A patch plate welding task dynamic scheduling system based on multi-station cooperation

Abstract

The application provides a kind of based on multi-station cooperation's patch plate welding task dynamic scheduling system, it is related to industrial automation control and production scheduling technical field, it is by constructing the task station adaptation matrix of fusing equipment real-time reliability and process residual distribution, combined with the time urgency reordering of fusing predicted completion deviation and optional station scarcity, and with effective idle time length matching and time fragment suppression mode Dynamic insertion task, while introducing the disturbance suppression rescheduling mechanism based on normalized comprehensive cost comparison, in this way, dynamically balance each station load, improve equipment utilization and overall capacity, shorten the task completion cycle, and effectively reduce the additional transfer and change disturbance caused by rescheduling under abnormal working conditions, realize the adaptive efficient collaborative scheduling of patch plate welding task in multi-station dynamic environment.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control and production scheduling technology, and more specifically, to a dynamic scheduling system for patch board welding tasks based on multi-station collaboration. Background Technology

[0002] With the continuous improvement of production efficiency and welding quality requirements in heavy equipment sectors such as shipbuilding and large steel structures, patch plate welding operations are gradually evolving from manual operation to robotic automation. Modern manufacturing workshops typically deploy multiple fixed-station automated welding machines, each undertaking multiple processes such as positioning, pre-welding, main welding, and inspection, forming a flexible production line with collaborative discrete workstations. In this scenario, the scheduling of welding tasks has become a bottleneck affecting overall output cycle time and equipment utilization. Early systems relied heavily on Manufacturing Execution Systems (MES) or upper-level scheduling software to issue batch plans according to preset cycles, using static priority rules or offline optimization models to determine the allocation and processing sequence of tasks at each workstation in one go. These methods can maintain basic operation in small-batch production with simple task types and stable equipment status. However, with the increasing demand for product customization, the differences in patch plates in terms of plate thickness, material, weld geometry, and quality grade are becoming increasingly significant, resulting in a high degree of dispersion in the actual occupancy time of different tasks at each workstation; at the same time, the availability and processing capacity of the welding equipment itself are also dynamically changing due to consumable replacement, maintenance, power fluctuations, and other factors. In recent years, some studies have attempted to introduce online scheduling mechanisms to reassign tasks by collecting workstation status in real time. However, existing dynamic scheduling schemes are mostly limited to load adjustment of a single bottleneck workstation, or only use queue length and equipment busy / idle status as switching criteria. They fail to fully integrate the deep impact of welding process characteristics and equipment degradation on the processing cycle, and are also difficult to balance the urgency of tasks, the matching degree of remaining workstation capacity, and the transfer and reloading costs brought about by rescheduling in multi-workstation collaboration.

[0003] Currently published patch board welding task scheduling methods generally suffer from the following shortcomings: First, the suitability assessment of tasks and workstations is mostly based on a binary judgment of rated process parameters and equipment availability, lacking a quantitative integration of workstation real-time reliability, historical processing cycle residual distribution, and process capability edge effects. This leads to unexpected timeouts and quality risks in actual processing due to implicit equipment degradation or process capabilities approaching their limits. Second, task priority ranking is mainly based on a single dimension of delivery deadline or waiting time, without considering the scarcity of feasible workstations and the impact of workstation load pressure on the overall schedulable space. This results in tasks that appear to have multiple available workstations having their actual selection set ignored. The existing technology suffers from several shortcomings. First, the allocation of idle time slots is rather crude, often based on the earliest available time or the minimum changeover time, without assessing the reusability of remaining idle time slots after insertion. This can easily lead to fragmentation of the timeline and weaken the system's ability to handle subsequent urgent tasks. Second, rescheduling under abnormal conditions mostly involves full rollback or only partial adjustments to affected tasks, lacking a disturbance suppression strategy based on comprehensive cost comparison. This can easily lead to large-scale task migration due to small state fluctuations, increasing the hidden costs of changeover, transfer, and program switching, and disrupting the continuity of processes and the stability of workshop logistics. These shortcomings make it difficult for the existing technology to achieve dynamic coordination and continuous balance among multiple workstations when facing patch board welding tasks with continuous arrivals, significant process differences, and frequent fluctuations in workstation status. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a dynamic scheduling system for patch board welding tasks based on multi-station collaboration. This system, to a certain extent, solves the problem that existing static scheduling for patch board welding cannot dynamically balance the load based on task characteristics and real-time status of the workstations, leading to imbalanced workstation utilization and reduced production capacity.

[0005] According to one aspect of the present invention, a dynamic scheduling system for patch board welding tasks based on multi-station collaboration is provided, comprising: The module includes a task data acquisition module, an adaptation matrix construction module, a task sorting module, a workstation allocation module, and a dynamic update module. The task data acquisition module is used to acquire the queue of welding tasks to be scheduled and the real-time operation data of each workstation; The adaptation matrix construction module is used to construct a task-workstation adaptation matrix based on task processing requirements, workstation process capabilities, equipment status, and remaining workstation capacity. The task sorting module is used to determine the time urgency index based on the task workstation adaptation matrix and generate a priority-adjusted task sequence. The workstation allocation module is used to insert tasks in the task sequence into the corresponding workstation timeline. The dynamic update module is used to reallocate tasks that have not yet started execution when the workstation status changes or new tasks arrive, and to generate a dynamically updated workstation task allocation scheme.

[0006] Furthermore, the module includes a task data acquisition module, an adaptation matrix construction module, a task sorting module, a workstation allocation module, and a dynamic update module; The task data acquisition module is used to acquire the queue of welding tasks to be scheduled and the real-time operation data of each workstation; The adaptation matrix construction module is used to construct a task-workstation adaptation matrix based on task processing requirements, workstation process capabilities, equipment status, and remaining workstation capacity. The task sorting module is used to determine the time urgency index based on the task workstation adaptation matrix and generate a priority-adjusted task sequence. The workstation allocation module is used to insert tasks in the task sequence into the corresponding workstation timeline. The dynamic update module is used to reallocate tasks that have not yet started execution when the workstation status changes or new tasks arrive, and to generate a dynamically updated workstation task allocation scheme.

[0007] Furthermore, the adaptation matrix construction module includes a dynamic process adaptation unit, which is used to integrate the normalized difference between the task process requirements and the stability capability of the workstation, the availability reliability of the workstation, the remaining available processing time before the delivery deadline, and the dispersion of the processing cycle residual into a dynamic process adaptation value.

[0008] Furthermore, the dynamic process adaptation value is determined according to the following formula:

[0009] in, For dynamic process adaptation values, Marking as physically feasible. For the reliability of the workstation, For the normalized difference vector, For dynamic sensitivity matrix, For the expected duration, The remaining available processing time before the delivery deadline. For reference process time of the task, This represents the degree of dispersion of the residual during the processing cycle.

[0010] Furthermore, the task sorting module includes an urgency calculation unit, which is used to determine the contribution weight based on the dynamic process adaptation value and workstation load pressure of each feasible workstation, and calculate the fusion of the expected completion time and the scarcity index of the available workstations based on the contribution weight, and then combine the task waiting time and delivery deadline to determine the time urgency index.

[0011] Furthermore, the workstation allocation module includes an idle interval matching unit, which is used to select workstations and idle intervals from a set of candidate insertion positions that meet the conditions of physical feasibility, process connection, logistics accessibility and time axis capacity, based on dynamic process adaptation value, logistics and process connection accessibility coefficient, ratio of effective idle time to task requirement time and residual idle availability.

[0012] Furthermore, the dynamic update module includes a locked interval retention unit and an unstarted task reassignment unit; the locked interval retention unit is used to retain started tasks and locked process intervals when the workstation status changes or new tasks arrive; the unstarted task reassignment unit is used to release the original allocation intervals of unstarted tasks, and regenerate the task workstation adaptation matrix, time urgency index and workstation task allocation scheme based on the updated equipment status, workstation remaining capacity and new tasks.

[0013] According to another aspect of the present invention, a dynamic scheduling method for patch board welding tasks based on multi-station collaboration is provided, comprising: Receive the queue of patch board welding tasks to be scheduled, extract the task process feature parameters of each patch board welding task, and obtain the equipment process capability, real-time equipment status and current task backlog status of each station. Calculate the process adaptation value and expected entry time of each patch board welding task at each station, and construct the task station adaptation matrix. Based on the task station adaptation matrix, the deviation between the expected entry time, expected completion status and delivery time of each patch board welding task is quantified into a time urgency index, and the queue of patch board welding tasks to be scheduled is reordered according to the time urgency index to generate a priority-adjusted task sequence. For the priority-adjusted task sequence, each patch board welding task is sequentially searched for an insertable free interval in the candidate workstation time axis. The workstation and free interval that best match the effective free time of the workstation with the task requirement time are selected. The patch board welding task is then inserted into the corresponding workstation time axis, and the workstation task backlog and equipment occupancy time distribution are updated. When a change in equipment status, abnormal evolution of processing cycle, or arrival of a new patch board welding task is detected at any workstation, the task workstation adaptation matrix and the time urgency index are recalculated, and the workstations for patch board welding tasks that have not yet started are reallocated to generate a dynamically updated workstation task allocation scheme.

[0014] Compared with existing technologies, the patch board welding task dynamic scheduling system based on multi-station collaboration provided by this invention constructs a task station adaptation matrix that integrates the real-time reliability of the equipment and the distribution of process residuals. It combines the expected completion deviation with the time urgency of the available station scarcity and dynamically inserts tasks by matching effective idle time and suppressing time fragmentation. At the same time, it introduces a disturbance suppression rescheduling mechanism based on normalized comprehensive cost comparison. In this way, the load of each station can be dynamically balanced, improving equipment utilization and overall capacity, shortening the task completion cycle, and effectively reducing the additional transfer and replacement disturbances caused by rescheduling under abnormal working conditions. This enables adaptive and efficient collaborative scheduling of patch board welding tasks in a multi-station dynamic environment. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a system block diagram of a patch board welding task dynamic scheduling system based on multi-station collaboration according to an embodiment of the present invention.

[0016] Figure 2 This is a comparison diagram of load balancing before and after multi-station dynamic scheduling in the patch board welding task dynamic scheduling system based on multi-station collaboration according to an embodiment of the present invention. Detailed Implementation

[0017] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention; it should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0018] As mentioned in the background section, existing patch board welding task scheduling methods use preset rules or static priority queues to allocate tasks to each workstation at once upon task issuance, lacking dynamic awareness of the real-time status of each workstation. When a workstation experiences an extended processing cycle due to equipment malfunction or process fluctuations, subsequent tasks continue to wait in the original sequence and cannot be transferred to idle workstations, resulting in uneven workload. Furthermore, the differences in weld length and plate thickness among different patch boards directly affect the actual occupancy time of each workstation. Static scheduling cannot match and optimize based on task characteristics and remaining workstation capacity, causing high-load workstations to queue and low-load workstations to remain idle, extending the overall completion cycle and disrupting the production rhythm.

[0019] Figure 1This is a system block diagram of a dynamic scheduling system for patch board welding tasks based on multi-station collaboration according to an embodiment of the present invention. Figure 1 As shown, the dynamic scheduling system for patch board welding tasks based on multi-station collaboration includes: The module includes a task data acquisition module, an adaptation matrix construction module, a task sorting module, a workstation allocation module, and a dynamic update module. The task data acquisition module is used to acquire the queue of welding tasks to be scheduled and the real-time operation data of each workstation; The adaptation matrix construction module is used to construct a task-workstation adaptation matrix based on task processing requirements, workstation process capabilities, equipment status, and remaining workstation capacity. The task sorting module is used to determine the time urgency index based on the task workstation adaptation matrix and generate a priority-adjusted task sequence. The workstation allocation module is used to insert tasks in the task sequence into the corresponding workstation timeline. The dynamic update module is used to reallocate tasks that have not yet started execution when the workstation status changes or new tasks arrive, and to generate a dynamically updated workstation task allocation scheme.

[0020] Specifically, the task data acquisition module first reads the queue of patch plate welding tasks to be scheduled from the production execution system, process management system, or task issuing terminal. Each patch plate welding task in the task queue has a unique task identifier. Then, it reads the patch plate length, patch plate width, plate thickness, material, bevel type, effective weld length, number of weld segments, number of weld corners, weld curvature change, welding process type, quality grade, current pending process, task arrival time, and delivery deadline corresponding to each task. When the task belongs to a process in the continuous process of positioning, pre-welding, main welding, and inspection, the task data acquisition module simultaneously reads the completion time of the preceding process, the inter-process transfer time, the fixture release status, and the necessary process waiting time, and combines the above information to form the task process characteristic parameters of the task.

[0021] Furthermore, the task data acquisition module reads the process capability information of each workstation from the workstation controller, welding robot control system, welding power supply, fixture control unit, inspection equipment, and equipment ledger. This process capability information includes the range of processable plate thickness, the range of processable plate size, the robot's reachable working area, the types of welding processes that can be executed, the types of available fixtures, the rated welding speed curve, changeover time, positioning assistance time, welding program loading time, inspection capability, and stable processing intervals under different process modes. Subsequently, it reads the real-time equipment status of each workstation, including running status, waiting status, maintenance status, fault shutdown status, speed-limited operation status, process mode switching status, fixture occupancy status, and inspection equipment occupancy status. Simultaneously, it reads the current task backlog status of each workstation. This current task backlog status is represented by the intervals of currently executing tasks, locked tasks, assigned but not started tasks, maintenance intervals, and unavailable intervals on the workstation's timeline, rather than simply by the number of queued tasks.

[0022] Once the task process characteristic parameters and station status information are prepared, the task data acquisition module iterates through all patch board welding tasks and for each task iterates through all candidate stations. For each task-station combination, the task data acquisition module first performs a physical feasibility judgment: if the task plate thickness exceeds the stable welding thickness range of the station, or the task size exceeds the positioning range of the station fixture, or the weld trajectory exceeds the robot's reachable area, or the current process to be executed by the task is inconsistent with the station's process capability, or the welding materials, shielding gas, welding power mode, and inspection level required by the process document cannot be met by the station, then the task-station combination is marked as infeasible and the reason for infeasibility is recorded; if the above equipment operating boundaries and process boundaries are met, then the task-station combination is marked as feasible and enters the calculation process of expected occupancy time and dynamic process adaptation value.

[0023] In one implementation, the task data acquisition module calculates the expected duration and dynamic process adaptation value for feasible task workstation combinations using the following formula:

[0024]

[0025] in, Indicates the current scheduling time Next The patch board soldering task was in the first The expected time spent at each workstation, in minutes; This indicates the effective length of the weld seam in the task, in meters, and is determined by weld seam trajectory data or process documents. This represents the task's process feature vector, which includes plate thickness, material, bevel type, weld complexity, quality grade, and the current process to be executed. Indicates the first The state vector of each workstation at the current scheduling moment includes the equipment operating status, speed limit status, fixture status, and welding power supply status; This represents the effective welding speed obtained by combining the task process characteristics and the station status, with the unit being meters per minute. It is determined by the station's rated welding speed curve, the combination of plate thickness and material, and the current speed limit status. The geometric complexity of the weld is dimensionless and is obtained by normalizing the number of weld segments, number of turns, curvature changes, and number of arc stops and starts relative to samples of similar tasks. This represents the quality grade correction amount, which is dimensionless and is obtained from the process cycle data corresponding to the flaw detection grade, weld formation requirements, and rework control grade. This indicates the auxiliary time required for positioning, clamping, changing, program loading, and process confirmation, in minutes. It represents the availability reliability of the workstation, is dimensionless, and is obtained by fusing the effective running time within the sliding time window, the speed-limited running time, the fault recovery status, and alarm records. This indicates the degree of dispersion of the processing cycle residual for similar tasks at this workstation, expressed in minutes. The reference process duration for the task is expressed in minutes and is determined by the processing time of historical tasks with the same plate thickness range, the same weld type, and the same quality grade. This represents the sliding correction amount for the residual of the processing cycle of similar tasks, in minutes.

[0026] The Indicates the first The first task and the first Dynamic process adaptation values ​​between individual workstations, dimensionless; Indicates physical feasibility; The normalized difference vector represents the difference between the task process requirements and the stability capability of the workstation. The components of the normalized difference vector include the plate thickness deviation, the degree of dimensional proximity to the tooling boundary, the weld complexity deviation, the quality grade and inspection capability matching deviation, and the current process and workstation process mode matching deviation. Indicates the first The dynamic sensitivity matrix of each workstation is dimensionless and is obtained by fitting the sensitivity of historical processing residuals to each normalized difference component. This indicates the time from the current scheduling moment to the task delivery deadline, i.e., the first... The remaining available processing time for each workstation after deducting locked tasks and unavailable intervals, in minutes; and This is a dimensionless correction coefficient, calibrated by the impact of capacity constraints and periodic fluctuations on completion deviations in historical scheduling.

[0027] Once the expected duration and dynamic process adaptation value of a certain task station combination are calculated, the task data acquisition module continues to search for the expected entry time of the task in the timeline of that station. Specifically, the system starts from the current scheduling time and searches along the timeline of the task station. The time axis of each workstation is scanned backward, skipping the intervals of currently executing tasks, locked tasks, maintenance intervals, and unavailable intervals, and identifying the free intervals that can accommodate the expected occupancy time and necessary changeover time. If the task has a preceding process, the start point of the free interval should not be earlier than the sum of the completion time of the preceding process and the transfer time. If the fixture required for the task has not yet been released, the start point of the free interval should not be earlier than the time when the fixture is released and the positioning preparation is completed. When the earliest free interval that meets the above conditions is found, the start time of the free interval is recorded as the expected entry time of the task at the workstation, and the expected entry time plus the expected occupancy time is recorded as the expected completion time.

[0028] For example, a patch panel welding task has an effective weld length of 3.6 meters, a plate thickness of 12 millimeters, and the weld includes two corners, with a quality level corresponding to high forming requirements. A main welding station, under its current state, has an effective welding speed of 0.18 meters per minute for this plate thickness and material, an auxiliary occupancy time of 6 minutes, a current station availability reliability of 0.90, a residual dispersion of 2 minutes for similar tasks, and a reference process duration of 32 minutes. The system first obtains the basic welding time based on the weld length and effective welding speed, then combines the weld geometric complexity, quality level, auxiliary time, reliability, and residual dispersion to obtain the expected occupancy time. Next, it obtains a dynamic process adaptation value based on the differences between the task and the station in terms of plate thickness, size, process mode, remaining capacity, and cycle stability. Finally, it finds the earliest available interval on the station's time axis that can accommodate the task and outputs the expected entry time and expected completion time. This example is only for illustrating the calculation process and does not limit the scope of protection of this invention.

[0029] After all task station combinations have completed the above processing, the task data acquisition module generates a task station adaptation matrix. Each row of the task station adaptation matrix corresponds to a patch board welding task, and each column corresponds to a candidate station. Each matrix element includes a physical feasibility marker, infeasibility reason, expected occupancy time, dynamic process adaptation value, expected entry time, expected completion time, and remaining capacity occupancy before the delivery deadline. The task station adaptation matrix is ​​then input into the adaptation matrix construction module to calculate the time urgency index of each patch board welding task.

[0030] On the other hand, the adaptation matrix construction module first reads the task workstation adaptation matrix formed by the task data acquisition module, and then, for the first... Each patch board welding task extracts its set of feasible workstations, the dynamic process adaptation value corresponding to each feasible workstation, the expected entry time, the expected occupancy time, and the expected completion time. Then, it reads the load pressure of each feasible workstation in the current rolling planning cycle. The load pressure is determined by the ratio between the occupancy time of the locked task at the workstation, the allocated time for the unstarted task, and the available time of the rolling cycle. After the feasible workstation information and load pressure information are prepared, the system calculates the contribution weight of the task on each feasible workstation, so that the workstation with a high adaptation value and low load pressure has a greater impact on the expected completion status of the task, while the impact of the workstation with a low adaptation value or high load pressure is correspondingly weakened.

[0031] In one implementation, the time urgency index can be determined according to the following formula:

[0032]

[0033]

[0034]

[0035] in, Indicates the first A set of feasible workstations for each task; express The workstation for the first The contribution weights of each task at its estimated completion time are dimensionless. Indicates the first The first task and the first Dynamic process adaptation values ​​between individual workstations; Indicates the first The load pressure of each workstation within the current rolling planning cycle, dimensionless; It represents the load suppression coefficient, which is dimensionless and is determined by the degree of queue extension caused by high-load workstations in historical plans; Indicates the task is in the The expected entry time for each workstation; Indicates the task is in the Expected occupancy time at each workstation; Indicates the first The estimated completion time for the integration of these tasks; Indicates the first Delivery timeframes for each task; This indicates the waiting time for the task from its arrival in the scheduling system to the current scheduling time, in minutes; This indicates the reference process time for the task, in minutes; Indicates the first The weighted average load pressure of a set of feasible workstations for each task; , and All are dimensionless feedback coefficients, which are updated based on the impact of historical task waiting, scarcity of available workstations, and workstation load pressure on completion deviation.

[0036] The Indicates the first The scarcity index of available workstations for a task is dimensionless. When multiple feasible workstations exist for a task, but the contribution weight is concentrated on a few workstations, the information entropy term is relatively small. An increase indicates that the actual transferability of the task is relatively small; when the task has a relatively balanced contribution weight across multiple workstations, A smaller value indicates that the task has greater scheduling flexibility. This is a dimensionless indicator representing the urgency of time, with values ​​ranging from 0 to 1.

[0037] When the After the time urgency index of each task is calculated, the adaptation matrix construction module performs the same calculation process for the next patch board welding task until all tasks in the task queue have obtained the corresponding time urgency index. Subsequently, the system reorders the task queue according to the time urgency index from highest to lowest. If the time urgency indices of two tasks are in the same stable sorting range, the task with the earlier delivery deadline is prioritized; if the delivery deadlines are the same, the task with higher availability of workstations is prioritized; if they still cannot be distinguished, the original arrival order of the tasks is maintained to reduce unnecessary task order disturbances. Optionally, when multiple adjacent tasks use the same fixture, the same welding program, or the same material batch, the adaptation matrix construction module can perform local adjacency adjustments without changing the main time urgency order, reducing the time spent on changeovers and program switching when inserting workstation timelines later.

[0038] For example, Task A's estimated completion time for fusion is 15 minutes later than the delivery deadline under the current matrix, and it has been waiting for 20 minutes. Its contribution weight is mainly concentrated in one main welding station. Task B arrives earlier, but its estimated completion time for fusion is still earlier than the delivery deadline, and its contribution weight is relatively balanced across multiple stations. Task C has an earlier delivery deadline, but its available station load pressure is low. After calculating the contribution weight, estimated completion time for fusion, available station scarcity, and time urgency index for Tasks A, B, and C respectively, the system may output the priority order of Tasks A, C, and B. This example illustrates that the time urgency ranking does not solely rely on task arrival time or delivery deadline, but simultaneously considers estimated completion deviation, waiting time, available station scarcity, and station load pressure. This example does not limit the specific number of tasks or parameter values.

[0039] After the reordering is completed, the adaptation matrix construction module generates a priority-adjusted task sequence. Each task in the task sequence carries a corresponding time urgency index, a set of feasible workstations, a fusion estimated completion time, a scarcity index of available workstations, and information on the load pressure of feasible workstations. The priority-adjusted task sequence is then input into the task sorting module for workstation insertion and allocation according to dynamic priorities.

[0040] On the other hand, the task sorting module first reads the priority-adjusted task sequence and selects the patch board welding task that is currently ranked first in the sequence and has not yet been assigned a workstation. Then, it reads the set of feasible workstations for the task from the task workstation adaptation matrix and removes workstations that are currently in an unprocessable state, whose process capabilities do not meet the task requirements, or that have no available processing intervals within the time axis. Once the candidate workstation set is determined, the system reads the time axis of each candidate workstation and identifies the executing task interval, locked task interval, maintenance interval, unavailable interval, assigned but not started task interval, and available idle interval in the time axis. The executing or locked intervals are reserved as immovable occupied intervals.

[0041] Furthermore, the task sorting module calculates the required task duration for the current task and each candidate workstation. This required duration includes the expected duration calculated by the task data acquisition module, the fixture changeover time required to switch from the previous adjacent task to the current task, the welding program loading and verification time, the positioning adjustment time, and the necessary process confirmation time. If the current task and the previous adjacent task at the candidate workstation use the same fixture, the same welding program, or the same batch of materials, the system reduces the changeover and program switching time based on the workstation's historical action log. If the two tasks differ significantly in plate thickness, bevel type, welding posture, inspection level, or fixture type, the system determines the changeover time based on the corresponding workstation's actual changeover record and process program switching record.

[0042] In one implementation, the task sorting module uses the following formula to select the task insertion position:

[0043]

[0044] in, This indicates the selected workstation and its candidate available space. Indicates the first The set of candidate insertion positions for each task that meets the conditions of physical feasibility, process connection, logistics reachability and time axis accommodation at the current scheduling time; This represents the dynamic process adaptation value between the task and the workstation; Indicates the task insertion at the first The first workstation The logistics and process connection availability coefficient during a free interval is dimensionless and is determined by the current position of the patch plate, the availability status of the transfer equipment, the completion time of the preceding process, the release status of the fixture, and the arrival preparation time. Indicates the first The first workstation The effective idle time of each candidate idle interval after deducting necessary prior preparations, in minutes; Indicates the required duration of the task, in minutes; This indicates the time for changing equipment or switching processes, in minutes. This indicates the time for program verification, location confirmation, or test preparation, in minutes. Indicates the first The representative remaining idle demand for the processing time of tasks at each workstation in the current set of unstarted tasks, in minutes, can be determined by the median or weighted median of the distribution of processing time for tasks at that workstation. This is a dimensionless fragmentation suppression coefficient, updated by the proportion of remaining idle intervals in historical production scheduling that cannot be utilized by subsequent tasks.

[0045] In the above formula, The term represents the accommodative capacity of a candidate workstation's effective idle time relative to the current task's required time, and is the core calculation item for workstation insertion and allocation. The exponential term is used to suppress the formation of unusable time fragments after insertion. When the residual idle time generated after the current task's insertion is close to the representative requirements of subsequent processable tasks, the candidate position retains a high matching value; when the residual idle time does not significantly match the distribution of subsequent task requirements, the matching value of the candidate position decreases. This calculation ensures that task allocation considers not only which workstation is currently idle, but also whether the idle interval is suitable for the current task and whether the remaining time after insertion can still be utilized by subsequent tasks.

[0046] Once the matching values ​​for all candidate insertion positions have been calculated, the task sorting module selects the workstation with the largest matching value and its idle interval as the insertion position for the current task. If multiple candidate positions have the same matching value, the position with the earlier expected completion time is selected. If the expected completion times are still the same, the position with a more even distribution of remaining available processing time after insertion is selected. Subsequently, the system determines the planned start time of the current task, which is the later of the following: the start time of the candidate idle interval, the time when the task can be processed, the time when the preceding process is completed plus the transfer time, the time when the fixture is changed, and the time when the program verification is completed. The planned completion time is then obtained by adding the task's required duration to the planned start time, and the task's occupied interval is written into the time axis of the selected workstation.

[0047] Once the current task is inserted, the task sorting module updates the task backlog and equipment occupancy time distribution for the selected workstation. This update includes increasing the planned occupancy interval for the task, updating the number of incomplete tasks at the workstation, the total required duration of incomplete tasks, the proportion of rolling cycles already occupied, the distribution of remaining free intervals, the expected entry time of subsequent tasks, and the workstation load pressure. Subsequently, the system returns to the priority-adjusted task sequence, reads the next uninserted patch board welding task, and repeats the candidate workstation screening, free interval search, task required duration calculation, matching value calculation, insertion, and update process based on the updated workstation timeline until all tasks in the priority-adjusted task sequence have completed workstation insertion and allocation.

[0048] For example, the expected time for task D at workstation W1 is 18 minutes, with a total of 3 minutes for changeover and verification, and a dynamic process adaptation value of 0.88. W1 has an effective idle interval from 9:10 to 9:45, with an effective idle time of 35 minutes. After inserting task D, the remaining idle time is 14 minutes, while the representative value of the time W1 can process for subsequent tasks is 15 minutes. The expected time for task D at workstation W2 is 27 minutes, with a total of 5 minutes for changeover and verification, and a dynamic process adaptation value of 0.80. W2 has an effective idle interval from 9:05 to 10:05, with an effective idle time of 60 minutes, but the remaining idle time after insertion differs significantly from the representative requirement of subsequent tasks. After calculating the matching values ​​for the corresponding candidate positions in W1 and W2, the system inserts task D into the 9:10 to 9:31 interval of W1 and updates the timeline and task backlog status of W1. This example is only used to illustrate the process of workstation selection, timeline insertion, and status update, and does not limit the specific number of workstations, time period length, or parameter values.

[0049] Once all tasks have been inserted, the task sorting module outputs a workstation task allocation scheme. This scheme includes the task identifier, assigned workstation, planned start time, planned finish time, current process, preceding and following process connection information, and corresponding workstation timeline update results for each patch board welding task. This workstation task allocation scheme then serves as the basis for the dynamic update module's dynamic update processing.

[0050] On the other hand, the workstation allocation module first continuously receives status events from the workstation controller, welding robot control system, welding power supply, fixture control unit, inspection equipment, and production execution system. When it receives status events such as fault shutdown, fault recovery, maintenance start, maintenance end, speed limit operation, unstable welding power supply, process mode switching, fixture unavailability, change in inspection equipment occupancy, or unavailability of transfer equipment, the system marks the event as a workstation status change event. At the same time, the system reads the robot trajectory completion ratio, effective arc burning time, real-time welding length, remaining weld length, and actual occupancy time of the current task, and compares the above real-time progress data with the stable fluctuation envelope formed by the historical processing cycle residual of similar tasks. When the actual remaining occupancy time continuously deviates from the stable fluctuation envelope, the system marks the situation as a processing cycle abnormal evolution event. When the production execution system or task issuing terminal sends a new patch plate welding task, the system marks it as a new task arrival event.

[0051] Upon the occurrence of any of the aforementioned events, the workstation allocation module uses the event occurrence time as the starting point of the new rolling plan, reads the current workstation task allocation scheme and the time axis of each workstation, and then iterates through all incomplete tasks, performing an execution status determination for each task: if the task has already started welding, or has completed positioning and clamping and transfer would disrupt process continuity, or has entered the inspection process and interruption would affect inspection consistency, then the time axis interval corresponding to the task is marked as a locked interval and retained in the original workstation; if the task has not yet entered the workstation, or has not yet completed clamping, or has not yet loaded the welding program, or is still in a waiting-for-transfer state, then the task is marked as a reassignable task, and the unstarted time interval occupied in the original scheme is released. For tasks that are being executed but not yet completed, the system re-estimates the remaining occupied time based on the real-time progress and corrects the corresponding workstation time axis with the updated expected completion time.

[0052] Furthermore, once the set of reassignable tasks is determined, the workstation allocation module will add new patch board welding tasks to the set and reread the equipment status and process capabilities of each workstation at the current moment. If a workstation changes from an available state to a fault-stopped state, the system will remove the time interval of that workstation before recovery from the available time axis and update its availability reliability and effective welding speed. If a workstation recovers from a fault state, the system will re-include the recovered available interval into the candidate workstation set after obtaining information that the equipment self-test is complete, the welding power supply is stable, the fixture status is valid, and the process mode verification is complete. If a workstation is in a speed-limited operation state, the system will adjust the effective welding speed according to the speed limit information fed back by the controller or the real-time welding speed, so that the expected occupancy time of the task at that workstation changes with the equipment status.

[0053] Once the latest workstation status and the set of reassignable tasks are ready, the workstation allocation module recalculates the physical feasibility markers, expected occupancy durations, dynamic process adaptation values, expected entry times, and expected completion times between reassignable tasks and each candidate workstation according to the processing flow described in the task data acquisition module, forming an updated task workstation adaptation matrix. Subsequently, the time urgency index of reassignable tasks is recalculated according to the processing flow described in the adaptation matrix construction module, and a new priority-adjusted task sequence is generated. Then, according to the processing flow described in the task sorting module, on the workstation time axis that retains the locked task interval, candidate workstation screening, idle interval search, task requirement duration calculation, matching value calculation, task insertion, and workstation time axis update are performed one by one on the new priority-adjusted task sequence to form a candidate dynamic update scheme.

[0054] On the other hand, to avoid unstarted tasks repeatedly migrating between multiple workstations due to minor status fluctuations, the workstation allocation module compares the candidate dynamic update scheme with the original workstation task allocation scheme before the event occurs, using a normalized comprehensive cost comparison, before adopting the candidate dynamic update scheme. The normalized comprehensive cost includes the deviation of the expected completion time from the delivery deadline, the dispersion of load pressure within the rolling planning cycle of each workstation, task transfer time, fixture changeover time, welding program withdrawal and reloading time, process connection waiting time, and task sequence disturbance. Among these, time-related quantities are dimensionless based on the task's reference process duration or the remaining available time of the shift, while load-related quantities are normalized based on the distribution of available processing time for each workstation within the current rolling cycle. If a candidate dynamic update scheme has a lower normalized overall cost after taking into account transfer, reloading, program withdrawal, and sequence disturbances, the workstation allocation module adopts the candidate dynamic update scheme; if the candidate dynamic update scheme does not reduce the normalized overall cost, the workstation allocation module retains the allocation of unstarted tasks in the original scheme that are not affected by the event, and only performs necessary reallocation for tasks directly affected by faults, speed limits, new tasks, or previous delays.

[0055] For example, in the original task allocation scheme, tasks E and F were both assigned to workstation W3 but had not yet started. At 9:20, the welding power supply of W3 reported a fault and stopped, and the expected completion time of the currently executing task was postponed. At the same time, workstation W1 completed its preceding task at 9:25 and created an available idle interval. Although workstation W2 was available, its dynamic process adaptation value for the plate thickness and bevel type required by task E was low. The system uses 9:20 as the new rolling plan starting point, locks the occupied interval of the currently executing task in W3, releases the unstarted interval of tasks E and F in W3, and includes tasks E and F in the set of reassignable tasks. Then, it recalculates the expected occupancy time, dynamic process adaptation value, and expected entry time of tasks E and F in W1, W2, and other available workstations, recalculates the time urgency index and priority order of the two tasks, and performs insertion according to the new workstation time axis. If the calculation results show that task E can be completed before the delivery deadline after being transferred to W1, and the overall cost of inserting task F into W2 or waiting for W3 to recover is lower, then the system outputs the corresponding dynamic update plan. This example is only used to illustrate the event triggering, recalculation, and reassignment process, and does not limit the specific fault type, time, or number of workstations.

[0056] Finally, the workstation allocation module generates a dynamically updated workstation task allocation scheme. This scheme includes the task identifier for each incomplete patch board welding task, the current process, the allocated workstation, the planned start time, the planned completion time, the connection information between preceding and following processes, whether reassignment has occurred, the reason for reassignment, the scheme version number, and the distribution of time slots occupied by each workstation after the update. The system then sends the dynamically updated workstation task allocation scheme to the production execution system and the corresponding workstation controller, and records the triggering event, input status, adaptation matrix, time urgency ranking result, and insertion result for this dynamic update. When subsequent equipment status changes, abnormal evolution of the processing cycle, or the arrival of new tasks occur, the workstation allocation module repeats the above dynamic update process based on the latest scheme.

[0057] In summary, the patch board welding task dynamic scheduling system based on multi-station collaboration of the present invention is explained. It constructs a task station adaptation matrix that integrates the real-time reliability of the equipment and the distribution of process residuals, and reorders tasks based on the expected completion deviation and the time urgency of the available stations. Tasks are dynamically inserted by matching effective idle time and suppressing time fragmentation. At the same time, a disturbance suppression rescheduling mechanism based on normalized comprehensive cost comparison is introduced. In this way, the load of each station can be dynamically balanced, the equipment utilization rate and overall capacity can be improved, the task completion cycle can be shortened, and the additional transfer and replacement disturbances caused by rescheduling under abnormal working conditions can be effectively reduced, realizing adaptive and efficient collaborative scheduling of patch board welding tasks in a multi-station dynamic environment.

[0058] Here, those skilled in the art will understand that the specific operations of each step in the above-described dynamic scheduling method for patch board welding tasks based on multi-station collaboration have been referenced above. Figure 1 and Figure 2 The description of the dynamic scheduling system for patch board welding tasks based on multi-station collaboration is detailed here, and therefore, its repeated description will be omitted.

[0059] In summary, the dynamic scheduling method for patch board welding tasks based on multi-station collaboration based on embodiments of the present invention is explained. It constructs a task station adaptation matrix that integrates the real-time reliability of the equipment and the distribution of process residuals, and reorders tasks based on the expected completion deviation and the time urgency of the available stations. Tasks are dynamically inserted by matching effective idle time and suppressing time fragmentation. At the same time, a disturbance suppression rescheduling mechanism based on normalized comprehensive cost comparison is introduced. In this way, the load of each station can be dynamically balanced, the equipment utilization rate and overall capacity can be improved, the task completion cycle can be shortened, and the additional transfer and replacement disturbances caused by rescheduling under abnormal working conditions can be effectively reduced, realizing adaptive and efficient collaborative scheduling of patch board welding tasks in a multi-station dynamic environment.

Claims

1. A dynamic scheduling system for patch board welding tasks based on multi-station collaboration, characterized in that, include: The module includes a task data acquisition module, an adaptation matrix construction module, a task sorting module, a workstation allocation module, and a dynamic update module. The task data acquisition module is used to acquire the queue of welding tasks to be scheduled and the real-time operation data of each workstation; The adaptation matrix construction module is used to construct a task-workstation adaptation matrix based on task processing requirements, workstation process capabilities, equipment status, and remaining workstation capacity. The task sorting module is used to determine the time urgency index based on the task workstation adaptation matrix and generate a priority-adjusted task sequence. The workstation allocation module is used to insert tasks in the task sequence into the corresponding workstation timeline. The dynamic update module is used to reallocate tasks that have not yet started execution when the workstation status changes or new tasks arrive, and to generate a dynamically updated workstation task allocation scheme.

2. The dynamic scheduling system for patch board welding tasks based on multi-station collaboration as described in claim 1, characterized in that, The task station adaptation matrix includes a physical feasibility marker, expected occupancy time, dynamic process adaptation value, expected entry time, and remaining available processing time before the delivery deadline for each task station combination; wherein, the expected entry time is determined by the locked occupancy interval, the unavailable interval, and the idle interval that can accommodate the task requirement time in the time axis of the corresponding station.

3. The dynamic scheduling system for patch board welding tasks based on multi-station collaboration according to claim 2, characterized in that, The adaptation matrix construction module includes a dynamic process adaptation unit, which is used to integrate the normalized difference between task process requirements and workstation stability capability, workstation availability reliability, remaining available processing time before delivery deadline, and the degree of dispersion of processing cycle residual into a dynamic process adaptation value.

4. The dynamic scheduling system for patch board welding tasks based on multi-station collaboration according to claim 3, characterized in that, The dynamic process adaptation value is determined according to the following formula: ； in, For dynamic process adaptation values, Marking as physically feasible. For the reliability of the workstation, For the normalized difference vector, For dynamic sensitivity matrix, For the expected duration, The remaining available processing time before the delivery deadline. For reference process time of the task, This represents the degree of dispersion of the residual during the processing cycle.

5. The dynamic scheduling system for patch board welding tasks based on multi-station collaboration according to claim 4, characterized in that, The task sorting module includes an urgency calculation unit, which is used to determine the contribution weight based on the dynamic process adaptation value and workstation load pressure of each feasible workstation, and calculate the fusion of the expected completion time and the scarcity index of the available workstations based on the contribution weight, and then combine the task waiting time and delivery deadline to determine the time urgency index.

6. The dynamic scheduling system for patch board welding tasks based on multi-station collaboration according to claim 5, characterized in that, The workstation allocation module includes an idle interval matching unit, which is used to select workstations and idle intervals from a set of candidate insertion positions that meet the conditions of physical feasibility, process connection, logistics accessibility and time axis capacity, based on dynamic process adaptation value, logistics and process connection accessibility coefficient, ratio of effective idle time to task requirement time and residual idle availability.

7. The dynamic scheduling system for patch board welding tasks based on multi-station collaboration according to claim 6, characterized in that, The dynamic update module includes a locked interval retention unit and an unstarted task reassignment unit. The locked interval retention unit is used to retain started tasks and locked process intervals when the workstation status changes or new tasks arrive. The unstarted task reassignment unit is used to release the original allocation intervals of unstarted tasks and regenerate the task workstation adaptation matrix, time urgency index and workstation task allocation scheme based on the updated equipment status, workstation remaining capacity and new tasks.

8. A dynamic scheduling method for patch board welding tasks based on multi-station collaboration, characterized in that, include: Receive the queue of patch board welding tasks to be scheduled, extract the task process feature parameters of each patch board welding task, and obtain the equipment process capability, real-time equipment status and current task backlog status of each station. Calculate the process adaptation value and expected entry time of each patch board welding task at each station, and construct the task station adaptation matrix. Based on the task station adaptation matrix, the deviation between the expected entry time, expected completion status and delivery time of each patch board welding task is quantified into a time urgency index, and the queue of patch board welding tasks to be scheduled is reordered according to the time urgency index to generate a priority-adjusted task sequence. For the priority-adjusted task sequence, each patch board welding task is sequentially searched for an insertable free interval in the candidate workstation time axis. The workstation and free interval that best match the effective free time of the workstation with the task requirement time are selected. The patch board welding task is then inserted into the corresponding workstation time axis, and the workstation task backlog and equipment occupancy time distribution are updated. When a change in equipment status, abnormal evolution of processing cycle, or arrival of a new patch board welding task is detected at any workstation, the task workstation adaptation matrix and the time urgency index are recalculated, and the workstations for patch board welding tasks that have not yet started are reallocated to generate a dynamically updated workstation task allocation scheme.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the dynamic scheduling system for patch board welding tasks based on multi-station collaboration as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the dynamic scheduling system for patch board welding tasks based on multi-station collaboration as described in any one of claims 1 to 7.

Images