A Multi-Constraint Scheduling Optimization Method for Shipbuilding

Abstract

This invention relates to a multi-constraint scheduling optimization method for shipbuilding. It acquires shipbuilding task sets and process sequence information, work surface status information, key resource occupancy status information, handover relationship information, delivery node information, and execution schedule. When a disturbance event is detected, the affected tasks are identified. Disturbance absorption verification is performed on tasks associated with the affected tasks, and then, based on preset diffusion conditions, it is determined whether they should be included in the rescheduling scope. Tasks not included in the rescheduling scope are set as frozen tasks. Within the rescheduling scope, scheduling adjustments are performed according to a preset adjustment order, generating and outputting an updated schedule. This invention can simultaneously improve scheduling adjustment efficiency and global optimization effect under multi-constraint conditions, effectively shortening the production cycle, reducing the probability of resource conflicts, and improving the overall collaborative efficiency and delivery reliability of the shipbuilding process.

Description

Technical Field

[0001] This invention belongs to the field of shipbuilding production management and scheduling optimization technology, specifically a multi-constraint scheduling optimization method for shipbuilding. Background Technology

[0002] Existing technologies, such as the ship or marine engineering segment scheduling method in patent application CN115099656A, have constructed a relatively complete scheduling system architecture through input layer, algorithm layer and output layer, and combined simulation module and heuristic rules to achieve matching optimization of segment and tire position resources. To a certain extent, they can improve tire position utilization and production efficiency. However, from the perspective of multi-constraint dynamic scheduling optimization, there are still obvious shortcomings and limitations. First, this type of method still belongs to the scheduling approach of centralized modeling plus overall solution. Its core is to generate a better segmented layout scheme through one or more simulation iterations. Its algorithm mainly revolves around segmented sorting, tire position screening and matching rules, such as sorting according to delivery date and processing time, selecting idle tire positions and allocating according to rules such as size and capacity matching. This approach focuses more on initial scheduling or periodic rescheduling and lacks fine-grained processing mechanism for frequent real-time disturbances in the production process. When equipment failure, resource fluctuations or process delays occur, it is often necessary to re-enter the overall solution process, resulting in a large range of scheduling adjustments, high computational costs, and easy to cause scheduling oscillations, which is not conducive to stable execution on site.

[0003] Secondly, while this method mentions dynamic information feedback and simulation optimization, its dynamism is mainly reflected in the updating of input information and simulation iteration. It lacks a control mechanism for the propagation path of disturbances and does not perform tiered processing of affected and unaffected tasks. The absence of strategies for local absorption and diffusion judgment means that once a local task changes, it often requires a complete recalculation, making it difficult to prevent the disorderly spread of disturbances in the process and resource chains, thus reducing system robustness. Thirdly, although the method covers timing constraints, resource constraints, and process constraints in constraint modeling, its scheduling decisions mainly revolve around resource matching and completion time optimization. It does not explicitly introduce strong correlation constraints between handover relationships and delivery nodes, nor does it protect the stability of the handover window. This can lead to disruption of upstream and downstream connections during scheduling adjustments, affecting the overall production rhythm. Furthermore, the method lacks fine-grained classification of task states in its task processing strategy. It does not distinguish between tasks that have entered the execution commitment stage and adjustable tasks, nor does it establish a freezing mechanism to constrain critical tasks from being disturbed or adjusted. This makes it easy to interfere with stable or committed tasks during dynamic rescheduling, affecting the controllability and reliability of production organization. Furthermore, this method primarily relies on heuristic rules and simple sorting logic for scheduling optimization, failing to construct evaluation indicators that reflect the system's bottleneck-resolving capabilities. It lacks a priority determination mechanism based on obstacle-resolving capabilities and cannot prioritize tasks that can simultaneously restore multiple blocked process chains or release critical work surfaces, thus limiting optimization efficiency in complex, multi-constraint coupled environments. Simultaneously, while its simulation optimization process can be used to evaluate the merits of different solutions, it lacks a mechanism for real-time re-evaluation of task priorities during scheduling adjustments and does not consider the transitions in task states due to prior adjustments, resulting in delayed scheduling order updates and difficulty in timely capturing optimization opportunities arising from system structural changes.

[0004] In summary, while existing technologies have some effect on static or quasi-dynamic scheduling optimization, they still have problems such as difficulty in absorbing local disturbances, excessive adjustment range, lack of protection for critical tasks, lack of targeted optimization sequence, and insufficient dynamic response capability when dealing with the complex environment of multiple disturbances, high coupling, and strong constraints in shipbuilding. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-constraint scheduling optimization method for ship manufacturing, thereby addressing some of the drawbacks and shortcomings pointed out in the background art.

[0006] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows: a multi-constraint scheduling optimization method for ship manufacturing, comprising: acquiring ship manufacturing task set and process sequence information, work surface status information, key resource occupancy status information, handover relationship information, delivery node information and execution schedule;

[0007] When a disturbance event is detected, the affected tasks are identified; disturbance absorption verification is performed on the tasks associated with the affected tasks, and it is then determined whether they are included in the reordering scope based on preset diffusion conditions. Tasks not included in the reordering scope are set as frozen tasks; within the reordering scope, scheduling adjustments are performed according to a preset adjustment order, and an updated schedule is generated and output.

[0008] Furthermore, the disturbance absorption verification is used to determine whether the associated tasks can continue to be executed without changing the process sequence, handover relationship and delivery node by utilizing the reserved buffer period, shift adjustment or alternative resources; associated tasks that can continue to be executed are not included in the rearrangement scope and are set as frozen tasks.

[0009] Furthermore, the frozen tasks include committed frozen tasks and observed frozen tasks; committed frozen tasks are tasks that have entered the commencement commitment window or whose downstream tasks have completed the preparation for receipt, while observed frozen tasks are tasks that are temporarily not included in the rescheduling scope after disturbance absorption verification; committed frozen tasks remain unchanged when scheduling adjustments are performed; when it is re-determined that observed frozen tasks do not change the delivery node, they are unfrozen and transferred to the rescheduling scope.

[0010] Furthermore, the preset adjustment order is determined according to the task's unblocking capability; the unblocking capability is determined based on the number of blocked handover relationships restored after the task adjustment and the number of newly released shared work surfaces; when performing scheduling adjustments, tasks that can simultaneously restore multiple blocked handover relationships or release shared work surfaces are given priority for adjustment.

[0011] Furthermore, the disturbance absorption verification is performed sequentially according to the reserved buffer period, shift adjustment, and alternative resources; when the previous absorption method meets the conditions for the related task to continue execution, the verification of the next absorption method is terminated, and the related task is set as a frozen task.

[0012] Furthermore, after setting the associated task as a frozen task, the original correspondence between the handover window and the delivery node remains unchanged, and only the execution time period, execution shift, or execution resources are adjusted.

[0013] Furthermore, the unblocking capability is classified according to the number of blocked handover relationships restored and the number of shared working surfaces released after task adjustment, and tasks that simultaneously satisfy the requirements of restoring blocked handover relationships and releasing shared working surfaces are identified as priority adjustment tasks.

[0014] Furthermore, when multiple tasks have the same deblocking capability, priority is given to adjusting the task that does not change the execution state of the frozen task; if they are still the same, priority is given to adjusting the task that does not change the delivery node.

[0015] Furthermore, after each task's scheduling adjustment is completed, the unblocking capability of the remaining tasks is reassessed; when the unblocking capability of subsequent tasks changes due to the adjustment of the previous task, the preset adjustment order is updated according to the reassessment result.

[0016] Furthermore, when reassessing the unblocking capability of the remaining tasks, tasks that have changed from a blocked state to a transferable state or from a shared work surface occupied state to a release state due to the previous task adjustment are marked as sequential transition tasks, and these sequential transition tasks are inserted at the beginning of the current preset adjustment order for priority execution.

[0017] The beneficial effects of this invention are as follows: This invention provides a multi-constraint scheduling optimization method for shipbuilding. By introducing a disturbance absorption verification and diffusion judgment mechanism after a disturbance occurs, it achieves fine control over the range of scheduling adjustments. By prioritizing the use of reserved buffer periods, shift adjustments, and alternative resources to locally absorb related tasks, tasks that can continue to be executed maintain their original process sequence, handover relationships, and delivery nodes, and are designated as frozen tasks, thereby effectively avoiding the chain reaction caused by large-scale rescheduling. Simultaneously, by dividing frozen tasks into committed frozen tasks and observed frozen tasks, and combining them with a dynamic unfreezing mechanism, the stability of critical operation commitments is ensured while also considering the flexibility of scheduling adjustments, significantly improving the stability and disturbance resistance of the scheduling system in complex production environments.

[0018] Furthermore, this invention constructs a preset adjustment sequence based on unblocking capabilities, using the number of restored obstructed handover relationships and the number of released shared work surfaces as core indicators to classify and prioritize tasks. This ensures that scheduling adjustments prioritize tasks that can clear critical operational links and release bottleneck resources. Simultaneously, a dynamic re-judgment mechanism and a sequential transition task identification strategy are introduced during the adjustment process, enabling tasks whose status has improved due to prior adjustments to be promptly moved forward for execution, thereby continuously optimizing the scheduling structure. Through these technical means, this invention can simultaneously improve scheduling adjustment efficiency and global optimization effects under multiple constraints, effectively shortening the production cycle, reducing the probability of resource conflicts, and improving the overall collaborative efficiency and delivery reliability of the shipbuilding process. Attached Figure Description

[0019] Figure 1 This is a functional relationship diagram of the multi-constraint scheduling optimization method for ship manufacturing according to the present invention.

[0020] Figure 2 This is a comparison diagram of the original plan and the plan after partial rearrangement in Embodiment 1 of the present invention.

[0021] Figure 3 This is a trajectory diagram of the cumulative contribution of disturbance absorption in Embodiment 1 of the present invention.

[0022] Figure 4This is a bubble matrix diagram of the joint control node margin in Embodiment 1 of the present invention.

[0023] Figure 5 This is a ranking diagram of the initial deblocking capabilities of candidate tasks in Embodiment 2 of the present invention.

[0024] Figure 6 This is a dynamically updated task priority transition trajectory diagram in Embodiment 2 of the present invention.

[0025] Figure 7 This is the updated multi-dimensional capability radar chart for critical tasks in Embodiment 2 of the present invention. Detailed Implementation

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] Combined with appendix Figure 1 This invention discloses a multi-constraint scheduling optimization method for shipbuilding, applicable to production scenarios such as ship section manufacturing, assembly, outfitting, painting, and transfer. It first acquires shipbuilding task sets and process sequence information, work surface status information, key resource occupancy status information, transfer relationship information, delivery node information, and execution schedule. Specifically, the shipbuilding task set represents the work content, duration, area, and sequential relationships of each production task; the process sequence information defines the irreversible execution order between tasks; the work surface status information represents the occupancy of work locations such as docks, wharves, section assembly areas, and compartment areas; the key resource occupancy status information represents the usage status of resources such as cranes, transportation equipment, special tooling, and work teams; the transfer relationship information represents the correspondence between upstream and downstream work surfaces after the completion of upstream tasks; the delivery node information represents target time points such as section completion, assembly completion, launching preparation, or whole-ship delivery; and the execution schedule represents the current planned start time, planned completion time, execution shift, and resource allocation results for each task. Based on the above information, the system establishes a multi-constraint scheduling model, ensuring that each task simultaneously satisfies process sequence constraints, work surface mutual exclusion constraints, critical resource capacity constraints, handover and connection constraints, and delivery node constraints, thereby forming an executable initial scheduling basis.

[0028] When a disturbance event is detected during production, the affected tasks are first identified. Disturbance events include critical equipment failures, delayed release of work surfaces, temporary resource shortages, delays in upstream tasks, or the entry of additional tasks. Subsequently, disturbance absorption checks are performed on related tasks that have process connections, resource competition, work surface sharing, or handover dependencies with the affected tasks. This determines whether the related tasks can continue execution through reserved buffer periods, shift adjustments, or alternative resources without altering the established process sequence, handover relationships, and delivery milestones. Related tasks that can absorb disturbances themselves are excluded from the rescheduling scope and designated as frozen tasks to maintain the basic stability of their original execution arrangements. Tasks that cannot absorb disturbances or have met preset diffusion conditions are included in the rescheduling scope. Preset diffusion conditions are used to determine whether the disturbance has spread to subsequent processes, shared work surfaces, or critical resource chains. Subsequently, within the rearrangement scope, scheduling adjustments are performed according to the preset adjustment order, prioritizing tasks that can unblock, restore handover connections, or release shared work surfaces. After each adjustment, the compatibility between the frozen tasks and the impact on delivery nodes are simultaneously verified until the tasks within the rearrangement scope meet all constraints again. Finally, an updated schedule is generated and output to achieve rapid response and stable optimization of the shipbuilding process under complex disturbance conditions.

[0029] Disturbance absorption verification is used to assess the feasibility of each associated task to determine whether it can continue to execute without triggering an overall schedule refactoring. Specifically, for each associated task, it first determines whether there is an available reserved buffer period. By comparing the time margin between the original planned start time and the latest allowed start time, it assesses whether the task still meets its downstream handover relationship and delivery node requirements after delayed execution. If it does, disturbance absorption is achieved by occupying the buffer period. If the reserved buffer is insufficient to cover the impact of the disturbance, it further determines whether absorption can be achieved through shift adjustments. That is, without changing the task's process sequence and handover connection, by increasing work shifts, adjusting shift work time, or compressing non-critical idle intervals, the task can be completed within a new time window without affecting the established delivery node. If shift adjustments still cannot meet the requirements, it further determines whether there are alternative resources. By selecting equipment, work shifts, or work areas with the same or equivalent capabilities to replace the original resources, the task can continue to execute under the given constraints. The above checks are performed sequentially according to the reserved buffer period, shift adjustments, and alternative resources. Once a certain method meets the condition for continuous task execution, subsequent checks are terminated, and the associated task is considered to have completed disturbance absorption. For associated tasks that have completed disturbance absorption, their original process sequence, handover relationship, and delivery node remain unchanged. Only partial adjustments to their specific execution period, execution shift, or resource configuration are allowed, and they are removed from the rescheduling scope and set as frozen tasks to reduce the scope of scheduling adjustments and improve overall scheduling stability.

[0030] Frozen tasks are categorized into committed frozen tasks and observation frozen tasks based on their stability and constraint strength. Committed frozen tasks represent tasks that have entered the commencement commitment window or whose downstream tasks have completed acceptance preparation. Adjustments to these tasks will directly impact on-site organization or subsequent continuous operations; therefore, their execution status remains unchanged during scheduling adjustments, including planned commencement time, planned completion time, work shifts, and resource allocation, to ensure the stability of production commitments and on-site coordination. Observation frozen tasks represent tasks that, after disturbance absorption verification, meet the conditions for continued execution but are not currently included in the rescheduling scope. These tasks maintain their original scheduling arrangement at the current stage, but the system continuously tracks and evaluates their constraint status, including their impact on delivery nodes, handover relationships, and resource usage.

[0031] During the scheduling adjustment process, only tasks included in the rescheduling scope undergo sequence optimization and resource reallocation, while committed frozen tasks remain completely unchanged, and observed frozen tasks are dynamically reviewed. If, in subsequent iterations, the system re-determines that a certain observed frozen task will not cause changes in delivery nodes after participating in the scheduling adjustment, and will not disrupt existing handover relationships or critical resource constraints, then the frozen state of the task is lifted, and it is transferred to the rescheduling scope to participate in subsequent optimization, thereby improving the overall scheduling flexibility and global optimization effect. If the re-determination result indicates that adjusting the task will cause delivery node shifts or critical constraints to fail, then its frozen state is maintained, thus achieving dynamic control of the scheduling adjustment scope while ensuring the stability of critical nodes.

[0032] The preset adjustment order is determined based on the unblocking capability of each task, so that scheduling adjustments are prioritized for critical tasks that can significantly improve the system's constraint status. Specifically, the unblocking capability of each task included in the rescheduling scope is assessed. Unblocking capability is comprehensively determined by the degree to which the task restores blocked handover relationships in the system and releases the occupied state of shared workspaces after adjusting the execution time, shift, or resource allocation. The number of restored blocked handover relationships represents the number of effective connections between upstream and downstream that can be re-established after the task's adjustment, and the number of newly released shared workspaces represents the number of job spaces available for other tasks after the task's adjustment. By statistically analyzing and ranking these two indicators, the unblocking capability priority of each task is formed, and the preset adjustment order is generated accordingly.

[0033] During the scheduling adjustment process, tasks are optimized sequentially according to a preset adjustment order. Priority is given to tasks that can simultaneously restore multiple blocked handover relationships or release shared workfaces, thereby quickly opening up critical operation links and expanding available workface resources in local adjustments. After each task adjustment is completed, the handover relationships and workface occupancy status in the system are updated in real time, and the unblocking capabilities of the remaining tasks are reassessed to ensure that subsequent adjustments are always based on the latest constraint status.

[0034] Disturbance absorption verification is performed sequentially, following the order of reserved buffer time, shift adjustments, and alternative resources, to ensure that the method with the least impact on the original schedule is prioritized for disturbance absorption. Specifically, the reserved buffer time is first verified for related tasks. By calculating the time margin between the original planned execution interval and the latest allowed completion time, it is determined whether the disturbance impact can be absorbed by extending the execution while still meeting the existing handover relationship and delivery node constraints. If the reserved buffer is insufficient, shift adjustment verification is further performed. By increasing the number of work shifts or optimizing the work time distribution of the shift group, the task duration is compressed without changing the process sequence. If shift adjustment still cannot meet the conditions, alternative resource verification is performed. By matching equipment, personnel, or work surface resources with equivalent capabilities, the task can continue to be executed under the new resource configuration. The above verifications are performed sequentially. Once the previous absorption method has met the conditions for the related task to continue execution, the verification process of the subsequent absorption methods is terminated, and the task is deemed to have completed the disturbance absorption process.

[0035] After absorbing the disturbance, related tasks are designated as frozen tasks to prevent them from participating in subsequent rescheduling and triggering a chain reaction of adjustments. For related tasks designated as frozen tasks, the original correspondence between their handover windows and delivery nodes remains unchanged, thereby ensuring the stability of upstream and downstream operations and the overall delivery target. Based on this, only necessary local adjustments are allowed to the execution time, shift, or execution resources of the task to adapt to changes brought about by the disturbance, while maintaining its logical position and constraints in the overall schedule. This allows for the smooth operation of the production system while controlling the scope of adjustments.

[0036] The ability to resolve obstructions is categorized based on the number of obstructed handover relationships restored and the number of shared work surfaces released after task adjustments, forming a hierarchical adjustment priority system. Specifically, firstly, the number of restored obstructed handover relationships serves as the primary criterion for categorization, classifying tasks that can restore more handover connections as higher-level tasks. Secondly, the number of released shared work surfaces serves as a secondary criterion, prioritizing tasks that release more workspace under the same restoration conditions. For tasks that simultaneously meet the conditions of restoring obstructed handover relationships and releasing shared work surfaces, they are identified as priority adjustment tasks and placed at the top of the adjustment order. Through this categorization method, scheduling adjustments are prioritized for key node tasks that can simultaneously improve process connections and workspace utilization, thereby accelerating the system's transition from an obstructed state to an executable state.

[0037] When multiple tasks have the same level of unblocking capability, stability constraints are further introduced for prioritization. Tasks that do not change the execution status of frozen tasks are prioritized for adjustment to avoid disturbing already stable tasks. Furthermore, if tasks with the same priority still exist, those that do not change the delivery node after adjustment are prioritized to ensure the overall delivery target remains unaffected. During scheduling adjustments, after each task adjustment is completed, the handover relationship status and shared workspace occupancy status in the system are updated in real time, and the unblocking capability of remaining tasks is reassessed. When the unblocking capability of subsequent tasks changes due to the adjustment of a previous task, the preset adjustment order is dynamically updated based on the reassessment results, ensuring that subsequent adjustments are always based on the latest constraints, thereby improving the responsiveness and global coordination of the scheduling optimization process.

[0038] After each task's scheduling adjustment, the remaining tasks are reassessed for their unblocking capabilities to reflect real-time changes in system constraints. Specifically, based on the updated handover relationships and workspace occupancy status, each remaining task is identified to determine whether it has transitioned from a blocked state to a handoverable state, or from a shared workspace occupancy state to a release state, due to the previous task's adjustment. Tasks satisfying either of these transition conditions are marked as sequential transition tasks, indicating that they possess higher execution urgency and stronger unblocking capabilities in the current scheduling environment. This marking mechanism enables the system to promptly capture structural improvements resulting from local adjustments and identify potentially critical tasks from the original order.

[0039] When updating the preset adjustment order, sequential transition tasks are inserted at the beginning of the current adjustment sequence for priority execution, giving them higher priority in subsequent scheduling adjustments. The insertion position is determined based on their unblocking capability level. When multiple sequential transition tasks exist simultaneously, they are sorted according to the number of restored handover relationships and the number of released shared working surfaces, and further filtered based on constraints that do not change the frozen task status and do not affect delivery nodes. By processing sequential transition tasks in advance, the restoration of critical paths and the release of shared resources can be accelerated, thereby continuously optimizing the overall scheduling structure during continuous iteration and improving the efficiency and stability of scheduling adjustments under multiple constraints.

[0040] Example 1:

[0041] In this embodiment, a coastal shipyard is building an 82,000-ton bulk carrier, currently in the coordinated stage of stern section manufacturing and pre-outfitting. This stage involves continuous operations in the section assembly area, pre-outfitting area, painting area, and transfer area. After the upstream tasks are completed, they need to be delivered to the downstream workstations according to the established handover window. The shipyard scheduling system pre-acquires the task set, process sequence, work surface occupancy status, key resource occupancy status, handover relationship, phased delivery nodes, and current execution schedule, and establishes a multi-constraint scheduling model. This model requires each task to simultaneously satisfy process sequence relationships, work surface mutual exclusion relationships, resource capacity relationships such as gantry cranes and transport vehicles, section handover connection relationships, and phased delivery target constraints. In this embodiment, the phased delivery node N1 is defined as the joint control node that is ready to be completed before the stern section completes pre-outfitting, completes transfer, and enters the final group for acceptance. Therefore, the task set corresponding to this node is A5, A6, and A7.

[0042] Seven tasks most closely related to the disturbance were selected as the implementation targets, and the original scheduling data is shown in Table 1. The planned start and end times in the table are expressed using relative reference times, where T0 is the starting point of this stage of scheduling. Task A1 is the assembly of the double bottom section floor plate of the stern, Task A2 is the welding of the longitudinal skeleton, Task A3 is the pre-assembly of the main engine room outfitting components, Task A4 is the pre-treatment of section painting, Task A5 is the section transfer and hoisting, Task A6 is the installation of the compartment piping system, and Task A7 is the preparation for receiving the final assembly. After A1 is completed, it is handed over to A2. After A2 is completed, it is handed over to A3 and A4 respectively. A5 can only be handed over after both A3 and A4 are completed. After A5 is completed, it is handed over to A7. A6 and A3 share the pre-outfitting work surface and compete for the piping system-related auxiliary resources.

[0043]

[0044] During execution, the on-site monitoring system detected an anomaly in the hydraulic circuit of gantry crane G1, causing G1 to be unavailable for 6 consecutive hours in the pre-outfitting and transfer areas. Since A3 and A5 directly occupy G1, they were identified as directly affected tasks. The system further identified related tasks with process integration, resource competition, and work surface sharing relationships with A3 and A5, determining A4 and A6 as related tasks. A4 and A5 have a handover dependency relationship, and A6 shares work surface W2 with A3, potentially leading to work surface conflicts due to A3's postponement. Meanwhile, A7 had entered the downstream receiving preparation stage, with receiving tooling, transportation channels, and work teams all pre-arranged according to the original plan. Adjusting A7 would directly affect on-site organization and subsequent continuous operations; therefore, A7 was designated as a frozen task, maintaining its planned start time, planned completion time, work shifts, and resource allocation unchanged in subsequent adjustments. Figure 2 As can be seen, the original and partially rearranged task periods are compared and displayed on the same timeline. Among them, A7, as the commitment freeze task, always maintains its original scheduling position, which can intuitively reflect the implementation status of the commitment window not being disturbed.

[0045] The system performs disturbance absorption checks on related tasks in the order of reserved buffer time, shift adjustment, and alternative resources. For A4, the system first performs a buffer check, verifying that it has a 2-hour available buffer before A5, and the indirect impact of the G1 failure is only 1 hour. Therefore, A4 can absorb the disturbance by extending the buffer without changing its original handover window and delivery node correspondence, and is set as an observation and frozen task. For A6, simple buffering is insufficient to absorb the W2 work surface conflict caused by the extension of A3. The system continues to perform a shift adjustment check, changing A6 from a single-shift operation to a long day shift and night shift connection operation, and compressing non-critical idle intervals so that it can still be completed within the original local handover window. Therefore, A6 is set as an observation and frozen task. For A3, the system first performs a buffer check, finding that its available buffer is insufficient to fully cover the 6-hour loss. After performing a shift adjustment check, there is still a gap. Subsequently, an alternative resource check is performed, and it is found that the spare small crane G2 in the adjacent area can undertake some light-load lifting operations, thus providing equivalent time compensation for A3. A3 is finally determined to be able to continue execution and is converted into an observation and frozen task. As for A5, it is affected by both the relocation of A3 and the failure of G1. Furthermore, the transfer and hoisting require full load to be used by G1 and R1 in coordination. The alternative resources cannot meet the tonnage requirements of the entire transfer section. Therefore, even after the three-level verification, it still cannot absorb the disturbance, and A5 is included in the rearrangement scope. Figure 2 The different colors used to distinguish between observation freeze tasks, commitment freeze tasks, and rearrangement tasks in this embodiment can be used to explain that A3, A4, and A6 are retained in the freeze set, while A5 is identified separately as a local rearrangement object, thus reflecting the technical idea of ​​narrowing the rearrangement scope.

[0046] The disturbance absorption margin is determined by the following formula:

[0047]

[0048] in, For the task The disturbance absorption margin, For the task The latest allowed start time, For the task The original planned start time, The effective working hours that can be additionally compensated through shift adjustments. The equivalent time that can be compensated by alternative resources. For the time loss caused by the disturbance, and This is the conversion factor. In this embodiment, we take... , When calculating A3, the system, based on the original handover window of A5, determined that the latest allowed start time for A3 is T0+36h, while the original planned start time was T0+32h. Therefore... A3 can compensate by partially extending the number of shifts. Compensation can be achieved using the backup small crane G2. The disturbance loss is Substituting, we get:

[0049]

[0050]

[0051] because This indicates that A3 can absorb disturbances without changing the process sequence, handover relationship, and delivery node. Therefore, A3 is not included in the rearrangement scope, and only its execution time and some hoisting resource configuration are allowed to be adjusted. Figure 3 The cumulative contribution trajectory shows that after adding buffer, shift compensation and alternative resource compensation, although the 6h disturbance loss was deducted, the final value of A3 is still in the positive range, with a corresponding final value of 0.65h, indicating that the task has self-absorption capability.

[0052] The same calculation is performed for A5. Based on the A7 receive preparation window, the system calculates that the latest allowed start time for A5 is T0+46h, while the originally planned start time was T0+45h. Therefore… A5 is a heavy-load, full-distance transfer service; adjustments to the schedule can only provide... There was no equivalent heavy-duty alternative crane on site, therefore The direct loss from the disturbance is Substituting, we get:

[0053]

[0054]

[0055] because This indicates that A5 cannot absorb the disturbance on its own while maintaining the existing process sequence, handover relationships, and delivery nodes. Therefore, A5 is included in this round of reordering. At this point, the tasks not included in the reordering are A3, A4, A6, and A7, with A7 being a committed frozen task, and A3, A4, and A6 being observation frozen tasks. During subsequent review, the system found that A3 no longer creates new constraints on A5 after resource replacement, and that unfreezing and re-participating in optimization would reignite competition on the W2 workbench. Therefore, A3 will remain in the observation frozen state. Figure 3 The cumulative contribution trajectory of A5 is also given, showing that in the absence of equivalent alternative resources, the final value of this task falls into the negative region, eventually reaching -4.2h, thus contrasting with A3.

[0056] With only a partial rescheduling of A5, the system maintains the commitment to freeze A7 unchanged, while verifying the compatibility between A5 and A3, A4, and A6. After partial optimization, A5, which previously relied on a single G1 for whole-segment hoisting, is now adjusted to first complete segmented temporary positioning preparation, and then continuously transfer with R1 after G1 is restored. Furthermore, by reducing the waiting interval before transfer and optimizing the hoisting path, the actual pure operation time of A5 is reduced from 6 hours to 5.5 hours. The completion time of A3 is updated to T0+44.5 hours, A4 to T0+42 hours, and A6 to T0+45 hours. A5 can start as early as after A3 and A4 are completed and complete at T0+50 hours. A7 still follows the original commitment schedule, starting at T0+52 hours and completing at T0+58 hours. For A4 and A6, the system only makes minor adjustments to their execution time periods and shifts, without changing the original correspondence between handover windows and delivery nodes. Figure 2 The updated time period results show that although A3, A4, and A6 have undergone local adjustments, they have not deviated from their original logical positions. A5 was limited to a small range of corrective rearrangement and did not trigger a comprehensive reconstruction of the remaining tasks.

[0057] Delivery node deviations are determined using the following formula:

[0058]

[0059] in, For delivery nodes deviation, For delivery nodes The completion time of the relevant tasks, For delivery nodes The required completion time To deliver the node The corresponding task set. The updated completion time is substituted into the joint control node N1 in this embodiment, where... , ,get:

[0060]

[0061]

[0062] because This indicates that the phased delivery nodes were not disrupted, and the established delivery requirements were still met after partial rearrangement. Figure 4 The updated completion status of A5, A6, and A7 is given in the form of a joint control node margin bubble matrix. The horizontal position represents the updated completion time, the vertical position represents the remaining margin relative to the joint control node N1, and the bubble size represents the task duration. The remaining margins of A5, A6, and A7 relative to the joint control node N1 are 8h, 13h, and 0h, respectively. A7 is located at the critical boundary but has not exceeded the boundary. A5 and A6 are both located before the node, indicating that the set of critical tasks is still within the controllable range after local rearrangement.

[0063] After detecting a crane malfunction disturbance, the system did not refactor all tasks. Instead, it first narrowed down the reordering scope by absorbing the disturbance through buffering, shift adjustments, and verification of alternative resources. Then, it made local repair adjustments only to A5, which could not absorb the disturbance. At the same time, it kept the promised frozen tasks completely unchanged and maintained the original logical position of the observed frozen tasks, allowing only local changes in their execution time, execution shift, or execution resources.

[0064] Example 2:

[0065] In this embodiment, a large coastal shipyard is building a 16,000 TEU container ship, currently in the dockside outfitting and partial painting transition phase after the overall assembly is completed. This phase involves multiple shared work areas, including the hatch coaming area, the upper engine room area, the starboard dockside work area, and the dockside logistics channel, requiring continuous coordination of gantry cranes, flatbed trucks, electrical installation teams, outfitting teams, painting teams, and inspection teams. The shipyard's scheduling system pre-acquires task sets, process sequences, work area status, key resource occupancy status, handover relationships, and phased delivery nodes, forming an original execution schedule. Among these, the starboard piping connection task and the pre-docking joint acceptance preparation task have entered the on-site commitment window, and their execution status cannot be changed; therefore, they are managed as frozen tasks.

[0066] To illustrate the dynamic scheduling process, eight tasks directly related to this disturbance were selected as implementation targets. These include: B1 (dock section replenishment and release), B2 (engine room section precision positioning welding), B3 (engine room main cable tray installation), B4 (upper deck outfitting bracket installation), B5 (funnel section painting and repair), B6 ​​(starboard side piping connection), B7 (dock logistics channel clearing and temporary tooling transfer), and B8 (preparation for joint acceptance before undocking). After B1 is completed, it is handed over to B2; after B2 is completed, it is handed over to B3 and B4 respectively; after B3 and B4 are completed, the upper engine room area and hatch coaming shared surface are released respectively; B5 and B6 share the starboard side work zone; and B8 relies on the closed-loop preparation chain formed by B2, B3, B4, and B6. In the original schedule, B1, B2, B3, B4, B5, B7, B6, and B8 were to proceed sequentially, but B7 was originally planned as a follow-up support task.

[0067] During execution, a positioning deviation occurred during the upstream section repositioning operation, causing a delay in the release of B1. Simultaneously, the shipyard temporarily received an emergency task to install a monitoring module, occupying the main logistics channel at the dock and a shared working area on the side. As a result, B2 could not proceed with the precision tack welding in its original order, B3 and B4 lost stable upstream handover conditions, the starboard working area occupied by B5 potentially conflicted with the frozen task B6, and the originally scheduled channel clearing task B7 was prematurely exposed as a bottleneck. The system further determined that B1 was in the deviation repair phase, and its work content was limited by the on-site positioning status, making it unsuitable for reordering or changing the sequence. Therefore, it was not considered an adjustable task in this round, but only participated in the calculation as a constraint source for subsequent task status updates. Accordingly, the adjustable candidate tasks for this round were determined to be B2, B3, B4, B5, and B7.

[0068] The initial status data for candidate tasks are shown in Table 2. The number of blocked handover relationships in the table indicates the number of effective upstream-to-downstream connections that can be restored once the task is prioritized for adjustment. The number of shared workfaces released indicates the number of newly released job spaces after the task adjustment. Whether the frozen task execution status is changed indicates whether the task adjustment will encroach on the locked resource windows of B6 or B8. Whether the delivery node is affected indicates whether prioritizing the task in its current state will compress the phased delivery buffer.

[0069]

[0070] The system uses a deblocking capability score to initially rank candidate tasks. The scoring formula is as follows:

[0071]

[0072] in, For the task The de-resistance rating, For the task The number of obstructed transfer relationships restored after adjustment For the task The number of newly released shared workfaces after the adjustment. Indicates task Whether to change the execution status of frozen tasks during adjustment; if so, set to 1; otherwise, set to 0. Indicates task The value is 1 if the adjustment affects the delivery schedule, and 0 otherwise. This example uses... , , , The four main candidate tasks in the first round are calculated. For B7, there are... , , , Substituting, we get:

[0073]

[0074] For B4, there is , , , Substituting, we get:

[0075]

[0076] For B2, there is , , , Substituting, we get:

[0077]

[0078] For B5, there is , , , Substituting, we get:

[0079]

[0080] Therefore, B7 received the highest score and was designated as the first priority adjustment task. For B4, B2, and B5, which all scored 11 points, the system was further sorted according to stability constraints. B5 was ranked first due to... This will change the execution status of frozen task B6, therefore it is placed later. Neither B4 nor B2 changes the execution status of frozen tasks, but if B2 is prioritized, it will compress the phased delivery buffer, therefore... B4 does not affect the delivery timeline, therefore B4 is prioritized before B2. This results in the first round of adjustments, with the order being B7, B4, B2, and B5. Figure 5The initial deblocking capability ranking of candidate tasks is given, with B7 scoring 14, B4, B2, and B5 all scoring 11, and B3 scoring 4. In addition, the figure further distinguishes between situations that affect frozen tasks and those that compress node buffers, indicating that although B4, B2, and B5 have the same value, the system still needs to perform a secondary ranking by adding the stability of frozen tasks and the security of delivery nodes.

[0081] The system first executed B7, advancing the clearing of the originally scheduled logistics channel and the transfer of temporary tooling. The temporary storage location for the emergency module was moved from the main dockside channel to the auxiliary work area. Simultaneously, a short-haul transport vehicle was reassigned for the recovery of enclosed tooling. After this adjustment, the main dockside logistics channel was reopened, and one new shared work surface was released around the hatch coaming, improving the access conditions for the previously blocked engine room cable trays and deck brackets. The system then updated the handover status and work surface occupancy status in real time and reassessed the unblocking capability of the remaining tasks.

[0082] Dynamic updates are performed using the following formula:

[0083]

[0084]

[0085] in, Indicates task In the Wheel relative to the first The state gain of the wheel, Indicates the updated dynamic priority. For the transition gain coefficient, this embodiment takes... For B3, there was a... , Therefore, its original score was:

[0086]

[0087] After the B7 was adjusted, the B3 was able to restore three blocked handover connections, instead of just one, and could also release one additional shared work surface. , Therefore, its updated rating is:

[0088]

[0089] Its state gain is:

[0090]

[0091] Its dynamic priority is:

[0092]

[0093] For B4, there are two [items] before and after the first round. , ,as well as , The updated rating will be:

[0094]

[0095] Its state gain is:

[0096]

[0097] Its dynamic priority is:

[0098]

[0099] For B2, due to logistical disruptions and incomplete upstream supply before the first round, its early implementation would compress the phased delivery buffer, therefore... After B7 completed its adjustments, the passageway was restored, the hatch coaming was released, and B3 and B4 both achieved more stable entry conditions. If B2 were to enter later in the second round of implementation, the phased delivery buffer would no longer be compressed; therefore, it was updated to... Meanwhile, the number of recoverable blocked handover relationships and the number of releasable working faces in B2 remain at [value missing]. , Therefore, its updated rating is:

[0100]

[0101] because Its dynamic priority is:

[0102]

[0103] Therefore, B3 satisfies... B3 is marked as a sequential transition task, and its dynamic priority 21 is significantly higher than B4 and B2. The system will insert B3 at the beginning of the current adjustment sequence, and the new adjustment order will be B3, B4, B2, B5. Figure 6 The sorting process is shown by prioritizing the transition trajectory. It can be seen that B7 is executed first and then no longer participates in the subsequent sorting. B3 jumps from the last position to the first position after B7 completes channel deblocking. The sorting is dynamically updated from B7, B4, B2, B5 in the first round to B3, B4, B2, B5 in the second round, which conforms to the dynamic control logic of deblocking and then re-evaluating, and then inserting forward.

[0104] To further illustrate the status changes of the remaining critical tasks after the first round of adjustments, Table 3 summarizes the number of hindered handover relationships, the number of shared workfaces released, the impact of delivery nodes on status, status gains, and dynamic priorities of B3, B4, and B2 before and after the first round. Since B7 has already been completed as a priority task in the first round and will no longer participate in subsequent rankings, and B5 remains in a lower position in the second round, Table 3 focuses on reflecting the changes in critical tasks that have a decisive impact on subsequent rankings.

[0105]

[0106] In the second round, the system prioritized the previously deployed B3, adjusting the engine room main cable tray installation from a single-sided advance method to a double-end opposing advance method. A group of electrical installation personnel who were previously waiting were preemptively moved to the upper engine room work zone, allowing the engine room outfitting chain to be restored first. After the adjustment of B3, B4 obtained a more stable upstream interface, and the system continued to execute B4 according to the updated sequence, ensuring the simultaneous completion of the upper deck bracket installation and hatch coaming area release. Subsequently, B2 entered the precision positioning welding without compromising the phased delivery buffer. The value has been updated to 0. B5 is kept at the end of the sequence and will be adjusted after frozen task B6 completes its critical interface work, thus preventing disturbances from spreading to frozen tasks. Throughout the process, the frozen status of B6 and B8 remains unchanged, and there are no instances of rescheduling that require migration of work teams, changes in commitment windows, or reduction of resources in the acceptance preparation area. Figure 7 The updated integrated status of B3, B4, and B2 is further presented from a multi-dimensional capability perspective, including dimensions such as updated hindered handover recovery capability, shared working surface release capability, status gain, updated score, and dynamic priority. Figure 7 It is evident that B3 excels most in three dimensions: hindered handover recovery capability, state gain, and dynamic priority. B4 shows a balanced improvement in shared workspace release capability and overall score, corresponding to While B2 doesn't excel in state gain, its shared working surface release capability remains high, and due to the removal of constraints related to delivery nodes, its score improves to 13 after the update. .

[0107] After two rounds of dynamic iteration, the previously blocked critical handover chain was restored by the overall conditions after B1 was released, unblocked via channel B7, then restored via bridge B3 and released via bracket B4, forming a continuous and executable link again. Regarding shared work surfaces, the main dockside logistics channel, hatch coaming area, and upper engine room work zone were restored sequentially, while the starboard quay work zone was delayed without affecting frozen task B6. The updated core execution sequence, originally B1, B2, B3, B4, B5, B7, B6, B8, has been optimized to prioritize B7 after B1 is released, followed by B3, B4, and B2, with B5 scheduled based on the frozen task status. B6 and B8 remain unchanged.

[0108] In the context of outfitting and final assembly at a container ship terminal, the system first determines the initial adjustment sequence based on unblocking capabilities. Then, after each task adjustment is completed, the unblocking capabilities of the remaining tasks are reassessed. This allows for the timely identification of tasks with sequential jumps caused by local improvements, prioritizing their execution. Scheduling adjustments prioritize tasks that can establish critical handover chains and release shared work surfaces. Furthermore, among tasks of the same priority, tasks that do not alter the execution status of frozen tasks or affect delivery milestones are selected first, effectively preventing disturbances from spreading to frozen tasks and interim delivery nodes. Ultimately, without implementing a global rescheduling, the system achieves prioritized restoration of critical handover chains, continuous release of shared work surfaces, and dynamic updates of the adjustment sequence based on constraint status, demonstrating high on-site responsiveness and global coordination efficiency.

Claims

1. A shipbuilding multi-constrained scheduling optimization method, characterized by, include: Obtain information on shipbuilding task sets and process sequence, work surface status, key resource occupancy status, handover relationships, delivery milestones, and execution schedules; When a disturbance event is detected, the affected tasks are identified; disturbance absorption verification is performed on the tasks associated with the affected tasks, and it is then determined whether they are included in the reordering scope based on preset diffusion conditions. Tasks not included in the reordering scope are set as frozen tasks; within the reordering scope, scheduling adjustments are performed according to a preset adjustment order, and an updated schedule is generated and output.

2. The method of claim 1, wherein, The disturbance absorption verification is used to determine whether associated tasks can continue to be executed without changing the process sequence, handover relationship and delivery node by utilizing reserved buffer periods, shift adjustments or alternative resources; associated tasks that can continue to be executed are not included in the rearrangement scope and are set as frozen tasks.

3. The method of claim 1, wherein, The frozen tasks include committed frozen tasks and observation frozen tasks; committed frozen tasks are those that have entered the commencement commitment window or whose downstream tasks have completed the preparation for receipt, while observation frozen tasks are those that are temporarily not included in the rescheduling scope after disturbance absorption verification; committed frozen tasks remain unchanged when the scheduling adjustment is performed; when it is re-determined that the observation frozen task does not change the delivery node, it is unfrozen and transferred to the rescheduling scope.

4. The method of claim 1, wherein, The preset adjustment order is determined according to the task's unblocking capability; the unblocking capability is determined based on the number of blocked handover relationships restored after the task adjustment and the number of newly released shared workfaces; when performing scheduling adjustments, tasks that can simultaneously restore multiple blocked handover relationships or release shared workfaces are given priority.

5. The method of claim 2, wherein, The disturbance absorption verification is performed sequentially according to the reserved buffer period, shift adjustment, and alternative resources. When the previous absorption method meets the conditions for the related task to continue execution, the verification of the next absorption method is terminated, and the related task is set as a frozen task.

6. The method of claim 2, wherein, After setting the associated task as a frozen task, the original correspondence between the handover window and the delivery node remains unchanged, and only the execution time period, execution shift, or execution resources are adjusted.

7. The method of claim 4, wherein, The unblocking capability is classified according to the number of blocked handover relationships restored and the number of shared working surfaces released after task adjustment, and tasks that simultaneously satisfy the requirements of restoring blocked handover relationships and releasing shared working surfaces are identified as priority adjustment tasks.

8. The method of claim 4, wherein, When multiple tasks have the same deblocking capability, priority is given to adjusting the task that does not change the execution status of the frozen task; if they are still the same, priority is given to adjusting the task that does not change the delivery node.

9. The method of claim 4, wherein, After each task's scheduling adjustment is completed, the unblocking capability of the remaining tasks is reassessed; when the unblocking capability of subsequent tasks changes due to the adjustment of the previous task, the preset adjustment order is updated according to the reassessment result.

10. A multi-constrained scheduling optimization method for shipbuilding according to claim 9, characterized in that, When reassessing the unblocking capability of the remaining tasks, tasks that have changed from a blocked state to a transferable state or from a shared work surface occupied state to a release state due to the previous task adjustment are marked as sequential jump tasks, and the sequential jump tasks are inserted at the beginning of the current preset adjustment order for priority execution.

Images