An energy-saving control method for an automated production line combining energy recovery and task reassignment

Abstract

The present application belongs to the field of automation technology, and particularly relates to an energy-saving control method for an automated production line combining energy recovery and task reassignment; the method comprises the following steps: step 1: in the automated production line, collecting equipment running state parameters and instantaneous power consumption parameters at the entrance of each production branch; labeling each record with an event category, a time stamp and a corresponding equipment number; step 2: on the basis of the generated event trigger table, simultaneously establishing a double-channel collaborative scheduling mechanism for energy recovery and task reassignment to ensure the simultaneous optimization of energy recovery efficiency and production rhythm stability, and finally solidifying the recovery execution instructions and task reassignment instructions; step 3: issuing the generated recovery execution instructions and task reassignment instructions to the corresponding execution units for implementation, and dynamically adjusting and compensating the scheduling according to the difference results; the simultaneous optimization of energy recovery and task reassignment is realized, and the energy utilization rate is significantly improved while the stability of the production rhythm is maintained.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of automation technology, and particularly relates to an energy-saving control method for an automated production line combining energy recovery and task reassignment. BACKGROUND

[0002] In the field of modern manufacturing, automated production lines have become the core infrastructure for improving production efficiency, reducing labor costs, and ensuring product consistency. With the increasing degree of industrial automation, the operation mode of production lines tends to be multi-variety, small-batch, and flexible, and the collaboration and scheduling requirements between devices are increasing. However, in actual operation, automated production lines generally face problems such as low energy utilization efficiency, large load fluctuations, and unstable production rhythm. In particular, in scenarios involving multiple devices and multiple stations, energy waste is more obvious. Existing production line control systems mostly focus on task scheduling and production capacity utilization optimization, and lack consideration of energy recovery and utilization, resulting in the remaining energy of some devices in idle state not being effectively utilized, or the recoverable energy window being occupied and lost due to device switching and task conflicts.

[0003] Among the disclosed automated production line control technologies, one common method is a task allocation strategy based on a fixed schedule. This strategy usually sets the task execution time and sequence of each device in the production planning stage, and strictly follows the schedule during operation. When there is idle time between device task execution, although some advanced systems can put the device into a low-power mode, they lack a dynamic scheduling mechanism for energy recovery, resulting in the inability to efficiently recover mechanical inertia energy, braking energy, or residual thermal energy during the idle period. In addition, this method relies on manual intervention or simple priority override for task conflict handling, lacks dynamic reassignment capabilities across tasks and devices, and is prone to mutual interference between energy recovery and task execution.

[0004] Another existing technology attempts to introduce an energy management module into the production control system, which monitors device power consumption in real time, determines whether it is in a recoverable energy state, and feeds the recovered electrical energy back to the energy storage system. This type of system can recover energy when a device switches from a working state to an idle state, such as converting mechanical energy into electrical energy using the regenerative braking function of a variable frequency drive. However, the scheduling logic of these methods is relatively simple, focusing only on the recovery process itself, and lacks close integration with task scheduling. When multiple devices simultaneously have energy recovery conditions, the system lacks global coordination and conflict resolution mechanisms, which can result in excessive concurrent energy recovery devices leading to reduced energy recovery efficiency, or resource contention with other critical tasks. SUMMARY

[0005] The main purpose of the present application is to provide an energy-saving control method for an automated production line combining energy recovery and task reassignment, which comprises the following steps:

[0006] Step 1: In the automated production line, collect the equipment operating state parameters and the instantaneous power consumption parameters at the entrance of each production branch; generate two types of event records in the monitoring unit according to the collection results, namely load demand events and energy recovery feasible events; store all event records in chronological order to form an event trigger table, and save the safety interlocking conditions in the monitoring unit as subsequent scheduling constraints; label each record with event category, time stamp and corresponding equipment number;

[0007] Step 2: Based on the generated event trigger table, a dual-channel collaborative scheduling mechanism is established simultaneously for energy recovery and task reassignment; according to the preset priority rules, the executable energy recovery window is locked in the recovery channel, and the available time slots for task allocation that do not conflict with recovery are allocated in the reassignment channel; through iterative promotion and conflict resolution, the window and time slot allocation are dynamically adjusted in multiple rounds of scheduling to ensure the optimization of energy recovery efficiency and production rhythm stability, and finally solidified into recovery execution instructions and task reassignment instructions;

[0008] Step 3: The generated recovery execution instructions and task reassignment instructions are issued to the corresponding execution unit for implementation, and the actual energy recovery amount and task completion status are collected during the execution process, which are compared with the instruction target, and dynamic adjustment and compensation scheduling are performed according to the difference results.

[0009] Further, in step 1, the equipment operating state parameters include two state identifiers, working and idle; the instantaneous power consumption parameter is the real-time detected electric power value; according to the collection results, two types of event records are generated in the monitoring unit, namely load demand events and energy recovery feasible events; the condition for triggering the load demand event record is that the state of the device at the collection point is switched from idle to working, and the condition for triggering the energy recovery feasible event record is that the state of the device at the collection point is switched from working to idle and the instantaneous power consumption parameter is lower than the preset power threshold.

[0010] Further, in step 2, the process of establishing a dual-channel collaborative scheduling mechanism simultaneously for energy recovery and task reassignment based on the generated event trigger table comprises: dividing the event trigger table into a recovery channel queue and a reassignment channel queue according to the event category; searching for load demand events in a preset time matching window for each energy recovery feasible event in the recovery channel queue, the time matching window is set by system parameters, forming an initial pairing set; generating a shadow task block for each load demand event in the reassignment channel queue, the shadow task block occupies the time slot resources of the task reassignment candidate set, but does not immediately trigger actual scheduling.

[0011] Further, the process of preferentially locking executable energy recovery windows in the recovery channel and allocating available time slots that do not conflict with recovery to tasks in the redistribution channel according to the preset priority rule comprises: in the recovery channel, starting from the highest priority pair, locking the corresponding energy recovery window and marking the time slot as unavailable for other events; in the redistribution channel, starting from the shadow task block with the lowest switching cost, attempting to move it into the earliest feasible time slot of the idle device, if the time slot conflicts with the locked energy recovery window, sequentially selecting the next device to rematch the time slot in the candidate set until a non-conflicting time slot is found or marked as waiting.

[0012] Further, the preset priority rule is: first, sorting according to the time relationship label; when the time relationship label is the same, sorting according to the device feasibility label; when both are the same, finally sorting according to the candidate sorting order; performing validity screening on the sorted pairs one by one, the screening condition is: the energy recovery window is in the open state, the window time range does not overlap with the locked window, the load demand event is not marked as invalid, and the corresponding device is not in the maintenance downtime state; the pair that does not meet the screening condition is marked as invalid in this stage and moved to the end of the queue, and does not participate in locking in this stage.

[0013] Further, in step 2, the process of dynamically adjusting window and time slot allocation in multiple rounds of scheduling through iterative promotion and conflict resolution to ensure simultaneous optimization of energy recovery efficiency and production rhythm stability comprises: using the iteration rule of recovery channel priority and redistribution channel supplement, in each iteration, first confirm all the windows locked in the recovery channel of this round, and then insert the shadow task block in the waiting state into the time slot position unlocked by this round; when the windows locked in the recovery channel in a certain iteration are less than the preset proportion, the preset proportion is set by the system parameter, roll back all operations in this round and execute the next round of matching from the pair set sorting; perform two-stage resolution on all allocated energy recovery windows and task time slots, the first stage deletes pairs that violate the safety interlocking condition, and the second stage solves the problem of multiple task competition in the same time slot, and the competition result is determined in order of pair priority, switching cost and device idle rate, and the low priority is returned to the next iteration.

[0014] Further, in step 2, the process of finalizing the recovery execution instruction and the task redistribution instruction comprises: converting the non-conflicting energy recovery window into a recovery execution instruction, converting the shadow task block into an actual redistribution instruction, and recording the event set that is not matched successfully for priority processing in the next cycle.

[0015] Further, step 3 specifically includes: routing the recovered execution instructions of step 2 to the energy recovery execution unit and the task reassignment scheduler respectively, and sequentially completing the reception confirmation and the readiness confirmation when issuing, and establishing the placeholder reservation on the corresponding device; setting the execution monitoring points on the recovery side and the reassignment side respectively during the implementation process, forming the recovery execution track and the task execution track, and performing consistency comparison with the target start time, the target end time and the device constraint, and outputting the deviation classification label; when any of the insufficient recovery, early termination, non-execution, delayed completion or interruption occurs, triggering dynamic adjustment and compensation scheduling, performing standby window replacement in the energy recovery window set, recovering by standby devices of the same type or segmenting and splicing recovery, performing the same device shift, cross-device handover or segmented execution in the task reassignment candidate set, and performing safety interlocking review and conflict rapid resolution on the newly generated instructions and the existing placeholder reservation, processing cross-channel conflicts according to the recovery priority and reassignment compensation rules, and processing same-side conflicts according to the rules of earlier start time priority and higher device idle rate priority; through the submission mechanism that takes effect all at once in the form of all success or all failure, all the instructions that pass the review are written to the real state of the execution side and the scheduling side at once, and if the submission fails, the whole is rolled back and the sorting result is reserved to enter the next round of adjustment.

[0016] Further, in step 3, when there is no instruction to be retried and no unresolved conflict in an execution cycle, it is determined to be convergent, and the execution result, adjustment record and deviation classification label are written back to the event trigger table, the energy recovery window set and the task reassignment candidate set, and the unfinished matters are moved to the next cycle priority queue.

[0017] The energy-saving control method of the automatic production line combined with energy recovery and task reassignment has the following beneficial effects: the synchronization optimization of energy recovery and task reassignment is realized, the energy utilization rate is significantly improved, and the stability of the production rhythm is maintained. By collecting the equipment operation state parameters and the instantaneous power consumption parameters at the entrance of the production branch in real time, the load demand events and the energy recovery feasible events can be accurately identified, and an event trigger table is constructed with unified time stamps, equipment numbers and safety interlocking conditions to provide reliable data basis for subsequent scheduling. In the scheduling process, the recovery channel preferentially locks the executable energy recovery window, and the reassignment channel assigns the available time slots that do not conflict with the recovery, and an iterative promotion and conflict resolution strategy is adopted to ensure that the energy recovery efficiency and the continuity of task execution are considered in multiple rounds of matching. The combination of priority rules and conflict resolution strategies can quickly solve the cross-channel and same-side conflicts, and avoid energy waste or production delay caused by unbalanced scheduling. In the execution stage, the recovery execution instructions and the task reassignment instructions are issued, the placeholders are reserved and the execution is monitored to form a closed-loop control, and dynamic adjustment and compensation scheduling are triggered in the case of insufficient recovery, premature termination, non-execution, delayed completion or interruption, etc. The system target is quickly restored through backup window replacement, cross-device transfer and segmented execution. The one-time effective submission mechanism ensures the consistency of execution and scheduling state, and prevents system instability caused by partial effectiveness. The present application realizes the whole-process automatic management, effectively reduces the mutual interference between the energy recovery process and the production task, improves the utilization rate of recoverable energy, reduces the energy and time loss caused by task switching, equipment idling or conflict, and finally improves the production efficiency while realizing the comprehensive effect of energy saving and consumption reduction. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of the provided drawings.

[0019] Figure 1 A method flowchart of the energy-saving control method of the automatic production line combined with energy recovery and task reassignment provided for the embodiments of the present application;

[0020] Figure 2 An experimental curve graph of the switching cost optimization effect of task reassignment provided for the embodiments of the present application;

[0021] Figure 3 An experimental data graph of the relationship between the equipment power consumption and the preset power threshold provided for the embodiments of the present application. DETAILED DESCRIPTION

[0022] The method of the present application will be further described in detail below in combination with the accompanying drawings and embodiments of the present application.

[0023] Reference Figure 1 An energy-saving control method for an automated production line combining energy recovery and task reassignment, the method comprising:

[0024] Step 1: In the automated production line, collect equipment operating state parameters and instantaneous power consumption parameters at the entrance of each production branch; generate two types of event records in the monitoring unit according to the collection results, namely load demand events and energy recovery feasible events; store all event records in time sequence to form an event trigger table, and save safety interlocking conditions in the monitoring unit as subsequent scheduling constraints; label each record with event category, time stamp and corresponding equipment number;

[0025] In the automated production line, set collection points at the entrance of each production branch, establish a stable data channel between the collection points and the monitoring unit, so that the collection points can continuously obtain equipment operating state parameters and instantaneous power consumption parameters, and return the two types of parameters to the monitoring unit in a unified data structure. The equipment operating state parameters are used to describe the state switching between work and idle of the equipment, and the monitoring unit performs time synchronization and consistency check on the data from the entrances of each production branch, and adds a time stamp to each data to ensure that the information reported by different collection points can be compared and processed under the same time reference. The monitoring unit generates two types of event records in the same processing flow according to the collection results, namely load demand events and energy recovery feasible events; when the equipment operating state parameters show a trend of switching from idle to work and the instantaneous power consumption parameters show corresponding rising characteristics, the monitoring unit generates load demand events; when the equipment operating state parameters show a trend of switching from work to idle and the instantaneous power consumption parameters show a continuous decline and stabilize in a low consumption characteristic interval, the monitoring unit generates energy recovery feasible events.

[0026] To avoid misjudgment caused by transient jitter, the monitoring unit implements de-jittering and short window confirmation strategy on the device running state parameters and instantaneous power consumption parameters, and only solidifies the event record after meeting the stable condition. All event records are stored in chronological order to form an event trigger table, which provides triple indexing by timestamp, device number and event category, supporting subsequent fast retrieval and cross verification of the same device, the same production branch or the same time slice. At the same time of writing into the event trigger table, the monitoring unit loads and saves the safety interlocking conditions, which are derived from the action exclusion constraints of the device, the allowed recovery interval on the energy side, the limitation of production rhythm on task switching and the start-stop rules related to human-machine safety. The monitoring unit takes the safety interlocking conditions as the basis for subsequent scheduling constraints and event validity filtering, used to block window locking or time slot occupation inconsistent with safety requirements in subsequent steps. Each record is labeled with event category, timestamp and corresponding device number, and the monitoring unit generates anti-duplicate identifier and traceability identifier at the same time of labeling, ensuring that only a single valid event is retained under the same timestamp and the same device number, and supporting mapping events with corresponding collection point physical location to quickly locate when an abnormality occurs at the production branch entrance. Through the above process, step 1 not only forms the event trigger table and completely saves the safety interlocking conditions as required, but also normalizes the expression of device running state parameters and instantaneous power consumption parameters in time dimension and device dimension, providing a calculable, traceable and constrained basic data environment for subsequent identification of executable energy recovery windows in the double channel and allocation of available time slots for tasks that do not conflict with recovery, thereby laying a reliable starting point for the overall synergy of the automatic production line energy-saving control method combining energy recovery and task reallocation.

[0027] Step 2: Based on the generated event trigger table, a double-channel collaborative scheduling mechanism is established simultaneously for energy recovery and task reallocation, which selects the executable energy recovery window in the recovery channel according to the preset priority rules, and allocates available time slots for tasks that do not conflict with recovery in the reallocation channel; through iterative promotion and conflict resolution, the window and time slot allocation are dynamically adjusted in multiple rounds of scheduling process to ensure the optimization of energy recovery efficiency and production rhythm stability, and finally solidified as recovery execution instructions and task reallocation instructions;

[0028] Based on the generated event trigger table, first, a flow-oriented double-channel collaborative scheduling mechanism is constructed for energy recovery and task reassignment in the monitoring unit. The monitoring unit scans the event trigger table under the same time reference, directs the records belonging to energy recovery to the recovery channel, and directs the records belonging to load demand to the reassignment channel. The records in the two channels are sorted and filtered by using the preset priority rules, so as to ensure that the executable energy recovery window is locked in the recovery channel, and the available time slot that does not conflict with the recovery is allocated to the task in the reassignment channel. The preset priority rules are used to establish a comparable order relationship for the candidates in the recovery channel and the reassignment channel, so that the records with closer time relationship, higher device feasibility and better candidate order sequence are preferentially involved in the attempt and locking. After the sorting is completed, the monitoring unit enters the round processing flow. In each round, the energy recovery window is attempted to be locked from the candidate with the highest priority in the recovery channel. After the locking is successful, the unavailable mark is set in the corresponding time range of the event trigger table, and the consistency check is performed with the safety interlocking condition. The energy recovery window that passes the check enters the effective set of the current round, and the candidate that does not pass the check is marked as invalid and waits for reevaluation in the next round.

[0029] Subsequently, in the reassignment channel, the available time slot that does not conflict with the recovery is attempted to be occupied from the task with higher priority and lower switching cost, aiming at reducing the switching influence and maintaining the production rhythm. If the current device overlaps with the locked energy recovery window in the target time slot, the next available time slot that does not conflict with the recovery is sequentially searched or the same type of device that is feasible is switched to, while maintaining the start-end relationship and the stability of the production rhythm. If it still cannot be satisfied, the task is recorded as waiting and its sorting position is reserved. In order to realize the iterative promotion and conflict resolution in the multi-round scheduling process, the monitoring unit performs conflict scanning on the locked energy recovery window and the allocated task time slot at the end of each round. The conflicts include the combination that violates the safety interlocking condition and the situation that multiple tasks compete in the same time slot. The monitoring unit first performs hard removal according to the safety interlocking condition, and then makes a decision in the time slot with competition according to the preset priority rules. The allocation with lower priority or greater impact on the stability of the production rhythm is returned to waiting, and the reason for the return is recorded to avoid repeated conflicts in the subsequent rounds.

[0030] To prevent the overall energy recovery efficiency from being reduced due to insufficient contribution of the recovery channel in a round, the monitoring unit sets a recovery channel locking proportion threshold value. When the number of successfully locked energy recovery windows or the covered time length in the recovery channel in the current round is lower than the threshold value, it is determined that the current round matching is not representative, the rollback mechanism is triggered to cancel all the locks and allocations in the round, and the system is restored to the available state before the start of the round. Subsequently, the next round is re-entered without changing the preset priority rules, ensuring that the goal of optimizing energy recovery efficiency and production rhythm stability is not dominated by local errors. To adapt to the growing characteristics of the event trigger table over time, the monitoring unit refreshes the candidate view of the two channels at the start of each round based on the timestamp, only including event participation ranking and matching within the current decision window, and increasing the priority of tasks that have been waiting for more than a set time in the reallocation channel to avoid the cumulative impact of long waiting times on production rhythm stability. At the same time, the energy recovery window that has not been used in the recovery channel for multiple consecutive rounds is lowered in the candidate ranking order in the next round to encourage resources to concentrate on higher value time periods. In the iterative advancement of multiple rounds of scheduling, the monitoring unit generates the intermediate results of the current round at the end of each round, including the list of locked energy recovery windows, the list of allocated available time slots that do not conflict with recovery, and the waiting set, and synchronizes with the state of the corresponding record in the event trigger table. If time overlap changes caused by newly entered events are found during the synchronization stage, the affected entries are immediately re-executed for conflict resolution, and if necessary, a small-scale rollback within the current round is triggered to maintain consistency.

[0031] As the round progresses, when the energy recovery windows of the recovery channel achieve stable coverage within the target time range and the tasks in the reallocation channel can obtain consecutive available time slots without introducing new conflicts, the monitoring unit determines that the iteration converges, enters the solidification phase, and finalizes the conflict-free results after iterative advancement and conflict resolution into recovery execution instructions and task reallocation instructions. The recovery execution instructions specify the start time, end time, involved devices, and safety condition reference identifier of the energy recovery window, the task reallocation instructions specify the available time slots and necessary switching order of the task on the specific device that do not conflict with recovery, and the events that fail to complete allocation in the current multiple rounds of scheduling are retained in the waiting set for subsequent processing period.

[0032] Step 3: The generated recovery execution instructions and task reallocation instructions are issued to the corresponding execution units for implementation, and the actual energy recovery amount and task completion status are collected during execution, compared with the instruction target, and dynamically adjusted and compensated based on the difference results.

[0033] Based on the recycling execution instructions and task reassignment instructions obtained in the preceding steps, the monitoring unit routes the two types of instructions to the energy recycling execution unit and the task reassignment scheduler respectively, and the instruction issuing process completes the reception confirmation and readiness confirmation in sequence, and establishes a placeholder reservation on the corresponding device to freeze the related time slots and energy recycling windows, preventing resource contention caused by concurrent operations. The implementation stage sets execution monitoring points on both the recycling side and the reassignment side at the same time, collects the actual energy recycling amount and task completion status in real time, forms the recycling execution trajectory and task execution trajectory, and performs consistency comparison with the target start time, target end time, and device constraints of the instruction target. The monitoring unit outputs deviation classification labels based on the comparison results to identify situations such as insufficient recycling, premature termination, non-execution, delayed completion, or interruption.

[0034] When the deviation classification labels indicate that there is a situation inconsistent with the instruction target, the monitoring unit immediately triggers dynamic adjustment and compensation scheduling: priority is given to backup window replacement in the energy recycling window set, if the same device cannot meet the requirement, the same type of backup device is used to compensate under the premise of meeting the safety interlocking condition, and if it is still insufficient, segmented splicing recycling is used to maintain the continuity of the energy recycling window as much as possible; in the task reassignment candidate set, the affected tasks are first tried to move backward on the same device to avoid recycling execution, if moving backward still conflicts with the locked energy recycling window, cross-device transfer is performed to migrate the task to a device time slot that is available and does not conflict with recycling, and if migration is not feasible, segmented execution is used to split the occupation and reduce the disturbance to the production rhythm.

[0035] The above dynamic adjustment generates new recycling execution instructions and task reassignment instructions, the monitoring unit performs safety interlocking review on the newly generated instructions and existing placeholder reservations, and executes conflict quick resolution; cross-channel conflicts are handled according to the recycling priority and reassignment compensation rules, and same-side conflicts are handled according to the earlier start time priority and higher device idle rate priority to restore timing consistency at the minimum switching cost. To ensure state consistency and atomicity, the monitoring unit uses a commit mechanism that takes effect all at once in the form of all success or all failure for all instructions that pass the review, simultaneously writes the real state to the execution side and the scheduling side at the commit moment, and if any sub-instruction does not meet the commit condition, the commit is determined to fail, the whole rollback is performed and the ordering result is preserved for the next round of adjustment, ensuring that there is no partial intermediate state left after a failure.

[0036] In the execution process, the monitoring unit continuously updates the recycling execution trajectory and the task execution trajectory, and synchronizes the new deviation classification label and the adjustment record with the event trigger table, the energy recycling window set and the task reassignment candidate set, so that the subsequent iteration can run on the latest constraints and progress. When there is no instruction to be retried and no unresolved conflict in an execution cycle, the monitoring unit determines that the current cycle has reached convergence, writes and archives the execution result, the adjustment record and the deviation classification label, and moves the unfinished items to the next cycle priority queue to ensure that the subsequent cycle continues to advance according to the recycling priority and the reassignment compensation strategy. Through the above closed-loop instruction issuance, consistency comparison, deviation identification, dynamic adjustment and one-time submission mechanism, the execution stage can implement recycling execution instructions and task reassignment instructions into specific occupation of energy recycling windows and available time slots that do not conflict with recycling under the constraints of real device conditions and safety interlocking conditions, ensuring the realization of energy recycling efficiency and maintaining the stability of production rhythm.

[0037] Further, the monitoring unit continuously reads the device running state parameters and instantaneous power consumption parameters at the collection points at the entrances of each production branch, and uniformly labels them with timestamps and corresponding device numbers. The device running state parameters include two state identifiers: working and idle. The actual acquisition can be mapped to the discrete labels of working and idle by the state bit of the controller, the start and stop signal or the task occupation flag. The monitoring unit performs debouncing and stability confirmation on the label sequence to ensure that short-time jitter does not misjudge the state switching. The instantaneous power consumption parameter is the real-time detected electric power value, which is continuously returned by the entrance side power measurement module. The monitoring unit performs effectiveness verification and outlier rejection on it, and establishes a consistency comparison channel with the state label.

[0038] According to the collection results, two types of event records are generated in the monitoring unit, and the logic of load demand events and energy recycling feasible events is realized by edge trigger rules: when the device state at the collection point is switched from idle to working and the switching is consistent within the stability confirmation window, the monitoring unit generates a load demand event in real time, and writes the event category, timestamp and corresponding device number into the event record; when the device state at the collection point is switched from working to idle and the instantaneous power consumption parameter is lower than the preset power threshold, an energy recycling feasible event is generated after stability confirmation, and the same fields are recorded. In order to avoid the time alignment error of state switching and power reading, the monitoring unit performs alignment processing under the same time base, and rearranges the late or out-of-order data according to the timestamp before event determination.

[0039] To ensure the reliability of events, the monitoring unit calls the safety interlocking condition for quick pre-check before generating event records, skips event solidification for devices in maintenance, emergency stop or limited mode, and only keeps monitoring tracks without triggering subsequent scheduling. For continuous occurrence of similar events, the monitoring unit generates a duplicate prevention identifier according to the timestamp and device number, avoiding the same state switching from being written multiple times. Finally, the load demand events and energy recovery feasible events enter the event trigger table in chronological order, and are archived together with the reference identifier of the preset power threshold and the source of the collection point, providing accurate and traceable data basis for subsequent selection of executable energy recovery windows in the recovery channel and allocation of available time slots that do not conflict with recovery in the redistribution channel.

[0040] Further, under the premise that the event trigger table has been generated and contains event categories, timestamps and device numbers, the channel division, pairing retrieval and shadow task block generation are completed according to the following process and rules, and the recovery channel queue, the redistribution channel queue, the initial pairing set and the shadow task block set are output, and the corresponding time slot occupation marks are completed in the task redistribution candidate set.

[0041] With the event trigger table as input, records are written into the recovery channel queue and the redistribution channel queue according to the event category field, where energy recovery feasible events are only written into the recovery channel queue, and load demand events are only written into the redistribution channel queue; the two queues are respectively sorted in chronological order, and when there are the same timestamps, the order is determined according to the ascending order of device numbers; read-only cursors are generated for the two queues respectively, recording the current processing position and the processed boundary, to ensure that subsequent retrieval is carried out in a one-way advancing and non-backtracking manner. Without changing the original content of the events, time indexes are established for the recovery channel queue and the redistribution channel queue respectively, for quick positioning of events according to time range; a device index is established for the redistribution channel queue, for screening load demand events within a specified device range; both indexes generate unique index identifiers and are associated with the original event row through the event unique identifier in both directions, to ensure that subsequent pairing retrieval and conflict checking can be directly located to the original event row.

[0042] The matching relationship in the same time period is defined as follows: taking the timestamp of the target energy recovery feasible event as a reference, find the load demand event that meets the condition within the preset time matching window; when the boundary of the time matching window crosses the production beat boundary, preferentially match within the current beat, and events that are not matched within the current beat are delayed to the next beat; when multiple load demand events are within the time matching window, sort the candidates according to the order of device idle first-come-first-served, event timestamp closer first, and smaller device number first. The above-mentioned time matching window and beat boundary are set by system parameters and are not modified in this step. From the current position of the read-only cursor of the self-recovery channel queue, take out the energy recovery feasible events one by one, call the time index to search for the load demand events in the reassignment channel queue within the defined time matching window, and call the device index to exclude the devices in the safety interlock prohibition list and the devices in the maintenance shutdown state; for each candidate load demand event retrieved, generate a matching item, which contains the matching identifier, the energy recovery feasible event identifier, the load demand event identifier, the time relationship label, the device feasibility label, and the conflict prediction label; when a certain energy recovery feasible event does not retrieve any candidate load demand event, generate a placeholder matching item and mark it as to be re-searched; write all matching items into the initial matching set according to the candidate sorting rule, and keep the matching items of the same energy recovery feasible event adjacent in the set, which is convenient for subsequent batch locking and rollback.

[0043] From the current position of the read-only cursor of the redistribution channel queue, take out the load demand events one by one, generate a shadow task block for each load demand event in the task redistribution candidate set, and the shadow task block contains the task identification, the source device number, the target device candidate list, the required time slot length label, the switching cost label and the safety interlock verification label; immediately after the creation of the shadow task block, an occupation mark is applied on the available time slot of the device corresponding to the task redistribution candidate set, the occupation mark contains two time points of occupation start time and occupation end time, and the state of "only occupying and not triggering actual scheduling" is recorded; when the target device candidate list is empty or all in the safety interlock prohibited state, the shadow task block is still created, but the occupation mark is empty and marked as waiting, and the shadow task block in the waiting state is summarized at the end of the step to be processed subsequently. A consistency check is performed on the initial pairing set and the shadow task block set: when there are multiple shadow task blocks in the same time matching window proposing occupation requests for the same time slot of the same device, only the occupation mark of the shadow task block with the highest ranking is retained, and the remaining shadow task blocks are marked as waiting; when a load demand event corresponding to a pairing item is in the waiting state and has no available time slot, the conflict prediction label of the pairing item is set to high risk, and the order of the pairing item in the initial pairing set is moved backward; when it is found that all pairing items corresponding to a certain energy recovery feasible event are high risk, a secondary search label is added to the energy recovery feasible event record for subsequent iteration to expand the search range.

[0044] When the event trigger table adds a new record in the current processing batch, the processing flow of the step is not interrupted, and the new record only enters the channel division in the next batch; when the time stamp is detected in reverse order, the event category is missing or the device number is invalid, the event record is moved to the exception queue and does not participate in processing in the current batch; at the end of the step, the recycling channel queue, the redistribution channel queue, the initial pairing set and the shadow task block set are persisted in the form of read-only snapshots, and the positions of the two read-only cursors and the occupation state of the task redistribution candidate set are saved synchronously, which are used as direct inputs for the matching and locking stages. Through the above process, the clear division of the event trigger table to the recycling channel queue and the redistribution channel queue is completed, the load demand event search within the limited time matching window and the construction of the initial pairing set are realized, and the time slot resources of the task redistribution candidate set are occupied by the shadow task block without triggering the actual scheduling, which provides a complete and consistent initialization basis for the subsequent matching and locking, alternating promotion and conflict resolution.

[0045] Further, the specific implementation process of the matching and locking stage is as follows: under the premise that the recycling channel queue, the redistribution channel queue, the initial matching set, the shadow task block set, the energy recycling window set and the task redistribution candidate set have been generated and passed through consistency check, the locking and placeholder are completed in the recycling channel and the redistribution channel respectively according to the fixed processing order and the atomized state change rule, the locked energy recycling window list and the settled shadow task block list are output, and the shadow task blocks that cannot be settled are marked with a waiting mark. The implementation steps are as follows.

[0046] Read all matching items in the initial matching set and all shadow task blocks in the shadow task block set, and respectively copy them as the input snapshot of this stage to ensure that this stage is read-only externally; a stage occupation view is established for the energy recycling window set and the task redistribution candidate set, the occupation view is bidirectionally associated with the real resource state, and is used to record the locking mark and the occupation mark in this stage, and does not directly change the real resource state. In the recycling channel, the matching items in the input snapshot of this stage are sorted according to the predetermined priority rule, and the priority rule is: time relationship label priority, device feasibility label priority, candidate sorting sequence priority; the sorted matching items are sequentially subjected to validity screening, and the screening condition is: the energy recycling window is in the open state, the window time range does not overlap with the locked window, the load demand event is not soft deleted, and the corresponding device is not in the maintenance shutdown state; the matching items that do not meet the screening condition are marked as invalid in this stage and moved to the end of the queue, and do not participate in the locking in this stage.

[0047] Starting from the highest priority valid matching item, the corresponding energy recycling window is located, and the occupation state of the window in the occupation view is checked. When the occupation state is idle, a locking mark is written for the window, the locking mark includes the locking start time, the locking end time and the matching identification, and the associated mark is written on the load demand event at the same time; when the occupation state is locked or occupied, the priority of the locking source is compared. If the priority of the matching item is higher, the original locking is executed to be revoked and a rollback mark is written, and then the locking of the matching item is completed; if the priority of the matching item is not higher than that of the original locking, the matching item is marked as conflict reservation, and the reservation state enters the next round of processing. After the locking of one matching item is completed, the energy recycling window in the occupation view is immediately marked as not being occupied by other events. When multiple matching items are continuously encountered and marked as conflict reservation due to window overlap, boundary processing is triggered: the overlapping windows are adjacent to each other, or adjacent to each other, and the window segments aligned with the beat boundary are preferentially generated, and only the segments that have not been locked in this stage are continuously tried to be locked; after a batch of locking operations are completed, the newly written locking marks of this batch are added to the to-be-committed list, but are not immediately written into the real resource state, and are waiting for the placeholder results of the redistribution channel to be submitted together.

[0048] In the redistribution channel, the shadow task blocks in the input snapshot of this stage are sorted by the switching cost label from low to high, and when the switching cost label is the same, the task identification is sorted in ascending order. The target device candidate list of each shadow task block is read, and the task redistribution candidate set is expanded into a candidate time slot sequence sorted by time according to the available time slot table. The candidate time slot sequence only contains time slots that pass the safety interlocking check label and the required time slot length label check. Starting from the first shadow task block after sorting, the candidate time slot sequence is traversed in turn, and for each candidate time slot, it is checked whether there is a time overlap with the locked energy recovery window in the occupation view. If there is no time overlap and the time slot is not occupied by other shadow task blocks, an occupation mark is written in the occupation view for the shadow task block, and the occupation mark contains the occupation start time, the occupation end time, the source device number and the target device number. If there is a time overlap or the time slot is already occupied, the time slot is skipped and the traversal continues. When the candidate time slot sequence is traversed and no available time slot is found, the next machine is selected according to the order of the target device candidate list, the candidate time slot sequence is regenerated, and the above matching and placeholder process is repeated. When all machines are traversed and no available time slot is found, the shadow task block is marked as waiting, and the waiting reason and the last conflict opponent are recorded.

[0049] The earliest feasible time slot is used in the candidate time slot sequence, which is defined as the time slot with the earliest start time and no overlap with the locked energy recovery window without violating the safety interlocking, required time slot length and device switching order. When there are multiple candidate time slots with the same start time, the machine with higher idle rate is selected, and when the machine idle rate is the same, the device number with smaller number is selected. A bidirectional consistency check is performed on the resources with written lock marks and occupation marks. When the occupation mark of a shadow task block and the lock mark of an energy recovery window appear a new time overlap, a rollback is performed according to the established priority rule, which is recovery first and redistribution placeholder second. When performing the rollback, only the latest mark on the lower priority side is removed, and the rollback reason is written into the rollback mark. When the rollback results in no available time slot for the shadow task block, the shadow task block is transferred to the waiting state.

[0050] At the end of this phase, the lock marks and the occupancy marks in the to-be-committed list are packaged in time sequence into a commit batch, and an atomic commit is performed: if the commit succeeds, the corresponding marks in the occupancy view are synchronously written to the real resource state, and the energy recovery window state is updated to be locked, and the task reassignment candidate set corresponding time slot state is updated to be occupied; if the commit fails, all newly added marks in this batch are rolled back, and the sorted pairs and shadow task blocks are kept for the next round of processing. The locked energy recovery windows are sorted into a locked energy recovery window list, the settled shadow task blocks are sorted into a settled shadow task block list, and the shadow task blocks marked as waiting are sorted into a waiting list; the three types of lists are persisted together with the rollback marks and conflict reservation records as direct inputs for the subsequent alternating advancement phase and conflict resolution phase.

[0051] Further, the preset priority rules are: first, sorting according to the time relationship label; when the time relationship label is the same, sorting according to the device feasibility label; when both are the same, finally sorting according to the candidate order; and performing validity screening on the sorted pairs one by one, the screening condition being: the energy recovery window is in an open state, the window time range does not overlap with the locked window, the load demand event is not marked as invalid, and the corresponding device is not in a maintenance downtime state; the pair that does not meet the screening condition is marked as invalid in this phase and moved to the end of the queue, and does not participate in the locking in this phase.

[0052] Further, the specific implementation process of the 2.3 alternating advancement phase is: on the premise that the locked energy recovery window list, the settled shadow task block list and the waiting state shadow task block list have been obtained, the recovery channel and the reassignment channel are advanced in the occupancy view in rounds, following the iteration rule of “recovery channel first, reassignment channel fills in the gaps”, which specifically includes the following steps.

[0053] A round identifier is generated, a round occupancy view is generated by copying from the real resource state, the locked energy recovery window list and the waiting state shadow task block list input in this round are loaded, and the to-be-committed list is emptied; by comparing the real resource state after the last round of submission with the occupancy view in this round, the idle sections generated due to rollback, cancellation or completion of release are summarized to form a set of unlocked time slots in this round. The energy recovery windows involved in this round are reviewed one by one in time sequence and priority order, and the review content includes consistency with safety interlocking conditions, time non-overlapping with existing effective windows, and window boundary validity; the review passers are recorded in the valid recovery window set in this round, and their time ranges are occupied in the occupancy view with lock marks; the non-passers cancel their lock in this round, record the cancellation reason and rewrite to the pair set for reordering processing in the next round.

[0054] For the waiting state shadow task block, according to the earliest feasible time slot priority strategy, the available time slot is preferentially searched in the current round unlocking time slot set; when the earliest feasible time slot overlaps in time with the current round effective recovery window, the time slot is skipped and the search is continued; when the current device has no available time slot, the candidate time slot sequence of the next device is expanded in the order of the target device candidate list and the search is repeated; once a time slot that does not overlap with the current round effective recovery window and meets the required time slot length and safety interlocking condition is found, an occupation mark is written in the occupation view, and the occupation mark is added to the to-be-submitted list; if no available time slot is found after traversing all candidates, the shadow task block is maintained in the waiting state and the source of the last conflict is recorded. The number of newly added effective recovery windows in the current round and the number of recovery window attempts in the current round are counted, and the locking ratio in the current round is calculated; when the locking ratio is lower than the preset ratio, the preset ratio is set by the system parameter, and the round rollback is performed: the to-be-submitted list in the current round is emptied, the occupation view is restored to the state at the initialization time of the current round, and the previous ordering of the waiting state shadow task block list and the pairing set in the current round is restored, and the next iteration is directly entered; when the locking ratio reaches or is higher than the preset ratio, the newly added locking marks and occupation marks in the current round occupation view are retained, and the conflict resolution phase is entered. The current round effective recovery window set, the to-be-submitted list, and the shadow task block list still in the waiting state are packaged as the current round output and provided to the conflict resolution phase for unified resolution and submission preparation.

[0055] The specific implementation process of the conflict resolution phase is: without changing the external real state of the current round occupation view, two-level resolution is performed on all allocated energy recovery windows and task time slots, and the effective submission batch and rollback list are output, which specifically includes the following steps.

[0056] First-level resolution: safety interlocking condition screening: the occupation marks in the current round effective recovery window set and the to-be-submitted list are checked one by one, and those that are inconsistent with the safety interlocking condition or that have cross-channel hard conflicts with the already effective occupation are directly deleted and the deletion reason is recorded; the energy recovery window is written to the pairing set and marked as to be re-searched, and the shadow task block is written to the waiting state shadow task block list and the candidate device order is preserved.

[0057] First-level resolution: cross-channel time overlap processing: the recovery window locking mark and the task time slot occupation mark still retained in the current round occupation view are checked for cross-channel time overlap; when overlap is found, the recovery window locking mark is retained and the corresponding task time slot occupation mark is revoked according to the fixed rule of recovery channel priority and redistribution channel position compensation, and the shadow task block after revocation is written to the waiting state.

[0058] Second level resolution: multi-task competition in the same time slot: on the timeline of each device, a competition group is established for multiple shadow task blocks pointing to the same time slot or partially overlapping with each other; a uniform sorting rule is applied to each competition group, and the sorting priority is: higher priority pair priority, lower switching cost priority, higher device idle rate priority, earlier starting time priority, and smaller device number priority; the front of the sorting retains its occupation mark, and the back of the sorting cancels its occupation mark and writes back to the waiting state, while recording the backtracking reason and backtracking time range.

[0059] Second level resolution: competition processing within the recovery window: when there are multiple recovery lock marks pointing to overlapping windows within the same time range, a pre-set priority rule is used for selection, and the priority rule is time relationship label priority, device feasibility label priority, and candidate sorting order priority; the selection result is to retain the lock mark of the priority, cancel the lock mark of the rest, and write the corresponding pair to the pairing set for the next round.

[0060] After completing the two-level resolution, a consistency review is performed for all retained lock marks and occupation marks, including consistency with safety interlocking conditions, non-overlapping across channels, and non-overlapping on the same side; the marks that pass the review are aggregated into a submission batch, and the marks that fail the review are backed up to the waiting or pairing set according to the corresponding reason. The submission batch is written to the real resource state in a one-time all-success or all-failure manner, and when the submission is successful, the energy recovery window state is updated to locked and the corresponding time slot state of the task reassignment candidate set is updated to occupied, and the effective time is recorded; when the submission fails, all new marks in this batch are revoked, and the sorting results and backtracking list of this round are retained to enter the next round of iteration. The recovery window and task time slot list of this round that successfully submitted, the backtracking list of the submission failure, and the waiting state shadow task block list are output; when the waiting state shadow task block and the available pairing item still exist, return to the alternate advancement stage to enter the next round, otherwise end this stage.

[0061] Further, in step 2, the process of final solidification into recovery execution instructions and task reassignment instructions includes: converting conflict-free energy recovery windows into recovery execution instructions, converting shadow task blocks into actual reassignment instructions, and recording a set of unmatched events for priority processing in the next cycle.

[0062] Further, the specific process of step 3 is as follows: the monitoring unit routes the recovered execution instructions and task reassignment instructions obtained in step 2 to the energy recovery execution unit and the task reassignment scheduler respectively, and completes the reception confirmation and readiness confirmation in the specified order during the delivery process. Only when the two types of confirmations return valid results, the placeholder reservation is established on the corresponding device to freeze the energy recovery window and available time slot occupied by the recovered execution instructions and task reassignment instructions, and to prevent other temporary operations from interfering with the same time range and device. In the implementation stage, the monitoring unit sets execution monitoring points on the recovery side and the reassignment side at the same time, continuously collects the actual energy recovery amount and the task completion status, generates the recovery execution trajectory and the task execution trajectory, and performs consistency comparison between the two types of trajectories and the target start time, target end time and device constraints recorded in the recovery execution instructions and task reassignment instructions. The comparison logic outputs a deviation classification label, which clearly labels the situations of insufficient recovery, early termination, non-execution, delayed completion or interruption, and attaches the reference identifier of the affected device, time range and source instruction.

[0063] After obtaining the deviation classification label, the monitoring unit triggers the dynamic adjustment and compensation scheduling in real time: in the energy recovery window set, the standby window replacement is preferentially performed. If the standby window of the same device cannot meet the recovery target, the same type of standby device is used to compensate back under the premise of not violating the safety interlocking condition. When the gap is still not covered, segmented splicing recovery is used to maximize the preservation of existing recovery results; in the task reassignment candidate set, the goal is to have less impact on the production rhythm. First, try to move the same device to avoid the locked energy recovery window and placeholder reservation. If there is still overlap after moving, perform cross-device handover, migrate the task to an available time slot that does not conflict with recovery, and use segmented execution to split the occupation and reduce the impact on critical periods when migration is not feasible. For the newly generated instructions formed by the above dynamic adjustment, the monitoring unit and the existing placeholder reservation jointly participate in the safety interlocking review, and quickly resolve the combined execution conflicts involving multiple devices and cross-channels; cross-channel conflicts are handled according to the rules of recovery priority and reassignment compensation, same side conflicts are handled according to the rules of earlier start time priority and higher device idle rate priority, and if necessary, the low priority occupation is rolled back and the rollback reason is recorded for subsequent round reference.

[0064] To ensure state consistency and traceability, the monitoring unit adopts a commit mechanism that takes effect all at once in an all-success or all-failure manner for all instructions that pass the review, simultaneously writes the final form of the two types of instructions to the real state of the execution side and the scheduling side at the moment of commitment, and synchronously updates the reference baseline of the recycling execution track and the task execution track; if the state of any sub-instruction changes or violates the device constraints before commitment, it is determined that the commitment fails, the overall rollback is performed, and the sorting result is retained for the next round of adjustment to avoid the appearance of a partially effective intermediate state. In one execution cycle, the monitoring unit continuously monitors the deviation classification label and the conflict state, and if there is no instruction to be retried and no unresolved conflict, it is determined to be convergent, and the execution result, adjustment record and deviation classification label are immediately written back to the event trigger table, energy recycling window set and task reallocation candidate set, so that these basic data continue to serve as the basis for sorting and screening in subsequent cycles, and unfinished matters are moved to the priority queue of the next cycle to ensure that they are processed preferentially in the new time range.

[0065] Further, the production line includes production branch A, production branch B and production branch C, and the devices include injection molding machine M1, punching machine P1, same model robots R1 and R2, and conveying unit C1, all of which are connected to the same DC bus and connected to the energy recycling execution unit. The monitoring unit continuously runs at a sampling period of minutes in a continuous running time interval minutes. The device running state parameter is defined as , where is the device identifier, is the time, takes "working" or "idle"; the instantaneous power consumption parameter is defined as , which represents the electrical power value of the device at time ; the preset power threshold is defined as , which is used to identify energy recycling feasible events, and in this example . When it is observed that the device switches from working to idle and continues to be stable, the monitoring unit generates an energy recycling feasible event and solidifies the energy recycling window; when it is observed that the device switches from idle to working, the monitoring unit generates a load demand event and solidifies the default plan of the corresponding task. To express subsequent calculations, the energy recycling window is defined as , where is the window number, and are the start and end times of the window, respectively; the default device of the task is defined as , the default start time is , the default end time is , and the approximately constant task power is In minutes, via the determination of the device running state parameters and the instantaneous power consumption parameters, the monitoring unit obtains the following energy recovery windows and default tasks in the event trigger table: M1 in switches from working to idle and remains until , forming ; P1 in switches from working to idle and remains until , forming ; R1 in switches from working to idle and remains until , forming ; R2 in switches from working to idle and remains until , forming ; M1 in switches from working to idle again and remains until , forming the standby window . The default tasks of the load demand event solidification include: in M1 execution, , ; in R1 execution, , ; in P1 execution, , ; in R2 execution, , ; in M1 execution, , . The historical statistics of the energy recovery execution side give the average recoverable power of each window in the interval , the actual recovery power trajectory is defined as , the theoretical recoverable energy in the window is defined as , and there are ; wherein the time is in minutes, and when converted to hours, it needs to be divided by . In this example, , , , , , and the units are all .

[0066] In the dual-channel cooperative scheduling mechanism, the monitoring unit first constructs the candidate set of the recovery channel and the redistribution channel and performs preset priority rule sorting. To form the initial pairing set of the recovery channel, the time matching window length is defined as , which represents the allowed time proximity range between the energy recovery window and the load demand event, and in this example, minutes. For each pair of candidate computes time relationship tags with device feasibility tags , and keeps candidate ordering sequence . Define ; where is the prerequisite for inclusion in pairing. The default priority rule is three-key lexicographic ordering, i.e., first by ascending, then by ascending, and finally by ascending. In this example, the valid pairings available include with , ; with , ; with , ; with , ; with , . The monitoring unit accordingly attempts to lock , , , , in sequence in the recycling channel. The safety interlock condition check is performed simultaneously in the locking process, in this example, the DC bus is allowed to have at most energy recycling windows concurrently at any time slice, and the concurrency of R1 and R2 in the same protection zone is limited to . After locking and , the concurrency upper limit has been reached within , so overlaps with the former two, and the first locking decision is invalid due to not meeting the safety interlock condition. At this time, the non-conflicting can be locked first, and will be re-evaluated in the next iteration after . To quantify the effectiveness of the recycling channel, the recycling channel locking ratio is defined as ; and the threshold is set as the rollback condition, which is in this example. The total duration of all feasible windows is calculated to be minutes, and the first round of locking , , , , , ,

[0067] In the reassignment channel, the monitoring unit generates a shadow task block for each load demand event and attempts to assign available time slots that do not conflict with energy recovery. The conflict criterion is that "the same device cannot be in both energy recovery window and in operation at the same time slice". After the first round of locking, the conflict cases are: with overlapping on R1 at ; with overlapping on P1 at ; with any recovery execution. To make the decision, a switching cost is defined to choose among "same device shift", "cross device handover", and "segment execution". The switching cost is defined as ; where is the extra preparation time (minute) due to device switching or sequence adjustment, is the extra auxiliary energy consumption ( ), is the delay (minute) from the relative default end time, is the weight coefficient to map the three items to a unified measure, which is taken as , , in this example. For , two candidates are constructed: same device shift to at , , , resulting in ; cross device handover to R2 and execute after end (re-tooling and communication check), , , , resulting in .

[0068] The monitoring unit thus chooses same device shift . For , two candidates are constructed: same device shift to at , , , resulting in ; cross device handover to M1 and execute after end (tooling and program switching), , , , resulting in . The monitoring unit thus chooses same device shift After this allocation, all shadow task blocks are located in available time slots that do not conflict with garbage collection.

[0069] To perform a two-stage resolution, combinations that violate the concurrency limit are first eliminated based on safety interlock conditions. During the time period, if permitted and Concurrency is allowed, but third-window concurrency is prohibited. It remains invalid within this section, but No longer restricted, the monitoring unit will be in the second iteration Reintegrate and lock its entire interval, because it is related to the already allocated interval. Located on different devices and without conflict. The second phase addresses multi-task contention within the same time slot, defining the device idle rate. For devices within the decision-making window The proportion of time not occupied by tasks, in this example Calculate R1 (because (Move back to fill), for R2 in calculate If competition occurs, the retention factor is determined by comparing pairing priority, switching cost, and device idle rate. In this example, there is no multi-task competition within the same time slot. At the end of the round, the theoretical recovery amount of the recovery channel is calculated and substituted into... have:

[0070] Unit is Total theoretical recovery in this round .

[0071] If no redistribution is carried out, then right Causes obstruction common Minutes, loss ,and right Causes obstruction common Minutes, loss The two combined By prioritizing the recovery channel and supplementing it with the redistribution channel, the aforementioned losses were avoided in this round, and the total switching cost was [missing information]. Within the preset scheduling cost budget.

[0072] Upon entering the solidification phase, the monitoring unit generates recovery execution instructions and task reallocation instructions, clarifying the energy recovery window. The system includes equipment identification, available time slots for each task that do not conflict with energy recovery, and the execution equipment. Two types of instructions are routed to the energy recovery execution unit and the task redistribution scheduler, respectively. Upon issuance, receipt confirmation and readiness confirmation are completed sequentially, and a placeholder reservation is established on the corresponding equipment. During execution, monitoring points are set up on both the recovery and redistribution sides to generate real-time recovery and task execution trajectories. These trajectories are then compared for consistency with the target start time, target end time, and equipment constraints, and deviation classification labels are output. In this example, two types of deviations were observed: firstly... exist Early termination, continued If no energy is recovered within minutes, the actual recovered energy is defined as... Recovery deviation is , can be obtained Secondly, During the execution period of R1 A brief material blockage caused a delay in completion; the task delay is defined as follows: For each minute, the monitoring unit labels the task as "delayed completion" to reflect the deviation. When any of the following occurs: insufficient energy recovery, early termination, non-execution, delayed completion, or interruption, the monitoring unit triggers dynamic adjustment and compensation scheduling. In the energy recovery window set, a backup window is first searched on the same device P1. P1 is idle and ,form ,estimate To compensate The required time is Therefore, in Internal segmented assembly and recycling The "alternate window replacement" can be completed in minutes. If unavailable, it can be replaced with a spare device of the same model or reclaimed by segmenting and splicing across devices but on the same busbar. In this example, activation is not required. For the task side, because... Minutes later, the monitoring unit prioritizes moving back to the same equipment to avoid intruding on any energy recovery window, constructing candidate... The extension and calculation of switching costs, with the additional switching cost shifted later being... If this extension conflicts with subsequent tasks, then cross-device handover or segmented execution will be performed within the task reassignment candidate set; in this example... Not with Overlapping and not with Competing on R2, because It is still on R1, so there is no need for cross-device handover or segmented execution.

[0073] All newly generated commands and existing placeholders must undergo safety interlock verification and rapid conflict resolution. Cross-channel conflicts are handled according to the rule of priority for reclamation and reallocation, while same-side conflicts are handled according to the rule of priority for earlier start time and higher device idle rate. In this example, P1 was found to be in... There is a short-term safety interlock condition window that prohibits entry to maintenance workstations. Minutes later, the monitoring unit will send the compensation command from Adjusted to and Two sections and in This minute contains an empty space, which is equivalent to... implement Minutes of recycling and implement Minutes of data recovery and security interlock verification passed. The monitoring unit then submits the data using a one-time submission mechanism that takes effect either all successes or all failures, defining the submission set. This includes two types of instructions: compensation recovery and task extension. If any sub-instruction's device status or safety interlock condition is not met at the moment of submission, the submission is deemed a failure and the entire process is rolled back. In this example, the first submission failed because the P1 maintenance signal was not cleared, and the monitoring unit rolled back the entire process. After a few seconds (less than the sampling period), the interlock signal is checked again and resubmitted. The second submission is successful, and the real status of the execution side and the scheduling side is written simultaneously.

[0074] Calculate the actual total recovered energy at the end of the execution cycle. (Excluding...) Early termination and by Aside from compensation, the remaining windows will proceed as planned, resulting in:

[0075] Consistent with the theoretical total recovery. (Task side) and Complete according to the adjusted time slot. Due to the delay in completion It extends but does not intersect with the energy recovery window. and Completed according to the default plan. At the end of the cycle, the monitoring unit checks for any retry instructions and unresolved conflicts. If none are found, convergence is determined, and the execution results, adjustment records, and deviation classification labels are written back to the event trigger table, energy recovery window set, and task reallocation candidate set. Simultaneously, unfinished items are moved to the priority queue for the next cycle. In this example, convergence was achieved, and the written-back content includes... "Early termination" deviation classification label, Compensation instruction reference identifier, The "delay completion" deviation classification label and the extended interval reference mark are used as the basis for the event trigger table to keep consistent with the real execution on the time axis and the device axis. So far, this example fully demonstrates the full-process technical solution of the double-channel cooperative scheduling mechanism constructed in the monitoring unit, the selection of the executable energy recovery window in the recycling channel according to the preset priority rule, the allocation of the available time slot for the task in the redistribution channel without conflict with the recycling, the dynamic adjustment through iteration and conflict resolution, and the final solidification into the recycling execution instruction and the task redistribution instruction, the triggering of dynamic adjustment and compensation scheduling according to the deviation classification label during the execution process, the atomic commit and necessary overall rollback under the one-time effective commit mechanism, and the write-back and moving into the next cycle priority queue after convergence. The switching cost is used as a unified measure, which is calculated in specific numerical values in each candidate allocation and used as one of the criteria for supplementing the redistribution channel, ensuring the optimization of energy recovery efficiency and production rhythm stability.

[0076] Figure 2 The experimental curve graph showing the task redistribution switching cost optimization effect of the present application quantitatively analyzes the optimization effect of the double-channel cooperative scheduling mechanism on the switching cost of the shadow task block. In the coordinate system, the horizontal axis represents the number of shadow task blocks, which increases from 5 to 50, and the vertical axis represents the switching cost (unit time), reflecting the time overhead in the task redistribution process. The graph contains two comparison curves: the solid line represents the switching cost before optimization, and the dashed line represents the switching cost after optimization. The curve before optimization shows a clear upward trend. When the number of shadow task blocks increases from 5 to 50, the switching cost increases sharply from 9.8 unit time to 1.5 unit time, indicating that the traditional scheduling method significantly decreases in efficiency when handling a large number of tasks. This is due to the lack of effective time slot allocation strategy, resulting in frequent task switching and waste of idle device time. In contrast, the optimized curve shows significant improvement. By implementing the shadow task block switching cost lowest priority allocation strategy, the switching cost remains stable at 5.0-5.6 unit time. Even in the case of handling 50 shadow task blocks, the switching cost is only 5.6 unit time, which is 73.3% lower than the peak value before optimization. This optimization effect is mainly due to the intelligent scheduling mechanism of the redistribution channel in the present application: the system preferentially selects the shadow task block with the lowest switching cost, moves it into the earliest feasible time slot of the idle device, and through the sequential selection algorithm in the candidate set, ensures that the time slot allocation does not conflict with the locked energy recovery window. The experimental results show that the double-channel cooperative scheduling mechanism effectively reduces the overall switching overhead of task redistribution through reasonable priority rules and conflict resolution strategies, improves the operation efficiency and rhythm stability of the automated production line, and verifies the practicality and advancement of the technical solution of the present application.

[0077] Figure 3The experimental data graph showing the relationship between the power consumption of the device and the preset power threshold in the application clearly reflects the identification mechanism of the energy recovery possible event and the load demand event. In the coordinate system, the horizontal axis represents time (minutes), and the vertical axis represents the instantaneous power consumption (W). Through 30 minutes of continuous monitoring data, the dynamic change rule of the device operation state parameter is shown. The preset power threshold in the graph is 20W, which is marked by a dashed line. This threshold is a key parameter for triggering the energy recovery possible event. When the instantaneous power consumption of the device is lower than the preset power threshold, the system determines that it is an energy recovery possible event, which is marked by a solid dot in the graph. The experimental data shows that at 5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes and 30 minutes, the device power drops to 12W, 15W, 8W, 12W, 15W and 8W, respectively, which are all lower than the preset power threshold of 20W. Therefore, these time points are identified as energy recovery possible events. At the same time, the device power curve shows periodic fluctuation characteristics. The time when the device power state switches from low power to high power is identified as a load demand event, which is marked by a solid square in the graph. Specifically, at 7.5 minutes, 12.5 minutes, 17.5 minutes, 22.5 minutes and 27.5 minutes, the device power jumps to 85W, 82W, 90W, 85W and 82W, respectively, indicating that the device operation state switches from idle to working, triggering the load demand event record. The experimental data graph verifies the effectiveness of the event trigger table generation mechanism of the application. Through real-time monitoring of the instantaneous power consumption parameter and comparison with the preset power threshold, the two key events can be accurately identified and classified, providing a reliable data basis for subsequent double-channel collaborative scheduling.

[0078] Although the specific embodiments of the application are described above, those skilled in the art should understand that these specific embodiments are only illustrative, and those skilled in the art can make various omissions, substitutions and changes to the details of the above method and system without departing from the principles and essence of the application. For example, combining the above method steps, performing substantially the same function in substantially the same manner to achieve substantially the same result is within the scope of the application. Therefore, the scope of the application is only limited by the appended claims.

Claims

1. An energy-saving control method for an automated production line that combines energy recovery and task redistribution, characterized in that, The method includes: Step 1: In the automated production line, collect equipment operating status parameters and real-time power consumption parameters at the entrance of each production branch; based on the collected results, generate two types of event records in the monitoring unit, namely load demand events and energy recovery feasibility events; store all event records in chronological order to form an event trigger table, and save the safety interlock conditions in the monitoring unit as subsequent scheduling constraints; label each record with event category, timestamp and corresponding equipment number; Step 2: Based on the generated event trigger table, establish a dual-channel collaborative scheduling mechanism for both energy recovery and task reallocation. According to preset priority rules, select and lock executable energy recovery windows in the recovery channel, and allocate available time slots that do not conflict with recovery to tasks in the reallocation channel. Through iterative advancement and conflict resolution, dynamically adjust the window and time slot allocation during multiple rounds of scheduling to ensure that energy recovery efficiency and production cycle stability are optimized simultaneously, and finally solidify them into recovery execution instructions and task reallocation instructions. Step 3: Issue the generated energy recovery execution command and task redistribution command to the corresponding execution units for implementation. During execution, collect the actual energy recovery amount and task completion status, compare them with the command target, and dynamically adjust and compensate the scheduling based on the difference results. In Step 2, through iterative advancement and conflict resolution, the window and time slot allocation are dynamically adjusted in multiple rounds of scheduling to ensure that energy recovery efficiency and production cycle stability are optimized simultaneously. This process includes: adopting an iterative rule of prioritizing recovery channels and filling in gaps with redistribution channels. In each round of iteration, first confirm all windows locked by the recovery channels in this round, and then insert the waiting shadow task blocks into the time slot positions unlocked in this round. When the number of windows locked by the recovery channels in a certain round of iteration is less than the preset proportion, the preset proportion is set by system parameters, roll back all operations in this round, and re-sort from the pairing set to execute the next round of matching. For all allocated windows... The energy recovery window and task time slots undergo a two-stage resolution process. The first stage removes pairings that violate safety interlock conditions, and the second stage resolves the issue of multiple tasks competing in the same time slot. The competition result is determined in the order of pairing priority, switching cost, and equipment idle rate, with the lowest priority pairing rolling back to the next iteration. In step 2, the process of establishing a dual-channel collaborative scheduling mechanism for energy recovery and task reallocation based on the generated event trigger table includes: dividing the event trigger table into a recovery channel queue and a reallocation channel queue according to event categories; retrieving load demand events for each feasible energy recovery event in the recovery channel queue within a preset time matching window, which is set by system parameters, to form an initial pairing set; and generating a shadow task block for each load demand event in the reallocation channel queue. The shadow task block occupies the time slot resources of the task reallocation candidate set but does not immediately trigger actual scheduling.

2. The energy-saving control method for automated production lines combining energy recovery and task redistribution as described in claim 1, characterized in that, In step 1, the equipment operating status parameters include two status indicators: working and idle; the real-time power consumption parameter is the real-time detected power value; based on the collection results, two types of event records are generated in the monitoring unit, namely load demand event and energy recovery feasibility event. The condition for triggering the load demand event record is that the equipment status at the collection point switches from idle to working, and the condition for triggering the energy recovery feasibility event record is that the equipment status at the collection point switches from working to idle and the real-time power consumption parameter is lower than the preset power threshold.

3. The energy-saving control method for automated production lines combining energy recovery and task redistribution as described in claim 2, characterized in that, The process of selecting and locking executable energy recovery windows in the recovery channel according to preset priority rules, and allocating available time slots that do not conflict with recovery to tasks in the redistribution channel, includes: in the recovery channel, starting with the highest priority pair, locking the corresponding energy recovery window and marking its time slot as unusable by other events; in the redistribution channel, starting with the shadow task block with the lowest switching cost, attempting to move it into the earliest feasible time slot of the idle device, if the time slot conflicts with the locked energy recovery window, then sequentially selecting the next device in the candidate set to rematch the time slot, until a non-conflicting time slot is found or marked as waiting.

4. The energy-saving control method for automated production lines combining energy recovery and task redistribution as described in claim 3, characterized in that, The preset priority rules are as follows: First, sort according to the time relationship label; if the time relationship labels are the same, then sort according to the equipment feasibility label; if both are the same, finally sort according to the candidate sorting order; perform validity screening on each of the sorted pairs, with the screening conditions being: the energy recovery window is open, the window time range does not overlap with the locked window, the load demand event is not marked as invalid and the corresponding equipment is not in maintenance shutdown state; pairs that do not meet the screening conditions are marked as invalid in this stage and moved to the end of the queue, and do not participate in the locking in this stage.

5. The energy-saving control method for automated production lines combining energy recovery and task redistribution as described in claim 4, characterized in that, In step 2, the final process of solidifying the energy recovery execution instructions and task redistribution instructions includes: converting conflict-free energy recovery windows into recovery execution instructions, converting shadow task blocks into actual redistribution instructions, and recording the set of unmatched events for priority processing in the next cycle.

6. The energy-saving control method for automated production lines combining energy recovery and task redistribution as described in claim 5, characterized in that, Step 3 specifically includes: routing the recovery execution command and task redistribution command solidified in Step 2 to the energy recovery execution unit and task redistribution scheduler respectively; upon issuance, sequentially completing reception confirmation and readiness confirmation, and establishing placeholder reservations on the corresponding devices; during implementation, setting execution monitoring points on the recovery side and redistribution side respectively to form recovery execution trajectory and task execution trajectory, and comparing them with the target start time, target end time, and device constraints for consistency, outputting deviation classification labels; when any situation occurs such as insufficient recovery, early termination, non-execution, delayed completion, or interruption, dynamic adjustment and compensation scheduling is triggered within the energy recovery window set. The system performs backup window replacement, replenishment by backup equipment of the same model, or segmented splicing and recycling. In the task redistribution candidate set, it performs same-device shift, cross-device handover, or segmented execution. It also performs safety interlock verification and rapid conflict resolution on newly generated instructions and existing placeholders. Cross-channel conflicts are handled according to the rule of recycling priority and redistribution replenishment. Same-side conflicts are handled according to the rule of earlier start time priority and higher equipment idle rate priority. Through a submission mechanism that takes effect all at once with all success or all failure, all verified instructions are written into the real state of the execution side and the scheduling side at once. If the submission fails, the whole process is rolled back and the sorting results are retained for the next round of adjustment.

7. The energy-saving control method for automated production lines combining energy recovery and task redistribution as described in claim 6, characterized in that, In step 3, if there are no retry instructions or unresolved conflicts within an execution cycle, it is considered convergence. The execution results, adjustment records, and deviation classification labels are written back to the event trigger table, energy recovery window set, and task reallocation candidate set. Unfinished items are moved to the priority queue of the next cycle.

Images