acceptable time for each affected task. The target transition time is determined by the later of the earliest process maturity time and the downstream acceptable time. When the transition time given by the production plan is earlier than the target transition time, a negative state debt is formed. When the actual executable time is later than the target transition time, a positive state debt is formed. The state debt is not directly used to change equipment parameters, but is used to determine which tasks need to enter the rolling optimization window and whether the adjustment action should be applied to task release, equipment reservation, or transition arrangement.

[0051] Process status debt can be calculated using the following formula:

[0052] ;

[0053] in, This represents the process state debt of production task i in process p. This indicates the actual moment when a state transition can occur based on the current execution state. This indicates the earliest time when the conditions for technological maturity are met. This indicates the earliest acceptable time when the downstream carrying capacity conditions are met. This indicates that the later value is taken. For example, a product can reach the demolding maturity condition at time 180, while the downstream trimming station can only be accepted at time 190. The current arrangement requires the transition at time 185. The actual feasible time is determined as 190, and a state debt with a value of -5 is formed. Based on this, the processor prohibits the issuance of the transition instruction at 185 and continues to propagate the difference along the activation constraint edge after the trimming process.

[0054] Please refer to Figure 3 , Figure 3The horizontal direction represents the main event stages experienced by the injection molding production line within a production batch, including raw material input, mold closing and start-up, injection filling, pressure holding and maintenance, cooling and shaping, demolding and transfer, trimming and inspection, and buffer waiting. The vertical direction represents the intensity of process state debt, that is, the comprehensive deviation of the process state vector after the event is executed from the expected state, constrained state, or historical stable state. The curves correspond to the state debt changes of different processes or combinations of processes. Based on the state debt calculation formula in the text, the state change amount, event impact weight, and process constraint intensity are integrated.

[0055] from Figure 3 It can be seen that the state debt rises rapidly in the early stages of mold closing and injection molding, indicating that mold closing accuracy, injection pressure build-up speed, and resin melt filling stability are most sensitive to system deviations. The curves in the holding and cooling stages show a slow release trend, indicating that the holding time, cooling water temperature, and product shrinkage process gradually absorb the previous deviations. However, if the cooling cycle is insufficient or the pressure holding time is too short, the state debt will still remain in the demolding and trimming stages. The curves in the inspection and buffering stages show a second rise in the later stages, reflecting the reverse impact of non-conforming part re-inspection, rework waiting, or logistics blockage on the production line's collaborative status.

[0056] When the process state debt propagates along the process constraint propagation graph to subsequent nodes, the processor does not unconditionally copy all the debt to every subsequent node. Instead, it calculates the propagation allocation relationship based on the remaining processable capacity of the subsequent process, the current work-in-process inventory, the allowed time interval for task reordering, and the reserved equipment time slots. If the subsequent process has an idle buffer and can digest the preceding delay before the work-in-process arrives, the corresponding propagation amount is absorbed. If the downstream capacity is already full or there is an uninterrupted continuous relationship between subsequent tasks, the state debt continues to be propagated to the next lower-level process until it reaches the frozen node, the quality locked node, or the inactive constraint edge.

[0057] In a preferred embodiment, please refer to Figure 4 The process state vector is associated with both the product identifier and the production task, and includes the raw material batch identifier, process version, current process identifier, completed process stage, actual injection molding end time, pressure holding completion status, remaining cooling or curing time, allowable demolding time interval, expected entry time of subsequent processes, and quality lock status. The processor updates the irreversible state field and variable state field in the process state vector according to the timestamped process event returned by the equipment controller, and retains the state version mapping relationship before and after the update.

[0058] Irreversible state fields include records of completed injection molding, demolding, trimming, and quality judgment that cannot be restored to the previous state through planning adjustments. Variable state fields include remaining cooling or curing time, expected start time of subsequent processes, equipment reservation time slots, and order of tasks to be executed. After receiving an event, the processor first determines whether the event changes the irreversible state. If it changes the irreversible state, a new state version is generated and the corresponding historical fields are locked. Subsequent rescheduling can only be adjusted after the irreversible state. If the event only changes the variable state, the relevant fields are updated under the current state version and the update log is retained so that when subsequent execution feedback arrives, it can be determined whether the feedback corresponds to the current plan or an expired plan.

[0059] The state version mapping relationship is stored using a parent version and child version association method. The parent version records the process state vector, constraint edge state, and task control sequence before adjustment. The child version records the event that triggered the adjustment, the affected fields, and the recalculation result. Any equipment execution feedback is located to the corresponding child version through the product identifier, process identifier, and version identifier. If the feedback still points to the parent version and the parent version has been closed, the feedback enters the isolation area. The processor does not use the feedback to update the current task state. The state vector is corrected through irreversible event supplementation after the actual action of the equipment is verified.

[0060] In a preferred embodiment, the processor maps injection end event, pressure holding completion event, cooling or curing stage event, demolding confirmation event, and quality release event to a unified process time scale axis, and reconstructs missing or out-of-order events based on the duration between adjacent valid events. During the reconstruction process, the set of historical tasks that can participate in the reconstruction is limited by the process version and raw material batch identifier in the process state vector. The candidate time intervals of missing events are determined by the stage duration distribution of the historical task set, and candidate time intervals that conflict with subsequent confirmed events are eliminated.

[0061] The unified process timescale axis uses the calibration time of the event received by the manufacturing execution system as the sorting basis, while retaining the local occurrence time recorded by the equipment controller. The processor corrects the local occurrence time based on the time difference of the most recent successful interaction between the equipment controller and the manufacturing execution system. If the pressure holding completion event of the same product is later than the demolding confirmation event, or the cooling start event is later than the cooling end event, it is determined that there is an out-of-order situation. The processor does not directly overwrite the original record, but puts the event into the set to be corrected, and constructs a stage duration reference interval based on adjacent tasks that have been completed under the same process version and the same raw material batch.

[0062] The candidate time for a missing event can be determined using the following formula:

[0063] ;

[0064] in, This indicates the time of reconstruction of the missing event for production task i. This represents the set of candidate times defined by the previous and next confirmed events, where m represents the number of reference stages adjacent to the missing event. This represents the confidence weight of the k-th reference stage. Indicates the time of adjacent confirmed events. This represents the median duration of the corresponding stage in the historical tasks for the same process version and raw material batch. This indicates the penalty when a candidate time conflictes with a confirmed event, a permitted demolding interval, or a device occupancy record. This represents the conflict penalty coefficient, which takes a finite value when the candidate time falls after the end of injection molding but before the demolding confirmation, and a lower value when the candidate time is later than the demolding confirmation. Take the larger value to exclude candidate moments with temporal inconsistencies.

[0065] Please refer to Figure 5 , Figure 5 The horizontal direction represents the Chinese sampling stage divided according to the production time sequence, and the vertical direction represents the cost of candidate missing events. The lower the cost, the more consistent the candidate event is with the state changes of the preceding and following processes, event chain constraints, and production line cycle time. Figure 5 The inner surfaces respectively display the cost curves for candidate events such as missing mold closing confirmation, missing injection start, missing pressure holding end, missing cooling completion, missing demolding and transfer, and missing inspection and entry. Figure 5 The results were calculated using the missing event reconstruction formula in the text, taking into account the occurrence time of candidate events, the state difference between adjacent events, the matching degree of process constraints, and the time continuity.

[0066] from Figure 5 It can be seen that the low-cost intervals of different candidate events do not completely overlap. Missing mold closing confirmation events typically have lower costs in the early stages, indicating that the system can infer that the mold closing confirmation event may have been missed even when equipment temperature, mold pressure, and mold opening / closing status have changed synchronously. Missing injection start events show a significant trough in the early to mid-stages, reflecting a strong coupling relationship between injection pressure increase and conveyor position maintenance. The cost troughs for missing pressure holding end and cooling completion events shift towards the mid- and late-stages, consistent with the physical sequence of pressure holding release, cooling and shaping, and demolding preparation in resin manhole cover molding.

[0067] The fact that the missing curve for inbound inspection only enters the low-cost range in the later part indicates that its judgment relies more on subsequent fields such as demolding and transfer, completion of trimming, and changes in the cache queue.

[0068] The time obtained through event reconstruction is only used to restore the temporal continuity in the process state vector and is not used to forge irreversible actions that the equipment has already performed. If the missing event is an event that requires clear execution evidence, such as demolding confirmation, quality release, or equipment action completion, the processor will only mark the reconstruction result as an estimated time and keep the corresponding irreversible state as pending verification. Only after the subsequent process terminal returns a confirmation event or the equipment controller returns a verifiable record will the estimated state be converted into a confirmed state and a new state version be generated.

[0069] In a preferred embodiment, the process constraint propagation graph uses the process state node corresponding to the production task as the vertex, the state dependency relationship between successive processes as directed edges, and encodes the process maturity condition, equipment availability, process continuity in the same batch, and downstream work-in-process capacity as edge constraints. The processor determines the activation state of the edge according to the constraint satisfaction interval corresponding to each directed edge, and only allows the process state debt to propagate through the directed edge that is in the active state and points to the incomplete process, and truncates the propagation path corresponding to the completed irreversible process.

[0070] The process maturity condition is encoded as a combination constraint of the completion time of the preceding state and the minimum duration. The equipment availability capacity is encoded as a matching constraint between the process version and the candidate equipment capability label. The process continuity of the same batch is encoded as the condition that allows the insertion of other batch tasks between adjacent tasks. The downstream work-in-process capacity is encoded as the difference between the acceptable quantity and the expected occupied quantity of the target process within a given time period. The processor saves the condition expression and data source for each constraint edge. When the process event or equipment reservation result changes, only the constraint edges that reference the changed field are recalculated, and the constraint edges that do not reference the changed field remain in their original state.

[0071] The activation state of a constraint edge is determined by whether the constraint constitutes an actual restriction on the next state transition of the current task. If the product has been demolded, the process maturity constraint edge pointing from the cooling or curing node to the demolding node is closed. If a piece of equipment is unavailable during the planned time period, the equipment capacity constraint edge pointing from the injection node to the corresponding equipment execution node remains active and prevents the task from passing. If the downstream process still has acceptable space and the expected occupancy will not exceed the capacity, the capacity constraint edge can absorb the preceding deviation without continuing to propagate. If the expected occupancy reaches the capacity boundary, the capacity constraint edge is in an active state and transmits the state debt to the upstream task release node.

[0072] Furthermore, for each active directed edge, the processor generates a propagation allocation coefficient that changes with the process event based on the remaining process capacity of the processes at both ends of the edge, the current work-in-process inventory, the allowed time interval for the target task to be transferred, and the reserved time period of the associated equipment. When there are multiple propagable paths for the same process state debt, the process state debt is allocated according to the continuous product of the propagation allocation coefficients on each propagation path, and the allocated debt is written to the corresponding subsequent process state node until a frozen node, a quality locked node, or an inactive directed edge is reached.

[0073] The propagation distribution coefficient can be calculated using the following formula:

[0074] ;

[0075] in, This represents the propagation allocation coefficient from node u to node v. This indicates the activation flag of a directed edge; the value is 1 when the edge is active and 0 when it is inactive. This indicates the current work-in-process inventory for the process corresponding to node v. This represents the number of tasks expected to flow from node u to node v. This represents the capacity that node v can accommodate within the corresponding time period. This indicates the idle time available for subsequent processes to absorb deviations within the task's allowed transfer window. Indicates the allowed transposition window length for the task. This indicates that a value less than 0 is taken as 0, for example, when the downstream occupied capacity equals the expected inflow capacity and there is no idle time. If we take 1, the state debt propagates completely backward, and there is still a considerable amount of idle time downstream. As the value approaches zero, the state debt is absorbed by downstream buffers.

[0076] When an upstream node connects to multiple candidate subsequent process nodes at the same time, the processor first excludes paths that do not meet the constraints of process version, quality status, or equipment capability. Then, it normalizes the propagation allocation coefficients of the remaining paths and allocates state debt according to the normalization result. If all candidate paths are excluded, the state debt is retained in the current node and the blocking source is marked in reverse to the corresponding task release node. The manufacturing execution system suspends the release of related new tasks without changing the tasks that are already in the irreversible execution stage.

[0077] In a preferred embodiment, please refer to Figure 6For each production task, the processor jointly compares the actual state transition time, the earliest transition time that meets the process maturity conditions, and the acceptable time that meets the downstream capacity conditions to form a process state debt with a direction mark. For multiple production tasks within the same rolling optimization window, a debt sequence is established according to the arrival order and accumulation level of the process state debt in the process constraint propagation diagram. The debt sequence is used to limit the release order of the tasks to be adjusted, the equipment reservation period, the transition time, and the insertion position of the inspection task.

[0078] The debt sequence is sorted primarily by the moment when the state debt first arrives at the affected node, by the number of activated constraint edges that the state debt passes through from the source node to the current node, and by the current process and irreversible state of the product as the adjustment boundary. State debts that arrive earlier and have a deeper propagation level are given priority to enter the rolling optimization window. Tasks in a quality-locked state do not participate in the normal sorting process but enter the lock queue. Tasks that have completed injection molding but have not yet reached the demolding maturity condition remain in the current equipment occupancy relationship. Tasks that have not yet started feeding or injection molding are allowed to absorb upstream or downstream deviations by adjusting the release time.

[0079] The time boundary of the rolling optimization window is determined by the current time of the earliest affected task and the upper limit of the allowed reordering of the latest adjustable task. The task boundary is determined by the production tasks covered by the state debt propagation path, adjacent tasks occupying the same equipment time period, and tasks sharing the same downstream capacity constraint. The processor does not include tasks that are not associated with the propagation path in terms of equipment, batch, time window, or capacity in the optimization window, and keeps the original task order, equipment reservation, and reordering arrangement unchanged in the stable area.

[0080] In a preferred embodiment, when the processor establishes a rolling optimization window, it sets the process tasks that have already started and are irreversible as frozen tasks, and uses their equipment occupancy range and the earliest subsequent release time as fixed constraints. For non-frozen tasks, a candidate adjustment set is constructed based on the debt sequence. Only the release time of the task to be fed, the order of injection molding tasks, the start time of the next production cycle, the product transfer time, and the insertion position of the inspection task are adjusted. Candidate adjustment schemes that violate the allowable demolding time range, the continuity of the same batch of processes, or the downstream work-in-process capacity constraints are eliminated.

[0081] The determination of frozen tasks is based on irreversible events. Tasks that have already started injection molding cannot be canceled to the waiting state. Tasks that have already completed demolding cannot be rescheduled for cooling or curing stages. Tasks that have reached a locked conclusion in quality inspection cannot enter the normal transfer queue. The processor writes the equipment time slot occupied by the frozen task, the expected completion time, and the earliest release time into the fixed constraint set. Candidate adjustment schemes can only move non-frozen tasks around the fixed constraint set and cannot cover fixed time slots or change the process version of frozen tasks.

[0082] The candidate adjustment set is generated through local enumeration and constraint pruning. The processor selects the task to be adjusted according to the debt sequence. For each task, it generates candidate arrangements that maintain the original time, move forward to the nearest available time period, move backward to the next available time period, or exchange the order with adjacent tasks that meet the same equipment capacity and batch continuity conditions. Each time a local arrangement is generated, the allowable demolition time interval, equipment reservation conflict, downstream capacity, and the acceptance status of the task are checked. Branches that violate any constraint are terminated immediately and do not continue to expand subsequent task arrangements.

[0083] When the injection cycle is extended, causing the subsequent demolding time to be delayed, the processor first checks whether the deviation can be absorbed by delaying the release time of tasks that have not yet been fed. If the downstream process has an idle interval, the injection sequence of the subsequent tasks is retained and the transfer time is adjusted. If the downstream capacity has reached the boundary, the non-frozen tasks that share equipment or capacity constraints with the affected tasks are included in the candidate adjustment set, and their equipment reservation and inspection insertion positions are rearranged. Throughout the process, the status and occupancy relationship of the frozen tasks that have been completed are not modified.

[0084] The candidate solutions for rolling optimization adopt hierarchical comparison instead of simply adding up indicators with different physical meanings. The processor compares the residual amount of process state debt, the number of layers of debt propagation across processes, the subsequent release offset of frozen tasks, and the change in the original task order in turn. Only when the indicators of the current layer are the same will the indicators of the next layer be compared, so that the resolution of state debt takes precedence over the stability of task order, while avoiding the need to change the task order on a large scale to reduce the amount of local debt.

[0085] The hierarchical evaluation of candidate solutions can be expressed as:

[0086]

[0087] ;

[0088] in, This represents a candidate adjustment scheme. This represents the globally optimal scheduling adjustment scheme obtained through filtering. This represents the set of candidate solutions that pass all constraint checks. This represents the priority evaluation vector. This represents the set of tasks in the scrolling optimization window. This represents the residual amount of the process state of task i after executing the candidate solution. This indicates the number of process layers that the state debt continues to traverse. This represents the total offset at the time when the frozen task is subsequently released. This indicates the number of task pairs whose order has changed relative to the original plan. This means minimizing the hierarchy from left to right according to the vector components. For example, if two candidate solutions have the same amount of debt residuals, the one with the smaller number of propagation layers is selected. If the first three items are still the same, the one with the smaller change in the original task order is selected.

[0089] In a preferred embodiment, when multiple candidate adjustment schemes meet the constraints, the processor calculates the residual amount of process status debt, the number of debt propagation layers across processes, and the subsequent release offset of frozen tasks after the execution of each candidate adjustment scheme. The processor then selects the target adjustment scheme in a hierarchical order, prioritizing the residual amount of process status debt, followed by the number of debt propagation layers across processes, and then prioritizing the subsequent release offset of frozen tasks. When different candidate adjustment schemes have the same value at the same level, the candidate adjustment scheme with the smallest change in the original task order is used as the basis for generating task release, equipment reservation, and sequence transfer instructions in the rolling optimization window.

[0090] Once the target adjustment plan is determined, the processor generates material feeding and injection molding task instructions based on the task release time in the plan, generates an injection molding equipment execution queue based on the equipment reservation time period, generates a process terminal receiving task based on the transition time, and generates a quality inspection queue based on the inspection task insertion position. Each instruction carries a product identifier, process identifier, planned execution time period, preceding completion conditions, quality status requirements, and control sequence version. Before execution, the equipment controller and process terminal verify the preceding completion conditions and version identifier. Instructions that do not meet the conditions remain waiting and are not triggered prematurely.

[0091] In a preferred embodiment, the processor generates a task control sequence with version identifiers for the rolling optimization window, and sends the task release instruction, equipment reservation instruction, and sequence switching instruction to the corresponding injection molding equipment controller and process terminal, respectively. The processor receives the execution start event, execution completion event, and execution rejection event of each instruction, verifies the task control sequence according to the task identifier, process identifier, and version identifier carried by the event, isolates and stores feedback with inconsistent versions, and updates the process state vector, process constraint propagation graph, and debt sequence only based on the execution feedback with consistent versions.

[0092] The version of the task control sequence is determined by the task status version, the rolling optimization window identifier, and the sequence generation order. Only one currently executable version can exist for the same task in the same process. After a new version is generated, the processor sends a version replacement instruction to the relevant equipment controller and process terminal. The equipment side saves the current executable version and refuses to execute the old version instruction. If the equipment has already started an irreversible action based on the old version, the equipment returns to the execution start event, and the manufacturing execution system puts the corresponding task into a frozen state and recalculates the local plan with the actual action as the new fixed constraint.

[0093] The execution start event is used to confirm that the device has occupied the corresponding time period. The execution completion event is used to update the irreversible state and release the device occupation. The execution rejection event records at least the rejection reason, the current device state, and the control sequence version saved on the device side. The processor classifies the rejection reason. If the preceding sequence is not completed or the quality is not released, the task is kept waiting. If the device is unavailable, the device capability constraint edge is reactivated. If the versions are inconsistent, the feedback is written to the isolation area and the current execution record of the device is queried to avoid the same feedback changing the task state and the device reservation state at the same time.

[0094] In a preferred embodiment, after completing a version-consistent execution feedback update, the processor extracts the debt difference and directed edge activation state changes of each process state node before and after the update, determines the actual debt resolution path corresponding to the current task control sequence, and when the actual debt resolution path is inconsistent with the predicted resolution path used when generating the task control sequence, only the production tasks and related processes involved in the difference path are re-formed into local rolling optimization windows, and the original task control sequence version, equipment reservation time period and transition time of the production tasks outside the difference path are used.

[0095] The predicted resolution path is saved when the target adjustment plan is generated. It records the state nodes where the debt volume is expected to decrease, the constraint edges that are expected to be closed or activated, and the downstream free intervals that are expected to absorb the deviation. After the feedback update is executed, the processor calculates the actual debt volume change of each node and the actual state change of each constraint edge. The actual changes are compared with the prediction records item by item. When the debt volume change of a node is opposite, the debt volume does not decrease as planned, the activation state of the constraint edge is inconsistent with the prediction, or the downstream free interval is occupied by other frozen tasks, the corresponding node and edge are marked as difference path elements.

[0096] The set of differential paths can be constructed as follows:

[0097] ;

[0098] in, This represents the set of difference paths that need to be recalculated. Represents the set of process status nodes. Represents the set of directed edges. This represents the actual change in the amount of debt at node u. This indicates a forecast of changes in debt volume. This represents the allowable difference determined based on the event time resolution and the fluctuation in stage duration. and These represent the actual and predicted activation states of a directed edge, respectively. For example, if the change in node debt falls within the allowable difference and the edge activation states are consistent, the edge will not be entered into the target state. When the change in debt exceeds the allowable difference or the edge state changes in the opposite direction, the corresponding node or edge will be included in the recalculation scope.

[0099] When re-forming the local rolling optimization window, the processor takes the nodes and edges in the differential path set as the core, expands forward to the nearest frozen node or unaffected task release node, and expands backward to inactive constraint edges or capacity-free nodes that can absorb the remaining state debt. Tasks that have no connection with the differential path in terms of equipment occupation, batch continuity, time window or capacity maintain the original control sequence version, and the equipment reservation time and turnaround time remain unchanged, thereby avoiding the rescheduling of the entire production line triggered by a single execution deviation.

[0100] Specifically, when a resin manhole cover has not reached the allowable demolding state by the expected cooling completion time, and the actual debt of the cooling or curing node decreases less than the predicted value, the processor will include the node, its process maturity constraint edge pointing to the demolding node, and adjacent tasks that share capacity with the demolding station into the differential path set. The current product that has completed injection molding and is in a frozen state will maintain the equipment occupancy relationship, and subsequent tasks that have not yet been fed will be allowed to postpone the release time. Other production tasks that do not share equipment and downstream capacity will continue to be executed according to the original control sequence.

[0101] In this embodiment, the manufacturing execution system forms a continuous closed loop. Process events update the process state vector, the process state vector changes the node and edge states of the process constraint propagation graph, the process constraint propagation graph limits the propagation direction and scope of process state debt, the process state debt determines the task boundary and adjustment priority of the rolling optimization window, the target adjustment scheme generates a task control sequence with version identifiers, and the execution feedback with consistent versions updates the process state vectors again. Each link retains the correspondence between product identifier, process identifier, and version identifier, so that the data processing results can be implemented into verifiable injection molding production line process control actions.

[0102] Furthermore, after a task is completed, the Manufacturing Execution System (MES) writes the final process state vector, state version mapping relationship, constraint edge change record, process state debt change, rolling optimization window range, target adjustment scheme, and execution feedback into the production history. If subsequent tasks of the same process version and raw material batch are missing events, the stage duration reference data can be selected from the production history. When equipment capacity or downstream capacity changes, the propagation path under the same constraint combination can be extracted from the production history. However, historical records are only used as candidate references, and the process maturity conditions, irreversible state, and quality lock-in state of the current task are still used as direct constraints for plan generation.

[0103] In a preferred embodiment, if the production line executes multiple resin manhole cover specifications or multiple process versions simultaneously, the processor maintains the rules for allowed demolding time intervals and equipment capacity matching relationships according to the process version. Cross-process version tasks only establish constraint edges when they share the same equipment reservation time period or downstream capacity, and do not automatically establish propagation relationships because they are in the same production plan. Tasks in the same raw material batch determine whether other tasks are allowed to be inserted according to the actual process continuity requirements, thereby making the propagation range of process status debt determined by the actual technical dependence rather than by the planned batch name.

[0104] In this embodiment, the quality lock status participates in the joint verification of the process state vector, process constraint propagation graph and task control sequence. For products that have not yet been released or have formed a lock conclusion after quality inspection, the constraint edges pointing to the normal subsequent process are closed, and the constraint edges pointing to the isolation processing status are activated. Rolling optimization does not bypass the quality status by adjusting the transfer time. After receiving the transfer instruction, the equipment controller or process terminal also needs to verify that the quality status requirements carried are consistent with the current task status version. If they are inconsistent, a rejection event is returned.

[0105] Specifically, the Manufacturing Execution System (MES) stores the planning status, equipment execution status, and product process status of the same task using different fields. The planning status indicates the next action and time arranged by the system, the equipment execution status indicates the action that the controller has started or completed, and the product process status indicates the actual maturity stage determined based on the injection completion, holding pressure, cooling or curing, and demolding events. The processor generates the next process instruction only when the three types of states meet the predetermined transition relationship. For example, when the planning status is pending demolding, the equipment execution status is cooling completed, and the product process status has entered the allowable demolding time interval, the demolding execution instruction is allowed to be generated.

[0106] In a preferred embodiment, the processor sets a closing condition for the rolling optimization window. When the state debt of all tasks within the window is reduced to an allowable range, all tasks are in a frozen state, and the remaining state debt reaches an inactive constraint edge that can absorb the deviation or a quality lock causes the task to exit the normal process route, the rolling optimization window is closed and the final control sequence version is saved. If a new process event changes the state of the relevant nodes again after closing, a new rolling optimization window is established according to the new state debt propagation path, and the candidate solutions in the already closed window are not directly restored.

[0107] This embodiment also allows the process terminal to input demolding confirmation, trimming completion, and quality release events that the equipment controller cannot directly provide. The input events must be associated with the product identifier, process identifier, occurrence time, and current control sequence version. The processor processes the events according to the same version verification rules as the equipment events. After the event passes verification, the corresponding irreversible state is updated and the constraint edge is recalculated. When the event fails verification, it enters the isolation area, thereby maintaining the consistency of the state update logic between manually confirmed events and automatic equipment events.

[0108] Through the above implementation methods, every state change of the resin manhole cover production task from release to quality inspection can be converted into a state field, constraint edge, or execution event with a reference source. Process maturity constraints, equipment capacity constraints, and upstream and downstream load-bearing constraints participate in the calculation through the same propagation graph. The process state debt expresses the inconsistency between the previous state and the subsequent acceptance state through directional time deviation. Rolling optimization only modifies non-frozen tasks that are actually related to the deviation. The task control sequence with version identification closes the calculation results with the on-site execution feedback, thus completing the collaborative management and control between various processes in the resin manhole cover injection molding production line.

Claims

1. A collaborative control system for resin manhole cover injection molding production line processes, characterized in that, Includes a processor, memory, and a data interface for communicating with the injection molding equipment controller and process terminals. The memory stores instructions executable by the processor, which instruct the processor to: establish a process state vector for each resin manhole cover production task that is updated with process events; A process constraint propagation diagram is constructed based on process maturity constraints, equipment capacity constraints, and upstream and downstream capacity constraints. Calculate the process state debt of each task's actual state transition time relative to the target transition time and propagate it along the process constraint propagation graph; A rolling optimization window is formed using the affected tasks and related processes to generate task release, sequence transfer, and equipment execution instructions, and the process state vector and process state debt are updated based on the execution feedback.

2. The collaborative control system for resin manhole cover injection molding production line processes according to claim 1, characterized in that, The process status vector is associated with both the product identifier and the production task, and includes the raw material batch identifier, process version, current process identifier, completed process stage, actual injection molding end time, pressure holding completion status, remaining cooling or curing time, allowable demolding time interval, expected entry time of subsequent processes, and quality lock status. The processor updates the irreversible state field and the variable state field in the process state vector according to the timestamped process event returned by the device controller, and retains the state version mapping relationship before and after the update.

3. The collaborative control system for resin manhole cover injection molding production line processes according to claim 1, characterized in that, The process constraint propagation graph uses the process state node corresponding to the production task as the vertex, the state dependency relationship between the preceding and following processes as the directed edge, and encodes the process maturity condition, equipment availability, process continuity in the same batch, and downstream work-in-process capacity as edge constraints respectively. The processor determines the activation state of each edge based on the constraint satisfaction interval corresponding to each directed edge, and only allows the process state debt to propagate through directed edges that are in the active state and point to processes that have not yet been completed, and truncates the propagation path corresponding to completed irreversible processes.

4. The collaborative control system for resin manhole cover injection molding production line processes according to claim 1, characterized in that, For each production task, the processor jointly compares the actual state transition time, the earliest transition time that meets the process maturity conditions, and the acceptable time that meets the downstream capacity conditions to form a process state debt with direction marking. For multiple production tasks within the same rolling optimization window, a debt sequence is established according to the arrival order and accumulation level of the process status debt in the process constraint propagation diagram, and the debt sequence is used to limit the release order, equipment reservation time period, transfer time and inspection task insertion position of the tasks to be adjusted.

5. The collaborative control system for resin manhole cover injection molding production line processes according to claim 2, characterized in that, The processor maps injection end events, pressure holding completion events, cooling or curing stage events, demolding confirmation events, and quality release events to a unified process time scale axis, and reconstructs missing or out-of-order events based on the duration between adjacent valid events. During the reconstruction process, the set of historical tasks that can participate in the reconstruction is limited by the process version and raw material batch identifier in the process state vector. The stage duration distribution of the historical task set is used to determine the candidate time intervals of missing events, and the candidate time intervals that have a time sequence conflict with the subsequently confirmed events are eliminated.

6. The collaborative control system for resin manhole cover injection molding production line processes according to claim 1, characterized in that, For each directed edge that is in an active state, the processor generates a propagation allocation coefficient that changes with the process event based on the remaining process capacity of the processes at both ends of the edge, the current work-in-process inventory, the allowed time interval for the target task to be transferred, and the time interval reserved by the associated equipment. When there are multiple propagable paths for the same process state debt, the process state debt is allocated according to the continuous product of the propagation allocation coefficients on each propagation path, and the allocated debt amount is written into the corresponding subsequent process state node until a frozen node, a quality locked node, or an inactive directed edge is reached.

7. The collaborative control system for resin manhole cover injection molding production line processes according to claim 4, characterized in that, When the processor establishes the rolling optimization window, it sets the process tasks that have already started execution and are irreversible as frozen tasks, and uses their equipment occupancy range and the earliest subsequent release time as fixed constraints. For non-frozen tasks, a candidate adjustment set is constructed based on the aforementioned debt sequence. Only the release time of the task to be fed, the order of injection molding tasks, the start time of the next production cycle, the time of product transfer, and the insertion position of the inspection task are adjusted. Candidate adjustment schemes that violate the allowable demolding time interval, the continuity of the same batch of processes, or the capacity constraints of downstream work-in-process are eliminated.

8. The collaborative control system for resin manhole cover injection molding production line processes according to claim 4, characterized in that, The processor generates a task control sequence with version identifier for the scrolling optimization window, and sends the task release instruction, equipment reservation instruction and sequence transfer instruction to the corresponding injection molding equipment controller and process terminal respectively. The processor receives execution start events, execution completion events, and execution rejection events for each instruction. It verifies the task control sequence according to the task identifier, process identifier, and version identifier carried by the event. It isolates and stores feedback with inconsistent versions and updates the process state vector, the process constraint propagation graph, and the debt sequence only based on execution feedback with consistent versions.

9. A collaborative control system for resin manhole cover injection molding production line processes according to claim 7, characterized in that, When multiple candidate adjustment schemes meet the constraints, the processor calculates the residual amount of process state debt, the number of debt propagation layers across processes, and the subsequent release offset of frozen tasks after the execution of each candidate adjustment scheme. The processor then selects the target adjustment scheme in a hierarchical order of priority: the residual amount of process state debt, the second priority of the number of debt propagation layers across processes, and the subsequent release offset of frozen tasks. When different candidate adjustment schemes have the same value at the same level, the candidate adjustment scheme with the smallest change in the original task order is used as the basis for generating task release, equipment reservation and sequence transfer instructions in the rolling optimization window.

10. A collaborative control system for resin manhole cover injection molding production line processes according to claim 8, characterized in that, After completing a version-consistent execution feedback update, the processor extracts the debt difference and directed edge activation state changes of each process state node before and after the update to determine the actual debt resolution path corresponding to the current task control sequence. When the actual debt resolution path is inconsistent with the predicted resolution path used when generating the task control sequence, only the production tasks and related processes involved in the difference path are re-formed into local rolling optimization windows, and the original task control sequence version, equipment reservation time period and transition time of the production tasks outside the difference path are retained.

Images