A method and system for intelligent scheduling of project tasks based on limited resources
By identifying the resource disturbance impact domain through the scheduling server, freezing completed tasks and historical resource occupancy, and forming the earliest feasible and minimum offset insertion scheme, the problem of task scheduling inconsistency caused by resource disturbance in projects with limited resources is solved, and the local repair of the scheduling scheme and the consistency of the execution end state are achieved.
Patent Information
- Application Number
- CN202611279053.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-21
- Publication Date
- 2026-11-17
AI Technical Summary
In the execution of projects with limited resources, when resource disturbances cause the task scheduling scheme to need to be readjusted, existing technologies are unable to effectively limit the scope of tasks that need to be rescheduled, resulting in duplicate resource occupation, omission of task dependencies, difficulty in controlling the scope of plan changes, and inconsistency in the status of different execution ends.
The scheduling server acquires task data and resource status, forms a baseline scheduling scheme and generates a snapshot, identifies the local impact domain of resource disturbances, freezes completed tasks and historical resource occupancy, forms the earliest feasible and minimum offset insertion schemes based on task dependencies and resource availability time intervals, compares and adjusts the schemes step by step, generates a local target scheduling scheme, and switches the scheduling version to maintain consistency after confirmation at the receiving end.
It enables the reasonable limitation of the scope of rescheduled tasks under resource disturbance, maintains the continuity of completed tasks and resource usage, reduces redundant resource allocation and task dependency omissions, and ensures scheduling consistency across all execution ends.
Smart Images

Figure CN123117633A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer project management and resource scheduling technology, and in particular to an intelligent scheduling method and system for project tasks based on limited resources. Background Technology
[0002] As projects such as engineering construction, equipment maintenance, manufacturing, and technological research and development expand in scale, project implementation typically involves a large number of interdependent tasks. It is necessary to determine the task execution sequence, execution time, and resource allocation relationships within limited equipment, personnel, and materials. Furthermore, there are interrelationships between equipment capacity, personnel skills, material inventory, and resource availability, meaning that the project task scheduling results directly impact the project schedule and resource utilization.
[0003] Existing project scheduling methods typically generate a baseline plan based on task durations, dependencies, and resource availability obtained during the project initiation phase, with the total project duration or resource balance as the primary optimization objective. Manual scheduling and standard project management software can handle basic task ordering and resource allocation, but as the number of tasks, resource types, or multiple tasks sharing the same resource increase, problems such as duplicate resource usage, missing task dependencies, and difficulty in controlling the scope of plan adjustments can easily arise. Using genetic algorithms, heuristic algorithms, or constraint-solving methods can improve the ability to solve complex scheduling problems, but these methods usually focus on generating a complete scheduling scheme based on given task and resource conditions.
[0004] Invention patent CN110991917A discloses a multi-mode resource-constrained project scheduling optimization method using a two-stage genetic algorithm. This method obtains project tasks, task temporal relationships, task execution modes, and resource availability. It determines the task execution mode and scheduling order through a two-stage genetic search and outputs the scheduling optimization results. This method can handle various task execution modes and constraints on both updatable and non-updatable resources. However, it primarily focuses on overall scheduling optimization under given project data and does not further address resource disturbance handling during the project execution phase. When resource disturbances such as equipment failures change existing resource availability and task resource bindings, directly re-executing the overall scheduling optimization requires handling a large range of tasks. It is difficult to limit the local impact of the disturbance and to transform the existing execution intervals and resource occupancy of unaffected tasks into fixed boundaries during the local recovery process. Therefore, the regenerated scheduling results may alter the arrangements of many tasks that have already been issued or are about to be executed.
[0005] Existing technologies have also proposed event-based reactive project scheduling methods for resource unreliability scenarios. These methods represent resource interruptions as scheduling events, establish recovery models after disturbances occur, and generate revised plans. Such methods can handle single or multiple resource disturbances and reduce the number of variables and solution complexity in reactive scheduling models. However, these methods primarily focus on constructing and solving recovery models under resource disturbance conditions, lacking a continuous processing relationship regarding how the actual resource state, task execution state, and existing scheduling records in the engineering system jointly define the scope of local repair. Especially in projects where equipment, personnel, and spare parts are all constrained, simply resolving the plan based on the disturbance event may still include tasks with no task dependency on the disturbed resource within the adjustment scope, and it is difficult to maintain the existing arrangements for completed tasks, resource occupancy, and other unaffected tasks.
[0006] Furthermore, existing scheduling systems typically output recalculated solutions as Gantt charts, reports, or adjustment suggestions, lacking clear limitations on the scope of task changes between the old and new solutions. If a local recovery plan is not feasible, the system can easily switch directly to a global rescheduling, increasing the calculation scope and expanding plan changes. When scheduling results are sent to task terminals, equipment control terminals, personnel terminals, and warehouse management terminals, different receiving terminals may experience inconsistent execution versions due to changes in resource status or asynchronous instruction reception; the task and resource statuses generated by the execution terminals may also not be promptly fed back to the next scheduling cycle, causing the calculation results to gradually deviate from the actual resource status.
[0007] Therefore, how to reasonably limit the scope of tasks that need to be rescheduled based on actual resource usage and task dependencies when resource disturbances occur during the execution of a limited resource project, maintain the continuity of effective task arrangement and existing resource usage, control the scope of plan changes under the condition of meeting the constraints of equipment, personnel, materials and task timing, and ensure that the recovery plan is consistent with the actual state of each execution end, is a technical problem that needs to be solved in the dynamic scheduling of tasks in limited resource projects. Summary of the Invention
[0008] The main objective of this invention is to provide a project task intelligent scheduling method and system based on limited resources. This method addresses the problem of limiting the scope of tasks that need to be rescheduled based on the actual execution status of tasks, resource occupancy status, and task dependencies when resource disturbances occur during project execution, the available time interval of existing resources, and task resource binding changes. It also addresses the problem of partially repairing tasks affected by resource disturbances while keeping completed tasks, tasks outside the local impact domain, and historical resource occupancy unchanged.
[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an intelligent scheduling method for project tasks based on limited resources, comprising the following steps: S1. The scheduling server obtains task data and resource status containing duration and task dependencies, forms task constraints and baseline scheduling scheme, and generates a baseline scheduling snapshot that records task execution interval, resource binding and resource consumption based on the baseline scheduling scheme. S2. Receive resource disturbance events containing the target resource identifier, disturbance time, and expected recovery time. Use the period from the disturbance time to the expected recovery time as the disabled time interval, update the available time interval of the target resource, obtain the task status and remaining duration at the disturbance time, identify unfinished tasks bound to the target resource in the baseline scheduling scheme and whose execution interval overlaps with the disabled time interval, as well as unfinished successor tasks determined along the task dependencies, form a local influence domain, fix tasks outside the local influence domain, completed tasks, and historical resource occupancy, and determine the tasks to be repaired and freeze constraints. S3. Remove the execution schedule after the disturbance of the task to be repaired, determine the insertion duration based on the task status, duration and remaining duration, and insert the task to be repaired into the candidate execution interval that satisfies the task constraints, resource available time interval and freeze constraints according to the task dependency, forming the earliest feasible insertion scheme and the minimum offset insertion scheme. S4. Compare the earliest feasible insertion scheme and the smallest offset insertion scheme in the order of the number of tasks to be adjusted, the total offset at the start time, and the increment of the total project duration to determine the local target scheduling scheme. If no local target scheduling scheme is formed, bounded retries are performed according to the updated recovery time. If no local target scheduling scheme is formed after bounded retries, global rescheduling is performed on the unfinished tasks. When a feasible recovery scheme is formed, a target scheduling scheme and target change record are generated. When no feasible recovery scheme is formed, a scheduling exception instruction is generated. S5. When a solution is found, send adjustment instructions and resource control data based on the target scheduling scheme and target change records to the task terminal and resource terminal. After confirmation by the receiving end, switch the scheduling version and control the resource terminal to update the job sequence, task assignment and resource reservation. When no solution is found, send the scheduling exception instruction to the project management terminal, receive execution feedback, update the task and resource status, and form the next cycle baseline scheduling snapshot with the effective scheduling scheme.
[0010] In the preferred embodiment, step S1 further includes the following processing.
[0011] The scheduling server obtains the scheduling baseline time from the unified project clock, and retrieves the task identifier, estimated duration, direct predecessor tasks, task resource conditions, earliest allowed start time, latest required finish time, and task failure recovery attributes. The scheduling server establishes task dependencies based on the direct predecessor tasks.
[0012] The scheduling server obtains the resource identifier, resource type, available capacity, available time interval, and capability attributes of limited resources. Limited resources include equipment resources, personnel resources, and spare parts resources; task resource conditions include equipment type and quantity, personnel skills and quantity, and spare parts type and quantity. The scheduling server forms hard constraints based on task dependencies, allowed execution time, available resource capacity, available resource time interval, resource capability attributes, and spare parts inventory.
[0013] The scheduling server generates candidate scheduling schemes based on task dependencies. For candidate scheduling schemes that do not meet hard constraints, it performs task timing adjustments, resource rebinding, or searches for the next available time interval. It then determines the baseline scheduling scheme from among the candidate scheduling schemes that meet the hard constraints. The scheduling server extracts the task start time, planned completion time, and baseline resource bindings from the baseline scheduling scheme, forms a baseline resource occupancy record according to resource identifiers, and stores the baseline scheduling time, baseline scheduling scheme, task data, task dependencies, resource status, hard constraints, task execution interval, baseline resource bindings, and baseline resource occupancy record as a baseline scheduling snapshot.
[0014] In the preferred embodiment, step S2 further includes the validation of equipment failure events and the updating of resource availability time intervals.
[0015] The scheduling server obtains the scheduling baseline time for the current scheduling period, limits the target resource to device resources, and receives the target device identifier, fault code or recovery status code, fault occurrence time, and expected recovery time or actual recovery time through the device management interface. The scheduling server verifies the resource disturbance event based on the status code, fault occurrence time, recovery time, scheduling baseline time, and the correlation between the fault event and the event that was not closed. After successful verification, a valid device fault event is generated.
[0016] The scheduling server defines the time interval from the time of the failure to the time of recovery as the disabled time interval for the target device. It excludes the disabled time interval from the original available time interval of the target device and uses the updated available time interval to form the post-event resource status. The scheduling server obtains the task execution status and remaining execution time from the task terminal at the time of the failure. The task execution status includes not started, in progress, paused, and completed.
[0017] In the preferred embodiment, step S2 further includes determining the local influence domain and forming a freeze constraint.
[0018] The scheduling server reads the tasks occupying the target device, their corresponding planned execution intervals, and resource bindings from the baseline scheduling scheme. Tasks whose resource bindings include the target device, whose execution status is not started, in progress, or paused, and whose planned execution intervals overlap with the target device's disabled time intervals are identified as directly affected tasks.
[0019] The scheduling server uses the directly affected task as the starting task in the task dependency relationship. It then identifies subsequent tasks whose execution status is not started, in progress, or paused as subsequent affected tasks, forming a local influence domain from the directly affected tasks and subsequent affected tasks. The scheduling server fixes the execution records, resource bindings, and resource usage of completed tasks and tasks outside the local influence domain.
[0020] The scheduling server reads the task failure recovery attributes of tasks within the local impact domain from the task data. For running or paused tasks that are not allowed to be migrated for recovery, the scheduling server retains the original resource bindings and historical execution records, and determines the recovery execution interval based on the recovery time and remaining execution duration. For running or paused tasks that are allowed to be migrated for recovery, the scheduling server fixes the historical execution records before the failure occurred, and retains the remaining part of the corresponding task in the scope to be repaired.
[0021] The scheduling server identifies unfixed tasks within the local impact domain as tasks to be repaired and forms freeze constraints based on fixed execution intervals, fixed resource bindings, historical resource usage, recovery execution intervals, target device disabling time intervals, and task constraints. Freeze constraints are used to limit the task execution records, resource bindings, and resource usage boundaries that cannot be changed during the local repair process.
[0022] In the preferred embodiment, step S3 further includes the formation of a partial scheduling scheme and a partial repair of the initial resource state.
[0023] The scheduling server divides the tasks to be repaired into directly affected tasks and subsequently affected tasks. The scheduling server replicates the baseline scheduling scheme and retains the task records, execution intervals, and resource bindings of the fixed tasks in step S2. It removes the planned start time, planned completion time, and resource occupation after the disturbance time of the directly affected tasks and the subsequently affected tasks, while retaining the task identifier, duration, task dependencies, task resource conditions, and historical execution records before the disturbance time, thus forming a partial scheduling scheme.
[0024] The scheduling server determines the insertion order of tasks to be repaired according to task dependencies and retains the source markers of whether the tasks to be repaired belong to directly affected tasks or subsequently affected tasks. The scheduling server excludes resources subject to freeze constraints from the resource availability time interval included in the resource status updated in step S2, and keeps the target resource disabled time interval in an unavailable state, forming the initial resource status for local repair.
[0025] In the preferred embodiment, step S3 further includes a dual-branch repair process involving the earliest feasible insertion and the minimum offset insertion.
[0026] The scheduling server establishes the earliest feasible insertion branch and the minimum offset insertion branch based on the partial scheduling scheme and the initial resource state of partial repair, respectively, and processes the tasks one by one according to the insertion order of the tasks to be repaired. For tasks that have not started, the scheduling server uses the estimated duration as the insertion duration; for tasks that are executing or paused, the scheduling server uses the remaining execution time as the insertion duration.
[0027] The scheduling server defines the initial boundary of the candidate execution interval based on the disturbance time, the earliest allowed start time of the task, and the completion time of the preceding task, and ensures that the duration of the candidate execution interval is consistent with the insertion duration. The scheduling server filters the candidate execution interval and resource bindings based on task resource conditions, the initial resource state of local repair, task constraints, and freeze constraints, ensuring that equipment capacity, personnel skills, resource availability time intervals, and spare parts availability meet the task resource conditions.
[0028] For tasks directly affected, the scheduling server prioritizes searching for candidate execution intervals bound to compatible alternative resources; if no candidate results are found, the scheduling server searches for candidate execution intervals that are rebound to the original target resource after the recovery time. For tasks subsequently affected, the scheduling server prioritizes searching for candidate execution intervals that maintain the binding of the baseline resource; if no candidate results are found, the scheduling server searches for candidate execution intervals bound to other compatible resources.
[0029] The earliest feasible insertion branch selects the candidate execution interval with the earliest start time, while the minimum offset insertion branch selects the candidate execution interval with the smallest offset relative to the baseline plan's start time. After each task insertion is completed, the scheduling server updates the corresponding branch's task execution interval, resource binding, resource availability time interval, and the preceding completion time of subsequent tasks, and uses the updated results to filter the candidate execution interval for the next task to be repaired. After all tasks to be repaired have been inserted, the scheduling server forms corresponding local recovery candidate schemes.
[0030] In the preferred embodiment, step S4 further includes the evaluation of local recovery candidate schemes and the determination of local target scheduling schemes.
[0031] The scheduling server compares the start time, completion time, and resource binding of the task to be repaired in the candidate and baseline scheduling schemes for the earliest feasible insertion scheme and the smallest offset insertion scheme, and determines the number of tasks to be adjusted. The scheduling server determines the total offset of the start time based on the degree of offset of the candidate start time relative to the baseline start time, and determines the increment of the total project duration based on the project completion time of the candidate scheme relative to the project completion time of the baseline scheduling scheme.
[0032] The scheduling server compares candidate schemes in order of adjusting the number of tasks, the total offset at start time, and the increment of the total project duration, and determines the top-ranked candidate scheme as the local target scheduling scheme. When all comparison results are the same, the scheme with the smallest offset is determined as the local target scheduling scheme. The scheduling server retains the task insertion record and the updated resource status corresponding to the local target scheduling scheme. Based on the task insertion record, it determines the tasks to be checked, compares the task start time, completion time, and resource binding in the local target scheduling scheme with those in the baseline scheduling scheme, generates a target change record for the local recovery path, and determines the updated resource status as the target resource status of the local recovery path.
[0033] In the preferred embodiment, step S4 further includes fault recovery time update, bounded local retry, and global rescheduling.
[0034] When no local target scheduling scheme is formed, the scheduling server limits the target resources to device resources, obtains the latest estimated recovery time of the target device through the device status acquisition interface, and updates the resource disturbance event, the target device disabled time interval and resource status. The scheduling server recalculates the remaining execution time and recoverable start time of the affected tasks based on the task execution status, redetermines the local influence domain and the tasks to be repaired based on the updated resource disturbance event, resource status and task execution status, and updates the freeze constraints based on the remaining execution time, recoverable start time and redetermined local influence domain.
[0035] The scheduling server re-executes local repair and candidate solution comparison within the maximum number of local retries. If the maximum number of local retries is reached and a local target scheduling plan has not yet been formed, the scheduling server fixes the execution records of completed tasks and the historical occupancy of tasks in execution or paused, and performs global rescheduling on all incomplete tasks, satisfying constraints such as task dependencies, resource availability time intervals, resource capabilities, personnel skills, and spare parts inventory. When a global recovery scheduling plan is formed, the scheduling server generates a target scheduling plan and target change records; if a global recovery scheduling plan is not formed within the preset scheduling time range, the scheduling server generates a scheduling exception instruction.
[0036] In the preferred embodiment, step S5 further includes pre-writing the scheduling scheme, multi-terminal confirmation, scheduling version switching, and execution feedback processing.
[0037] The scheduling server generates a recovery processing identifier. When forming a target scheduling plan, the scheduling server updates the resource availability time range and resource occupancy status according to the task execution range and resource binding in the target scheduling plan, thus forming the target resource status. The scheduling server creates a pending scheduling version based on the currently effective scheduling version, writes the task execution range and resource binding of the target scheduling plan and the target resource status into the pending scheduling version, and generates task adjustment instructions, equipment work queue update data, personnel task assignment data, and spare parts reservation data, respectively carrying the recovery processing identifier, the pending scheduling version, and the task identifier, based on the target change record.
[0038] The scheduling server sends task adjustment instructions to the corresponding task terminals and sends updated equipment work queue data, personnel task assignment data, and spare parts reservation data to the corresponding resource terminals. The receiving end checks its local status based on the task identifier and the scheduling version to be implemented. If the adjustment can be executed, it locks the corresponding resource and returns confirmation information carrying a recovery processing identifier, the scheduling version to be implemented, the task identifier, and an acceptance status. If the resource status has changed or the instruction version is lower than the locally confirmed version, the receiving end returns confirmation information carrying a recovery processing identifier, the scheduling version to be implemented, the task identifier, and a rejection status.
[0039] The scheduling server checks the confirmation information from all receivers within the confirmation waiting period. If any receiver fails to return confirmation information, the scheduling server resends the corresponding instruction only to those receivers and continues to wait for confirmation information as long as the cumulative number of transmissions is less than the maximum number of transmissions. If all receivers return confirmation information carrying an acceptance status, the scheduling server switches the pending scheduling version to the active scheduling version. If any receiver returns confirmation information carrying a rejection status, or if any receiver fails to return confirmation information after the cumulative number of transmissions reaches the maximum number of transmissions, the scheduling server unlocks locked resources, cancels task and resource records in the pending scheduling version, and continues to use the original active scheduling version. When generating a scheduling exception instruction, the scheduling server sends the scheduling exception instruction to the project management terminal while maintaining the original active scheduling version.
[0040] The scheduling server receives task execution feedback and resource status feedback, carrying a recovery processing identifier and the effective scheduling version. Based on the task execution feedback, it updates the task status, actual execution time, and remaining execution time. Based on the resource status feedback, it updates the resource availability time range, capacity validity status, and spare parts inventory status. When the resource status feedback indicates an equipment malfunction, and the equipment management system confirms that the equipment has stopped operating or its operation has been restricted, the scheduling server generates a resource disturbance event for the next cycle.
[0041] The scheduling server will create a baseline scheduling snapshot for the next cycle by including the effective scheduling scheme, updated task status, updated resource status, actual execution interval, future planned execution interval, resource binding, and resource occupancy records. When generating resource disturbance events for the next cycle, the scheduling server will use these events and the baseline scheduling snapshot for determining the local impact domain and repairing tasks in the next scheduling cycle, ensuring that the execution feedback from the current scheduling cycle becomes the basis for the task and resource status in the next scheduling cycle.
[0042] In a preferred embodiment, the present invention also provides an intelligent scheduling system for project tasks based on limited resources. The system includes a scheduling server, a task terminal, a resource terminal, and a project management terminal. The scheduling server communicates with the task terminal, the resource terminal, and the project management terminal through communication interfaces.
[0043] The scheduling server includes a processor, memory, communication interface, and scheduling database. The memory stores the project task scheduler, and the processor executes the project task scheduler to implement the baseline scheduling module, disturbance handling module, local repair module, recovery decision module, and execution feedback module.
[0044] The baseline scheduling module acquires task data and resource status, generating task constraints, a baseline scheduling scheme, and a baseline scheduling snapshot. The disturbance handling module updates the target resource availability time interval based on resource disturbance events, determining the local impact domain, tasks to be repaired, and freeze constraints. The local repair module removes future execution schedules for tasks to be repaired, generating the earliest feasible insertion scheme and the minimum offset insertion scheme based on task dependencies.
[0045] The recovery decision module determines the target scheduling scheme based on the number of tasks to be adjusted, the total offset at the start time, and the incremental total project duration. It also performs bounded retries or global rescheduling in case of local recovery failure. The execution feedback module sends adjustment instructions and resource control data to the task terminal and resource terminal via a communication interface. It switches the scheduling version based on the confirmation from the receiving end, receives task and resource execution feedback, and generates a baseline scheduling snapshot for the next cycle based on the effective scheduling scheme, task status, and resource status.
[0046] The scheduling database stores task data, resource status, baseline scheduling snapshots, local rescheduling contexts, candidate recovery schemes, scheduling versions, and execution feedback snapshots. The baseline scheduling module, disturbance handling module, local repair module, recovery decision module, and execution feedback module transmit task data, resource status, candidate schemes, scheduling versions, and feedback results for the corresponding scheduling period through the scheduling database.
[0047] This invention provides a project task intelligent scheduling method and system based on limited resources. When the status of equipment, personnel and spare parts comes from different sources and the update time is inconsistent, the available status of each resource and the actual status of the task are associated with the baseline scheduling snapshot that records the task execution interval, resource binding and resource occupation. This can provide the task timing and resource occupation basis under the same scheduling benchmark for disturbance recovery, and reduce the duplication of resource allocation, omission of task dependencies and rewriting of existing occupation.
[0048] After a resource disturbance occurs, directly affected tasks are identified based on the overlap between the target resource binding and the disabled time interval, and unfinished successor tasks are included along the task dependencies. The recovery arrangements for completed tasks, tasks with local impact outside the domain, historically occupied tasks, and tasks that are not allowed to migrate are kept unchanged by the freeze constraints, so that the scope of the re-solution corresponds to the actual impact of the disturbance and reduces irrelevant task changes caused by the overall rescheduling.
[0049] Under the resource constraints of freezing, the earliest feasible insertion scheme and the minimum offset insertion scheme are formed respectively. The local target scheduling scheme is determined step by step according to the number of tasks to be adjusted, the total offset at the start time, and the increment of the total project duration. This can control the plan changes while meeting the resource conditions. After a local recovery fails, bounded retries are performed according to the updated recovery time. After the retry boundary is reached, the unfinished tasks are rescheduled globally, taking into account both recovery timeliness and the stability of the original plan.
[0050] The target scheduling scheme applies task adjustment instructions and resource control data to each receiving end. The task execution interval, equipment operation queue, personnel task assignment, and spare parts reservation are controlled by the pending scheduling version and confirmation results. When confirmation times out, the corresponding instruction is resent to the unconfirmed receiving end. If the receiving end rejects the instruction or fails to confirm after reaching the maximum number of resentments, the pending record is revoked and the original effective scheduling version is maintained. The actual task status and resource status are then written into the next cycle's baseline scheduling snapshot, reducing the possibility of inconsistencies between receiving ends and subsequent scheduling data deviating from the actual operating status. Attached Figure Description
[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of an intelligent scheduling method for project tasks based on limited resources according to the present invention; Figure 2 This is a schematic diagram of an intelligent scheduling system for project tasks based on limited resources according to the present invention. Detailed Implementation
[0052] The technical solution of this application will be further explained below with reference to an implementation method for scheduling maintenance projects under multi-device sharing conditions. In this implementation method, the project task can be an equipment maintenance task, an engineering construction task, or a manufacturing task; the limited resources include equipment resources that can be actually occupied by the task, personnel resources with corresponding skill qualifications, and spare parts resources required to complete the task.
[0053] Example 1 like Figures 1 to 2 As shown, this invention provides an intelligent scheduling method for project tasks based on limited resources, comprising the following steps: S1. The scheduling server obtains task data and resource status containing duration and task dependencies, forms task constraints and baseline scheduling scheme, and generates a baseline scheduling snapshot that records task execution interval, resource binding and resource consumption based on the baseline scheduling scheme. S2. Receive resource disturbance events containing the target resource identifier, disturbance time, and expected recovery time. Use the period from the disturbance time to the expected recovery time as the disabled time interval, update the available time interval of the target resource, obtain the task status and remaining duration at the disturbance time, identify unfinished tasks bound to the target resource in the baseline scheduling scheme and whose execution interval overlaps with the disabled time interval, as well as unfinished successor tasks determined along the task dependencies, form a local influence domain, fix tasks outside the local influence domain, completed tasks, and historical resource occupancy, and determine the tasks to be repaired and freeze constraints. S3. Remove the execution schedule after the disturbance of the task to be repaired, determine the insertion duration based on the task status, duration and remaining duration, and insert the task to be repaired into the candidate execution interval that satisfies the task constraints, resource available time interval and freeze constraints according to the task dependency, forming the earliest feasible insertion scheme and the minimum offset insertion scheme. S4. Compare the earliest feasible insertion scheme and the smallest offset insertion scheme in the order of the number of tasks to be adjusted, the total offset at the start time, and the increment of the total project duration to determine the local target scheduling scheme. If no local target scheduling scheme is formed, bounded retries are performed according to the updated recovery time. If no local target scheduling scheme is formed after bounded retries, global rescheduling is performed on the unfinished tasks. When a feasible recovery scheme is formed, a target scheduling scheme and target change record are generated. When no feasible recovery scheme is formed, a scheduling exception instruction is generated. S5. When a solution is found, send adjustment instructions and resource control data based on the target scheduling scheme and target change records to the task terminal and resource terminal. After confirmation by the receiving end, switch the scheduling version and control the resource terminal to update the job sequence, task assignment and resource reservation. When no solution is found, send the scheduling exception instruction to the project management terminal, receive execution feedback, update the task and resource status, and form the next cycle baseline scheduling snapshot with the effective scheduling scheme.
[0054] In the preferred embodiment, step S1 specifically includes the following steps: S11. Synchronization and Constrained Representation of Task Data and Limited Resource Status The scheduling server reads the current time from the unified project clock and sets the read current time as the base time for this scheduling. and at the scheduling reference time The project management interface, equipment management interface, personnel management interface, and warehouse management interface are called respectively; the scheduling server uses the timestamps carried in the data returned by each interface as the basis for status synchronization; for cases where there are multiple status records for the same data object, the scheduling server selects a timestamp no later than the base time. And distance from the reference time The most recent record is used as the valid state for this scheduling, so that task data and resource state correspond to the same scheduling benchmark; The scheduling server will be the first Each task to be scheduled is represented as a task data object: (1); in, As a task identifier; The estimated duration of the mission; This is the set of tasks that are directly preceding this task; The set of resources required for the task includes at least the type and quantity of equipment, the skills and number of personnel, and the type and quantity of spare parts. The earliest allowed start time for the task; The latest time required to complete the task; This is a task failure recovery attribute, used to indicate whether a task in execution is allowed to migrate to a compatible resource to continue execution after being forced to stop due to a failure of its bound resource. This indicates that migration and recovery are permitted. This indicates that the original resource bindings need to be retained and execution will continue after the original resources are restored; the scheduling server will aggregate all tasks to be scheduled into a task set. And based on the set of direct predecessor tasks for each task. Establish a directed graph of task dependencies , where the set of nodes With task set The tasks in the text correspond one-to-one; when the task It belongs to the task direct prerequisite task set At that time, at the edge set Establish a task Pointing to task The directed edges; The scheduling server will be the first A finite resource is represented as a resource state object: (2); in, For resource identification; The resource type includes at least equipment resources, personnel resources, and spare parts resources; For resources at the reference time Available capacity; A set of time intervals in which resources are available; For resource capabilities attributes; for equipment resources, At a minimum, this includes the equipment model and the type of work that can be performed; regarding personnel resources, At least skill category and skill level should be included; for spare parts resources, At least the spare parts model and compatible equipment type should be included; the scheduling server will aggregate all resource status objects into an initial resource status set. ; The scheduling server based on the task set Task Dependency Directed Graph and initial resource state set Generate hard constraint set Hard constraint set At a minimum, this includes: the start time of a subsequent task must not be earlier than the completion time of its direct predecessor task; and the completion time of a task must not be later than its latest completion time. The total demand for the same type of resources allocated to each task within the same time interval shall not exceed the available capacity of the corresponding resources; the equipment models and personnel skills allocated to the task must meet the resource requirements of that task. The task execution interval must fall within the set of available time intervals of the bound resources. The number of spare parts required for the task must not exceed the available inventory of the corresponding spare parts at the start of the task.
[0055] S12. Generation of feasible scheduling schemes based on genetic search and hard constraint repair The scheduling server uses a directed graph based on task dependencies. For task sets Perform a topological sort and generate the results without changing the order constraint that predecessor tasks precede successor tasks. The tasks are arranged in a sequence; for each task sequence, the scheduling server determines the task based on its resource condition set. From the initial resource state set Filtering ability attributes With resource condition set Matching resources form the initial resource binding for the corresponding task; the task arrangement and the initial resource binding together constitute the first... Chromosomes ,in, For the first The task arrangement corresponding to each chromosome. A sequence of bindings between tasks and resources; The scheduling server schedules each chromosome Perform serial decoding; decode according to task order. Read the tasks sequentially, and determine their estimated durations. Scheduling reference time The earliest allowed start time Direct Pre-Task Set Completion time and resource binding sequence of each task and the set of available time intervals for the corresponding resources. The candidate start time and candidate completion time of the task are determined; the candidate start time must not be earlier than the scheduling reference time. and the earliest allowed start time The candidate's completion time must not be later than the latest completion time. This forms a relationship with chromosomes. Corresponding candidate scheduling schemes ; After completing an initialization, crossover, or mutation operation, the scheduling server calls the hard constraint set. Candidate scheduling schemes Perform constraint checks; if a candidate scheduling scheme has task prerequisite conflicts, the scheduling server will postpone the start time of the candidate conflicting task until all its directly preceding tasks are completed; if there are resource time conflicts, the scheduling server will proceed only after the task resource conditions are met. Search for the next available resource or the next available time interval from the available resources; if there is a mismatch in equipment model, personnel skills, or spare parts quantity, revoke the binding of the corresponding resource and select one that matches the capability attributes. and available capacity Re-establish bindings in the resources; if the repaired completion time is later than the latest completion time. Then continue searching for resources that meet the resource requirements and can be completed at the latest possible time. The available interval for completing the task; candidate scheduling schemes that still do not meet any hard constraints after the above repairs are eliminated. For candidate scheduling schemes that have completed constraint repair The scheduling server calculates the total project duration: (3); in, For the task In candidate scheduling schemes The completion time in the middle; The scheduling reference time generated in sub-step S11; Candidate scheduling schemes The total project duration is calculated from the scheduling baseline time; the scheduling server uses the total project duration as the basis. Perform the selection operation and follow the crossover probability. and mutation probability Update the task arrangement and resource binding sequence; new chromosomes generated by crossover and mutation continue to undergo decoding and hard constraint repair; when the maximum number of iterations is reached... At that time, terminate the genetic search; The scheduling server will select the set of elements that satisfy the hard constraints during the genetic search process. The candidate scheduling schemes are summarized into a set of feasible candidate schemes. And associate the total project duration with each candidate scheduling scheme. The total project duration is formed by the total project duration corresponding to all feasible candidate solutions. Set of feasible candidate solutions Arranged in ascending order of total project duration; if total project durations are the same, arranged according to the generation order of the corresponding candidate scheduling schemes; preset parameters. , , and These are used to control the initial candidate size, genetic search termination condition, crossover operation trigger condition, and mutation operation trigger condition, respectively, and their control functions are completed within this step.
[0056] S13. Determining the baseline scheduling scheme and generating resource usage snapshots. The scheduling server reads the set of feasible candidate solutions. and the total project duration When the set of feasible candidate solutions When the set of feasible candidate solutions is empty, the scheduling server stops performing minimum value calculations, sends the baseline scheduling scheme infeasibility information to the project management terminal, and re-executes sub-steps S11 and S12 after the task conditions or resource status are corrected in the project management system or resource management system; When not empty, the scheduling server will aggregate the total project duration. The minimum value among them is used as the baseline total project duration, and a set of feasible candidate solutions is selected. The candidate scheduling scheme ranked first in the middle is used as the benchmark scheduling scheme. Therefore, when multiple candidate scheduling schemes have the same total project duration, the determined baseline scheduling result is obtained according to the generation order retained in sub-step S12. The scheduling server uses the baseline scheduling scheme Extract each task The start of the plan Plan completion time and benchmark resource binding set Baseline resource binding set Record task The actual bound device identifier, personnel identifier, spare parts identifier, and quantity; the scheduling server further aggregates the planned execution intervals of all tasks according to the resource identifiers, forming the first... The baseline occupancy record set for each resource: (4); in, For resources The baseline occupancy record set; To occupy resources Task identifier; and These are the planned start time and planned completion time of the task, respectively. Representing resources It has been bound to the task. ; The scheduling server will schedule the reference time. Baseline scheduling scheme Task Set Task Dependency Directed Graph Initial resource state set Hard constraint set The planned execution period for each task Baseline resource binding set and the baseline occupancy record set of each resource Associated storage serves as the baseline for scheduling snapshots Baseline scheduling snapshot With scheduling reference time The corresponding version identifier remains valid until a new resource disturbance event is received; the superscript "" in this implementation means that the version identifier remains valid until a new resource disturbance event is received. "" indicates the baseline state of the current resource disturbance handling cycle; after completing a scheduling scheme execution and feedback update, sub-step S53 generates a baseline scheduling snapshot for the next cycle with the effective scheduling scheme and the resource state after feedback, and updates the baseline state at the beginning of the next cycle.
[0057] In the preferred embodiment, step S2 specifically includes the following steps: S21. Equipment Failure Event Validation and Resource Availability Range Update The scheduling server receives the target device identifier reported by the device management system through the device management interface. Fault codes or recovery status codes When the fault occurred and expected recovery time When this disturbance is triggered by the execution feedback of sub-step S53, the scheduling server will start from the next cycle resource disturbance event. The corresponding device identifier, status code, time of the exception, and estimated recovery time are read from the database and assigned to the appropriate values. , , and The scheduling server then retrieves the baseline scheduling snapshot from the current period. Read the scheduling reference time and initial resource state set And based on the target device identifier In the initial resource state set Mid-positioning target equipment resources When the status code This indicates that when a new stopped or restricted job status occurs in the current cycle, the scheduling server verifies the time of the fault occurrence. No earlier than the scheduling reference time And the expected recovery time Later than the time the fault occurred When the status code When updating the recovery time or device recovery status of an existing fault event that has not yet been closed, the scheduling server associates the existing fault event with the target device identifier, allowing the original fault occurrence time of the existing fault event to be updated. Earlier than the current cycle's scheduling reference time And verify the updated estimated recovery time or the actual recovery time. Later Status code When the device is restored, the scheduling server will use the moment when the device management system confirms the restoration as the time when the device is restored. After successful verification, the scheduling server generates a valid device failure event. (5); in, This is a valid equipment failure event; Identify the target device; For the target device's fault code or recovery status code; This refers to the time when the corresponding fault event occurred; The estimated recovery time or the actual recovery time confirmed by the equipment management system; when the time relationship, equipment status, or existing fault event correlation fails verification, the scheduling server re-requests the target equipment status and recovery time from the equipment management system, and generates a valid equipment fault event after obtaining data that meets the above conditions. ; The scheduling server will record the time of the failure. Until the expected recovery time Determine the time interval for disabling the target device. and from the original available time interval of the target device Subtract the target device's disabled time interval The updated set of available time intervals for the target device is obtained: (6); in, Baseline scheduling snapshot The set of original available time intervals for the target device; Disable the target device during the specified time interval; The set of available time intervals for the target device after deducting the disabled time intervals; operator " This indicates that time periods overlapping with the disabled time intervals of the target device are removed from the original set; the scheduling server uses the updated set of available time intervals for the target device. Replace the initial resource state set In And keep other resource state objects unchanged to form a post-event resource state set. ; For fault events directly reported by the equipment management system, the scheduling server requests the time of the fault occurrence from the task execution terminal. The corresponding task execution status; for the resource disturbance event in the next cycle triggered by sub-step S53, the scheduling server reads the feedback task status from the baseline scheduling snapshot of the next cycle; the scheduling server uniformly represents each task execution status record obtained from any of the above sources as... ,in, As a task identifier; For the task at the moment of failure The execution status includes not started, in progress, paused, and completed; For the task at the moment of failure The remaining execution time is recorded, with the remaining execution time for completed tasks set to zero; the scheduling server summarizes the execution status records of all tasks into an execution status set at the time of the event. .
[0058] S22. Identification of Directly Affected Tasks and Propagation of Subsequent Impacts The scheduling server receives valid device failure events. Read the target device identifier and in the baseline occupancy record set of the target device. The task identifier and corresponding planned execution interval occupying the target device are read from the database; for each task occupying the target device, the scheduling server executes the set of execution states based on the event time. Read the execution status of the task. And it will match tasks with execution statuses of not started, in progress, or paused with the time interval when the target device is disabled. Perform interval overlap detection; When the task's baseline resource binding set Includes target device identifier Task execution status For those not started, in progress, or paused, and within the planned execution period. Disabling time interval of the target device When there is overlap, the scheduling server adds the task to the set of directly affected tasks: (7); in, The set of tasks that are directly affected; Indicates a completed state; the operator " " indicates the intersection of intervals; Represents an empty set; a set of resource states after an event. The set of available time intervals for target devices Used to confirm the period during which the target equipment was no longer available for the directly affected task within the originally planned execution period; The scheduling server uses the set of tasks directly affected. Starting with the task in the graph, we follow the directed graph along the task dependencies. The directed edges are used to perform successor traversal; for each successor task obtained through traversal, the scheduling server executes the set of execution states based on the event time. Read its execution status and add the subsequent tasks whose execution status is not started, in progress, or paused to the set of subsequent affected tasks. The scheduling server will be directly affected by the set of tasks. With the set of subsequent impact tasks Merge to form a local rescheduling influence domain: (8); in, For the local rescheduling influence domain; operator " "" indicates set merging; the propagation boundary of the affected domain is limited to the unfinished successor tasks of the directly affected tasks. Other tasks that share the same type of resources and have no task dependencies are retained in the original scheduling scheme, and their resource conflicts are handled by the freezing constraints generated in sub-step S23 and the resource constraint checks in main step S3.
[0059] S23. Formation of Freeze Constraints and Determination of the Set of Tasks to be Repaired The scheduling server executes a set of states based on the event time. Read the execution status of each task and remaining execution time And combined with the local rescheduling influence domain and task set Task Fault Recovery Properties Categorize tasks; For tasks whose execution status is "completed", the scheduling server fixes their actual completion record and baseline resource binding set. And records of resource usage that have already occurred; for those located within the local rescheduling impact domain For tasks other than those scheduled for execution, the scheduling server fixes the planned execution interval. Baseline resource binding set and baseline occupancy record set The above tasks together form a set of full-scale freeze tasks. ; For those located in the local rescheduling influence domain Internal, execution status is in progress or paused and task failure recovery attribute For tasks where the scheduling server records stops caused by device failures as forced pauses, it retains the execution records of the task and the baseline resource binding set. and valid equipment failure events Expected recovery time Let this be the time when the remaining part of the task resumes execution on the original bound resource; the scheduling server combines the execution status set at the event time. Remaining execution time Calculate the constraint completion time after the task is restored: (9); in, The constraint completion time for tasks that are not allowed to be migrated and restored after the original bound resources are restored; The estimated recovery time for the target equipment; The remaining execution time returned by the task execution terminal; the scheduling server binds the original resources of the task, the execution records that have already occurred, and the time to resume execution. and constraints on completion time Write to the full freeze task set The corresponding resource usage interval is updated to the continuous execution interval after the target device recovers, and is used to constrain the start time of subsequent tasks of this task; for tasks with execution status of in progress or paused and task failure recovery attributes The task is to schedule the execution records that were formed before the failure occurred on the server and keep the remaining parts in the candidate range of the set of tasks to be repaired so that the main step S3 can search for compatible resources. The scheduling server will partially reschedule the affected domain. The task set that has not yet been fully frozen The tasks have been identified as the set of tasks to be repaired: (10); in, The set of tasks to be repaired; the scheduling server starts from the baseline occupancy record set. Extract the full set of frozen tasks The corresponding resource usage records will be maintained, and the execution or paused tasks will be recorded from their actual start time to the time of the failure. The existing resource usage records are used as historical fixed records. Combined with tasks that are not allowed to be migrated and restored, a set of frozen resource usage is formed within the resource usage range after the original bound resources are restored. Subsequently, the entire task set will be frozen. Fixed execution range, fixed resource binding, and frozen resource usage set Target device disabled time interval and hard constraint sets Merge into a set of frozen constraints ; The scheduling server will handle valid device failure events. Post-event resource state set Execution state set at the moment of the event Task Set Task Dependency Directed Graph Baseline scheduling scheme Planned execution period Baseline resource binding set The set of tasks directly affected The set of subsequent impact tasks Set of tasks to be repaired Full freeze task collection Frozen resource usage set Freeze constraint set and hard constraint set Association forms a local rescheduling context .
[0060] In the preferred embodiment, step S3 specifically includes the following steps: S31. Local Removal of the Influence Domain and Generation of the Sequence to be Inserted The scheduling server receives valid device failure events. Read the time of the fault occurrence in the middle and in the set of tasks to be repaired The system internally defines the set of faulty resource removal tasks and the set of successor chain removal tasks: (11); in, This refers to the set of directly affected tasks that require the original resource bindings and planned execution intervals to be revoked. After excluding directly affected tasks, it is necessary to cancel the original planned execution interval and recalibrate the set of subsequent tasks with dependent timings; the operator " " represents the intersection of sets, operator " " represents the difference between sets; The scheduling server replicates the baseline scheduling scheme. Preserve the entire frozen task set The task record, execution range, and resource binding are configured, and the task set is removed for faulty resources. and successor chain removal task set Perform partial removal; the partial removal cancels the corresponding task in the baseline scheduling scheme. The scheduler records the start and end times of the task, as well as subsequent resource usage; the scheduling server retrieves data from the task set. The system retains the task identifier, task duration, prerequisites, and resource conditions of the corresponding tasks, as well as the execution records formed before the event occurred, thereby forming a partial scheduling scheme. ; The scheduling server uses a directed graph based on task dependencies. Set of repair tasks Perform topological sorting to generate a sequence of tasks to be inserted: (12); in, The sequence of tasks to be inserted; This represents a directed graph with task dependencies. The scheduling server performs a topological sort on a specified set of tasks, ensuring that any preceding task precedes its corresponding succeeding task in the sequence; the scheduling server retains each task from the set of tasks removed from failed resources. Or the successor chain removes the task set The source marker is used in sub-step S32 to determine the resource binding processing method; The scheduling server uses a set of post-event resource states. Based on this, the set of frozen resources occupied is deducted from the available time interval of each resource. When the fault occurs The continued occupation interval forms the initial resource state set for local repair. Local repair of the initial resource state set The state that the disabled time interval of the target device has been deducted from its available time interval is retained.
[0061] S32. Two-branch repair using the earliest feasible insertion and the minimum offset insertion. The scheduling server replicates part of the scheduling scheme. and the initial resource state set for local repair Establish the earliest feasible insertion branch and the minimum offset insertion branch respectively; This indicates the repair branch, where, This indicates the earliest feasible insertion branch. This represents the minimum offset insertion branch; both branches follow the sequence of tasks to be inserted. Process tasks one by one; For the task to be inserted The scheduling server executes the set of states at the time of the event. Read execution status and remaining execution time When execution status The estimated duration of the task before it begins. This serves as the duration of the current insertion; when the execution status... If execution is in progress or paused, the remaining execution time will be displayed. This serves as the duration of the current insertion; thus, the effective insertion duration is obtained: (13); in, Task to be inserted The effective insertion duration; Indicates a state that has not yet started; Indicates the status of execution; Indicates a paused state; The scheduling server receives valid device failure events. Read the time of the fault occurrence in the middle According to the task dependency directed graph Read the predecessor tasks of the task to be inserted, and read the completion time of the corresponding predecessor tasks from the task records that have been retained or inserted in the current repair branch; the scheduling server uses the time of the failure. The earliest allowed start time for the task The maximum value among the completion times of all preceding tasks is used as the lower bound for the time search of the task to be inserted. The scheduling server determines the task based on the set of resource conditions. The current resource availability status of the repair branch, and the set of frozen constraints. and hard constraint set Determine candidate execution intervals for the task to be inserted; the start time of a candidate execution interval is no earlier than the lower bound of the time search, and the completion time of a candidate execution interval is determined by its start time and the effective insertion duration. Add them together to get; The scheduling server filters the corresponding equipment, personnel, and spare parts for each candidate execution interval; the capability attributes of the selected equipment and personnel must meet the set of task resource conditions. Furthermore, the candidate execution interval must fall within the remaining available time interval of each selected resource; the available inventory of the selected spare parts at the start of the candidate execution interval must cover the task resource condition set. The specified spare parts requirement; the scheduling server simultaneously calls the frozen constraint set. This ensures that the candidate execution interval and its resource binding remain unchanged, maintaining the execution interval, resource binding, and frozen resource usage of all frozen tasks. The candidate execution intervals and resource bindings, checked against the aforementioned time constraints, resource capacity constraints, inventory constraints, and freeze constraints, collectively form the insertion record. ,in, The insertion start time, The insertion completion time, This refers to the resource binding set used for this insertion; the scheduling server will place the current task in the repair branch. All inserted records in the data are summarized into a feasible inserted record set. ; For the set of faulty resource removal tasks In the task, the scheduling server first inserts records into the feasible set. Search for inserted records bound to a compatible alternative device; if no compatible alternative device inserted records are found, then search again at the expected recovery time. Then rebind the insertion record of the original target device; for the successor chain removal task set In the task, the scheduling server first searches for the baseline resource binding set. The system then searches for inserted records that utilize other compatible resources. In the earliest feasible insertion branch, the scheduling server selects the earliest inserted record at the start time: (14); in, For the task The earliest feasible insertion record; To insert a record The corresponding start time; In the minimum offset insertion branch, the scheduling server selects the start time relative to the baseline schedule. The record with the smallest offset to insert: (15); in, For the task Insert a record with the minimum offset; For the task The planned start time in the baseline scheduling scheme; symbol " " indicates taking the absolute value; when there are multiple insert records with the same offset, the insert record with the earlier start time is selected; For each selected insertion record, the scheduling server writes the task start time, completion time, and resource binding set into the corresponding repair branch, deducts the newly written resource occupancy interval from the resource availability time interval of that branch, and updates the preceding completion time of the task's successor. These updates are then used in the sequence of tasks to be inserted. Generation of feasible insert records for the next task; when the set of feasible insert records for a certain repair branch... If empty, mark the completion of the repair branch. Set to zero and terminate the branch; when the task sequence to be inserted... When all tasks in the process have completed the insertion, a completion marker will be set. Set to one; The scheduling server will generate a scheduling scheme from each repair branch. Selected Insertion Record Set Updated resource status set and completion mark Related branch repair results .
[0062] S33. Integrity verification and candidate set generation of local recovery scheme The scheduling server reads the earliest feasible branch insertion repair result in sequence. Minimum offset insertion branch repair results For completion markers For branches with a value of zero, the scheduling server marks the branch as a partial repair failure and stops generating candidate solutions based on that branch; for branches with a completion mark... For a single branch, the scheduling server inserts records based on the selected set. Check the set of tasks to be repaired Check whether each task in the process has a corresponding insertion record, and check whether the same task is inserted only once; After the integrity check is passed, the scheduling server will implement the branch scheduling scheme. Compared with the baseline scheduling scheme Compare item by item to confirm the complete set of frozen tasks. The fixed execution records and fixed resource bindings for each task remain unchanged, and are based on the frozen resource usage set. Check if the newly inserted task occupies a frozen resource period; subsequently, the scheduling server calls the frozen constraint set. and hard constraint set Check the task prerequisites, equipment disabled zones, resource capacity, equipment compatibility, personnel skills, and spare parts quantity constraints. When branch scheduling scheme During integrity checks, freeze consistency checks, and full constraint checks, the scheduling server will branch the scheduling scheme. Selected Insertion Record Set and the updated resource status set Association forms local recovery candidate schemes And add it to the local recovery candidate scheme set. Branches that fail any check will not be included in the local recovery candidate set. ; The scheduling server uses the set of partial recovery candidate schemes. Does it include local recovery candidate schemes to generate local repair status? Local recovery candidate solution set When at least one local recovery candidate scheme is included, the local repair state will be... Set to 1; for the local recovery candidate solution set When empty, the local repair state will be restored. Set to zero.
[0063] In the preferred embodiment, step S4 specifically includes the following steps: S41. Quantitative Evaluation of Candidate Local Restoration Solutions and Determination of Target Solution The scheduling server first reads the partial repair status. ;when For a period of time, the scheduling server sequentially reads the set of partial recovery candidate schemes. The candidate solutions are selected, and the selected records are inserted into the record set. From the branch scheduling scheme Obtain the task to be repaired Corresponding candidate start time Candidate completion time and candidate resource binding Meanwhile, from the baseline scheduling scheme Obtain the reference start time corresponding to the same task. Benchmark completion time and benchmark resource binding The scheduling server compares each of the above records and calculates the number of adjustment tasks for each candidate scheme relative to the baseline scheduling scheme. (16); in, Candidate solutions Adjust the number of tasks; This is an indicator function; it takes the value of one when the condition within the parentheses is true, and zero when the condition is false; the operator " " indicates a logical OR; this indicator is used to characterize the scope of plan changes caused by equipment failure; The scheduling server continues to utilize candidate start times. and the reference start time Calculate the total offset at the start time of the task to be repaired: (17); in, Candidate solutions The total offset at the start time is used to characterize the degree of disturbance to the original planned execution sequence of the repair task during local repair; The scheduling server based on the task set From the branch scheduling schemes respectively and baseline scheduling scheme Read the completion time of each task and calculate the total project duration increment of the candidate solutions: (18); in, Candidate solutions The increase in the total project duration; For the task In branch scheduling scheme The completion time in the middle; For the task In the baseline scheduling scheme The completion time in the middle; for tasks that have not been locally repaired or adjusted, its Use the baseline completion time ; The scheduling server will adjust the number of tasks, the total offset of the start time, and the increment of the total project duration in sequence to form a candidate evaluation vector. The target branch is determined by performing lexicographical comparisons in the following order: priority of adjusting the number of tasks, second priority of the total offset at the start time, and third priority of the total project duration increment. (19); in, Identifier for the target branch; superscript This means comparing the indicators in the candidate evaluation vectors level by level according to their order; when two candidate evaluation vectors are completely identical, the scheduling server selects the branch with the smallest offset to insert; the scheduling server reads the branch scheduling scheme corresponding to the target branch. Selected Insertion Record Set Updated resource status set and candidate evaluation vectors Related to form a local target solution package And effectively mark local targets. Set to one; when When the value is zero, the scheduling server will package the local target scheme. Set to empty, and mark the local target effectively. Set to zero.
[0064] S42. Fault recovery time-based updates and bounded partial retries The scheduling server will count partial retry attempts. Initialize to zero and read the valid local target markers. ;when For the time being, the scheduling server will package the local target solution. This is denoted as the updated local target scheme package. Set the global rearrangement trigger flag. Set to zero and change the local rescheduling context. This is recorded as the updated context. ; when When the value is zero, the scheduling server obtains the latest estimated recovery time of the faulty resource through the device status acquisition interface. And local retry count Add one; the scheduling server will update the latest estimated recovery time. Replace the original expected recovery time in the local rescheduling context Update the disabled time interval of the target faulty resource to And based on the task execution status set at the time of the event. The remaining execution time and recoverable start time of the affected tasks are recalculated. The scheduling server then calls main step S2 to redetermine the directly affected task set, the subsequently affected task set, the task set to be repaired set, the frozen task set, and the frozen resource occupancy using the updated device failure events, resource status, and task execution status. Then, main step S3 is called to generate a new set of local recovery candidate schemes and local repair status. Sub-step S41 is called to form a new local target scheme package and local target validity markers according to the same evaluation rules. The above processing forms an updated context containing the latest fault recovery boundary, influence domain boundary, and freeze boundary. ; When a new local target is effectively marked as one, the scheduling server will record the corresponding local target scheme packet as the updated local target scheme packet. Set the global rearrangement trigger flag. Set to zero and terminate local retries; when the new local target valid flag is still zero and the local retry count is zero. Less than the preset maximum number of local retries At that time, the scheduling server obtains the latest estimated recovery time again and repeats the above update, partial repair, and evaluation process; when the local retry count... Reaching the preset maximum number of local retries If no effective local target solution has yet been formed, the scheduling server will update the local target solution package. Set to empty and set the global rearrangement trigger flag. Set to 1; preset maximum number of local retries It is an integer greater than or equal to one, and its value is jointly limited by the allowed duration of project scheduling response and the maximum processing time of a single local repair, so that the cumulative processing time of all local retries is kept within the allowed duration of project scheduling response; The scheduling server will update the local target scheme package. Global Reordering Trigger Flag Local retry count and the updated context The context of recovery decision formation is related. .
[0065] S43. Determining the Layered Recovery Path and Generating a Unified Target Recovery Scheme The scheduling server from the recovery decision context Read the updated local target scheme package Global Reordering Trigger Flag Local retry count and the updated context ;when When the value is zero, the scheduling server retrieves the updated local target scheme package. Reading local target scheduling scheme Local target insertion record set Local target resource state set and local target evaluation vector , target scheduling scheme Set as Set the target resource status Set as , target evaluation vector Set as and restore mode Set to partial recovery mode; local target insertion record set Used to locate the task record written during this partial repair, and compared with the baseline scheduling scheme. Check each change item one by one; when For a time, the scheduling server will access the updated context. The system reads the latest task execution status, updated resource status, task dependency directed graph, and hard constraint set; fixes the execution records of completed tasks and the historical occupancy intervals of tasks in execution or paused before the current moment; and aggregates all incomplete tasks into a global evaluation task set. The scheduling server evaluates the global set of tasks. The genetic search and constraint satisfaction process described in main step S1 is invoked. The joint solution process generates global candidate solutions for remaining tasks based on task dependencies, updated resource availability intervals, resource capability attributes, personnel skills, and spare parts inventory, and performs feasibility screening according to the set of hard constraints. When at least one feasible candidate solution is obtained, the scheduling server selects the candidate solution with the earliest project completion time as the global recovery scheduling solution. Simultaneously form a global recovery resource status set The scheduling server evaluates the task set globally. Replace the set of tasks to be repaired in sub-step S41 Calculate the number of adjusted tasks, total offset at start time, and total project duration increment of the global recovery scheduling scheme relative to the baseline scheduling scheme, and form a global recovery evaluation vector according to the arrangement order specified in sub-step S41. The scheduling server will use the target scheduling scheme. Set as Set the target resource status Set as , target evaluation vector Set as Restore mode Set to global recovery mode and effectively mark the target scenario. Set to 1; when the joint solution process fails to obtain a feasible candidate solution that satisfies the set of hard constraints within the preset scheduling time range, the scheduling server will set the target scheduling solution. Set to empty, and set the target resource state set. Set the target evaluation vector to the latest resource state in the updated context. Set to empty, recovery mode Set to global recovery failure mode and mark the target scheme as valid. Set to zero; for partial recovery paths, the scheduling server will effectively mark the target scheme. Set to one; When the target scheme is effectively marked For a given time, the scheduling server inserts records into the local target set. Determine the scope of tasks to be verified for the local recovery path, or determine the scope of tasks to be verified for the global recovery path based on all incomplete tasks, and then set the target scheduling scheme. Compared with the baseline scheduling scheme The task start time, completion time, and resource bindings are compared item by item to generate a target change record set: (20); in, For the target change record set; As a task identifier; , and These are the task start time, completion time, and resource binding in the target scheduling scheme; the target scheme is marked as valid. When the value is zero, the scheduling server will set the target change record set. Set to empty to prevent subsequent execution from writing infeasible scheduling results; The scheduling server will target the scheduling scheme. Target resource status set Target change record collection Recovery Mode Target evaluation vector Target solution valid marking and local retry count Associating them to form a unified target recovery solution package .
[0066] In the preferred embodiment, step S5 specifically includes the following steps: S51. Target scheduling scheme pre-writing and multi-terminal execution instruction generation The scheduling server reads the target scheme's valid flag. ;when For a period of time, the scheduling server generates a identifier for this recovery process. And based on the currently active scheduling version number Generate the scheduling version number to be activated The scheduling server uses the pending scheduling version number. Create a shadow version in the scheduling database and store the target scheduling scheme. The task start time, completion time, and resource bindings are written into the shadow task table, and the target resource state set is recorded. The available equipment range, equipment operation queue, personnel occupied range, and spare parts reservation status are written into the shadow resource table, and then based on the target change record set. Lock the task records and resource usage records that need to be updated this time; the shadow version remains in a pending state until the required confirmation result from the receiving end is obtained in sub-step S52, and the current effective scheduling version number is... The corresponding task records continue to serve as valid data on the execution side; The scheduling server reads the target change record set item by item. For any target change record, the scheduling server binds it to the target resource. Extract the target equipment, target personnel, and target spare parts, and combine this with the target start time. Target completion time Generate task adjustment instructions: (twenty one); in, For the task Task adjustment instructions; To restore the processing identifier; The version number of the schedule to be implemented; As a task identifier; and These are the start time of the objective and the completion time of the objective, respectively. Bind to the target resource; the scheduling server binds to the target resource. Generate and adjust task instructions The associated equipment queue update data, personnel task assignment data, and spare parts reservation data enable the equipment control terminal to adjust the corresponding equipment's work queue, enable personnel terminals to update task reception and arrival times, and enable the warehouse management terminal to lock the required spare parts quantity according to task identifiers and target start times. The scheduling server aggregates all task adjustment instructions and their associated data into a set of instructions to be issued. ; When the target scheme is effectively marked When the value is zero, the scheduling server generates a recovery processing identifier. Use the current active scheduling version number. As the version number to be processed Maintain the current active state of the task table and resource table, and adjust according to the recovery mode. Generate scheduling exception instructions Dispatch exception instructions Carry recovery processing label Current effective scheduling version number Recovery Mode and local retry count This is used to notify the project management terminal that no executable recovery plan has been formed; the scheduling server will schedule abnormal instructions. Write the set of instructions to be issued ; Target evaluation vector Recovery Mode and local retry count Both processing paths are associated with the recovery processing identifier. Associate the writing of shadow versions or exception records, which are used in sub-step S52 to record the formation process of this scheduling scheme; The scheduling server will restore the processing flag. Current effective scheduling version number Pending scheduling version number Target scheduling scheme Target resource status set Target change record collection Set of instructions to be issued Recovery Mode Target evaluation vector Target solution valid marking and local retry count Associative formation instruction preparation context .
[0067] S52, Multi-terminal command issuance confirmation and scheduling version switching The scheduling server prepares the context from the instructions. Read the valid markers of the target scheme in the middle Recovery processing mark Current effective scheduling version number Pending scheduling version number Set of instructions to be issued ;when For a period of time, the scheduling server binds the target resources according to each task adjustment instruction, sends the task adjustment instruction to the corresponding task execution terminal, sends the equipment queue update data to the equipment control terminal, sends the personnel task assignment data to the personnel terminal, and sends the spare parts reservation data to the warehouse management terminal. Each receiving terminal checks the local task, equipment, personnel, or spare parts status based on the task identifier and the scheduling version number to be effective. When the corresponding adjustment can be executed, it locks the corresponding resource and returns a confirmation message. When the resource status has changed or the instruction version is lower than the locally confirmed version, it returns a rejection message. Each confirmation message is represented as follows: (twenty two); in, For the task Corresponding receiver Confirmation information; This refers to the task execution terminal, equipment control terminal, personnel terminal, or warehouse management terminal. To confirm the status; The timing for generating confirmation information at the receiving end; the scheduling server only receives messages with the recovery processing identifier equal to... And the version number equals The confirmation information will be sent and the verified confirmation information will be written into the confirmation information set. ; The scheduling server waits for a preset confirmation period. Internal check of pending instruction set Has all the receiving ends returned an acknowledgment? If there is a receiving end that has not returned an acknowledgment, the scheduling server will only resend the corresponding instruction to that receiving end and record the cumulative number of transmissions. If the cumulative number of transmissions is less than the preset maximum number of transmissions... At this point, the scheduling server continues to wait for the receiving end to return confirmation information; the cumulative number of transmissions reaches the preset maximum number of transmissions. The confirmation process ends when any receiver returns a rejection message or all receivers return confirmation messages. When all required receivers return confirmation messages and the confirmation status in each confirmation message is... When both are accepted, the scheduling server atomically switches the shadow task table and shadow resource table in the scheduling database to the active state, and sets the scheduling version number to be active. Record as the version number that takes effect after the switch. and the target scheduling scheme Target resource status set and target change record collection These are respectively denoted as the effective scheduling schemes. Effective resource status set and collection of effective change records At the same time, the command status will be issued. Set to one; When any of the requested receivers returns a rejection message, or the cumulative number of transmissions reaches the preset maximum number of transmissions. If any receiving end still fails to return confirmation information, the scheduling server sends a resource unlock command to the receiving end that has already returned confirmation information, thus revoking the target change record set. The corresponding shadow task records and shadow resource records will continue to use the currently effective scheduling version number. and its corresponding scheduling scheme and resource status; the scheduling server will switch to the new version number. Set as The set of change records will take effect. Set to empty and set the command issuance status. Set to zero; activate the scheduling scheme. and effective resource status set The current active scheduling version number is located under this path. The corresponding scheduling scheme and resource status; when When the value is zero, the scheduling server will have a set of instructions waiting to be issued. The scheduling exception command is sent to the project management terminal, and the received result returned by the project management terminal is written into the confirmation information set. The version number will take effect after the switch. Set to the currently active scheduling version number The scheduling scheme will take effect. and effective resource status set Set to the currently active scheduling version number respectively The corresponding scheduling scheme and resource status will be used to create a set of change records. Set to empty and set the command issuance status. Set to zero; The scheduling server will restore the processing flag. Target solution valid marking Version number to take effect after switching Effective scheduling scheme Effective resource status set Collection of effective change records Set of instructions to be issued Confirmation information set Command issuance status Recovery Mode Target evaluation vector and local retry count Association forms execution monitoring context .
[0068] S53, Execution Feedback Analysis and Next Scheduling Cycle Status Update The scheduling server executes monitoring context Read the recovery processing identifier Target solution valid marking Version number to take effect after switching Effective scheduling scheme Effective resource status set Collection of effective change records Set of instructions to be issued Confirmation information set and the status of instruction issuance ;when and All are temporary; the scheduling server operates according to the set of effective change records. Establish and adjust the monitoring scope of the task, and based on the confirmed information set. Determine the feedback source for each adjustment task; the task execution terminal generates task feedback when the task is received, started, paused, resumed, and completed; the equipment control terminal generates equipment feedback when the equipment enters an operating, idle, or faulty state; the personnel terminal generates personnel feedback when personnel arrive, leave, or become unavailable; and the warehouse management terminal generates spare parts feedback when spare parts are locked, issued, returned, or out of stock. Any execution feedback is represented as: (twenty three); in, For the first Execution feedback; Feedback type; The moment the feedback occurs; The data includes task status, resource status, or quantity data corresponding to the feedback type; the scheduling server only processes data where the recovery processing identifier is equal to... And the version number equals The execution feedback, and the use of task identifiers Write the feedback data into the corresponding task record and resource record; The scheduling server updates the task execution status and actual start or finish time based on task feedback; when a task is paused, the scheduling server updates the remaining execution time of the task based on the already executed duration; when a task is completed, the scheduling server releases the equipment time interval and personnel time interval occupied by the task, and, based on feedback from the warehouse management end, includes outbound spare parts in the consumption and restores unused spare parts to available inventory; the scheduling server updates the set of effective resource statuses based on equipment feedback, personnel feedback, and spare parts feedback. The available resource range, effective capacity status, and inventory status are used to form a set of task statuses after feedback. and the set of resource statuses after feedback ; The scheduling server will report the time of occurrence. With effective scheduling scheme The actual start or finish time of the task is compared with the planned time; when the deviation between the actual start or finish time and the planned time reaches a preset time deviation threshold... The device control terminal reports that the target device is unavailable, or based on the set of confirmation information... The determined device control terminal waits for a preset execution feedback time. If no status is returned within the specified time, the scheduling server queries the current status of the device bound to the relevant task through the device management interface; when the device management system confirms that the device is in a stopped or restricted operation state, the scheduling server marks the next cycle event trigger. Set to 1, and generate the next cycle's resource disturbance event based on the target device identifier, device fault code, time of anomaly occurrence, and latest estimated recovery time. When the equipment management system does not provide an estimated recovery time, the scheduling server uses the end time of the current scheduling time range as the estimated recovery time. The scheduling server will assign a temporary value to the target device and maintain its disabled state. Before the device management system returns a clear recovery status, the scheduling server must not restore the target device to usability based on the temporary value. Once the device management system confirms that the relevant device remains usable, or that no abnormal conditions have occurred as described above, the scheduling server will mark the next cycle event trigger. Set to zero and set the resource disturbance event for the next cycle. Set to empty; personnel feedback and spare parts feedback continue to be used to update the resource status set after feedback. Furthermore, it is not converted into a device failure event in this device failure implementation method.
[0069] when as one and When the value is zero, the scheduling server compares the set of instructions to be issued item by item according to the task identifier and the receiver identifier. With confirmation information set The system locates the receiving end that returns a rejection status or no confirmation information, and marks the relevant tasks as pending verification in the currently active task status to form a set of task statuses after feedback. And mark the relevant resources as pending verification in the current effective resource status to form a set of resource statuses after feedback. When the receiving end is a device control end and the device management system confirms that the corresponding device is in a stopped or restricted operation state, the scheduling server will mark the next cycle event trigger. Set to 1, and generate the next cycle resource disturbance event carrying the target device identifier, device fault code, time of anomaly occurrence, and expected recovery time. When the receiving end that rejects the instruction or does not return a confirmation message is a task execution terminal, a personnel terminal, or a warehouse management terminal, the scheduling server will mark the next cycle event trigger. Setting it to zero will reduce the resource disturbance event in the next cycle. Setting it to empty will leave the corresponding terminal status as pending verification and wait for the corresponding management system to return a verifiable status update; when When the value is zero, the scheduling server writes the current active task status and the current active resource status into the feedback task status set, respectively. and the set of resource statuses after feedback Maintain the current effective scheduling scheme and resource usage status, and mark the event trigger for the next cycle. Setting it to zero will reduce the resource disturbance event in the next cycle. Set to empty, and maintain the scheduling exception instruction already sent in sub-step S51; after the device management system returns new device availability information, the scheduling server updates the feedback resource status set. The project management terminal is then notified. When the project management terminal confirms the restart of scheduling, the scheduling server re-executes the main step S1 with the updated task and resource status, re-forms the baseline scheduling scheme, and then enters the subsequent resource disturbance processing process.
[0070] The scheduling server will resume mode. Target evaluation vector Local retry count Confirmation information set The execution feedback received during this period is used to form a recovery execution log. And will restore the processing mark Version number to take effect after switching Effective scheduling scheme , Task status set after feedback Resource status set after feedback and restore execution log Association forms execution feedback snapshot .
[0071] The scheduling server reads execution feedback snapshots from the unified project clock. The generation time; the next cycle event trigger marker. For a period of time, the scheduling server will start from the next cycle of resource disturbance events. When the read exception occurred And set it as the reference time for the next cycle. The next cycle event trigger flag. When the value is zero, the scheduling server will execute a feedback snapshot. The generation time is set as the reference time for the next cycle. The scheduling server will take effect with the scheduling scheme. Set as the baseline scheduling scheme for the next cycle and the set of resource statuses after feedback Set as the initial resource state set for the next cycle The scheduling server uses the task status set after feedback. The actual start time, actual completion time, and actual resources of the completed tasks are written into the execution interval of the next cycle plan. and the next cycle's benchmark resource binding set It also fixes the historical occupancy of tasks already in execution or suspended; for the future portions of unfinished tasks, the scheduling server will use the baseline scheduling scheme for the next cycle. Extract the corresponding planned execution interval and resource binding and write it. and The scheduling server then aggregates the actual and planned occupancy ranges of all tasks according to their resource identifiers, forming a baseline occupancy record set for the next cycle. The scheduling server will set the reference time for the next cycle. Next cycle baseline scheduling scheme Task Set Task Dependency Directed Graph The initial resource state set for the next cycle Hard constraint set , Task status set after feedback Next cycle plan execution interval Next cycle benchmark resource binding set and the next cycle's baseline occupancy record set The association forms the baseline scheduling snapshot for the next cycle. .
[0072] When the next cycle event triggers the flag. For a short period, the scheduling server will schedule the resource disturbance events for the next cycle. and the next cycle baseline scheduling snapshot Both are passed to the main step S2 and executed at the start of the next cycle. This makes the main step S2 , , , , and The corresponding fields in the next cycle's baseline scheduling snapshot are updated respectively; the next cycle event trigger flag. When the value is zero, the scheduling server continues to use the execution feedback snapshot. Receive subsequent feedback; when subsequent equipment failure events arrive, the scheduling server generates the next cycle baseline scheduling snapshot in the same way and then enters the main step S2; thus, the actual task status, physical resource status and effective scheduling scheme formed by the execution feedback become the baseline data for the next impact domain identification and local repair.
[0073] In the preferred embodiment, the intelligent scheduling system for project tasks based on limited resources is deployed in the project scheduling center, including a scheduling server, task terminals, resource terminals, and project management terminals. The scheduling server establishes data communication connections with the task terminals, resource terminals, and project management terminals through communication interfaces. The resource terminals include equipment control terminals, personnel terminals, and warehouse management terminals, which are used to receive equipment operation sequence update data, personnel task assignment data, and spare parts reservation data, and to provide feedback on the availability status and execution results of the corresponding resources.
[0074] This embodiment uses a Lenovo ThinkSystem SR250 V3 server as the scheduling server, configured with a server processor, RAM, redundant storage device and Ethernet communication interface; the server processor is used to execute the project task scheduling program, the RAM is used to save the task status, resource status and candidate recovery schemes of the current scheduling cycle, the redundant storage device is used to save the scheduling database, scheduling version and execution feedback data, and the Ethernet communication interface is used to connect the project business network and resource control network.
[0075] The task terminal uses a Zebra TC22 mobile data terminal, which has data processing, data storage, and wireless communication capabilities. It is used to receive task adjustment instructions and provide feedback on task reception, start-up, pause, resumption, and completion status. The equipment control terminal uses a Siemens SIMATIC IOT2050 industrial gateway, which has Ethernet and serial communication interfaces. It is used to receive equipment operation sequence update data and transmit the operation sequence and resource lock status to the corresponding equipment controller. The personnel terminal can use a Zebra TC22 mobile data terminal, and the warehouse management terminal can use a client program installed on the warehouse workstation. The project management terminal uses a Lenovo ThinkCentre M70q Gen 4 microcomputer, which has data processing, data storage, and network communication capabilities. It is used to display the target scheduling plan, receive scheduling exception instructions, and submit manual handling results.
[0076] The project task scheduling program includes a baseline scheduling module, a disturbance handling module, a local repair module, a recovery decision module, and an execution feedback module. Each functional module is called by the processor of the scheduling server and is associated with the input data and processing results of each stage through task identifier, resource identifier, scheduling version, and timestamp.
[0077] The baseline scheduling module obtains task duration, task dependencies, allowed task execution time, equipment capacity, personnel skills, resource availability time interval, and spare parts inventory status from the project management system, equipment management system, personnel management system, and warehouse management system to form task constraints. The baseline scheduling module generates feasible candidate scheduling schemes according to task dependencies and task constraints, determines the baseline scheduling scheme from the feasible candidate scheduling schemes, and writes the task execution interval, resource binding, and resource occupancy records into the baseline scheduling snapshot.
[0078] The disturbance handling module receives resource disturbance events carrying the target resource identifier, disturbance time, and expected recovery time through the communication interface, and updates the target resource availability time interval based on the resource disturbance events. The disturbance handling module reads the baseline scheduling snapshot and the task execution status fed back by the task terminal, identifies incomplete tasks that occupy the target resource and overlap with the resource disable time interval, and identifies incomplete successor tasks along the task dependencies, thereby determining the local influence domain, tasks to be repaired, and freeze constraints.
[0079] The local repair module replicates the baseline scheduling scheme, retains the execution interval, resource binding, and resource occupation of frozen tasks, and removes the execution schedule of the tasks to be repaired after the disturbance time. The local repair module determines the insertion order of the tasks to be repaired based on the task dependencies. Among the candidate execution intervals that satisfy the task constraints, the available time interval of updated resources, and the freezing constraints, the module selects the candidate execution interval with the earliest start time and the candidate execution interval with the smallest offset relative to the baseline start time, respectively, to form the earliest feasible insertion scheme and the smallest offset insertion scheme.
[0080] The recovery decision module determines the number of adjustment tasks, the total offset at the start time, and the increment of the total project duration for each of the two insertion schemes, and determines the target scheduling scheme according to the priority order of the number of adjustment tasks, the total offset at the start time, and the increment of the total project duration. When a feasible solution is not formed by local repair, the recovery decision module updates the resource disturbance event according to the latest expected recovery time returned by the equipment status acquisition interface, and re-executes the local repair within the preset maximum number of local retries to form a bounded retries. When a feasible solution is still not formed after the preset maximum number of local retries, the recovery decision module performs global rescheduling for the incomplete tasks, and generates a scheduling exception instruction when there is no solution for global rescheduling.
[0081] The execution feedback module generates task adjustment instructions and resource control data based on the target scheduling scheme and target change records. It sends the task adjustment instructions to the task terminal and sends updated equipment operation sequence data, personnel task assignment data, and spare parts reservation data to the corresponding resource terminals. The task terminal and resource terminals verify the task identifier and the scheduling version to be effective, lock the resources that can be adjusted, and return confirmation information. After all required receiving terminals return an acceptance status, the execution feedback module switches the scheduling version, controlling the equipment control terminal to update the equipment operation sequence, the personnel terminal to update task assignment, and the warehouse management terminal to update spare parts reservation. If a rejection status exists or a complete confirmation is not obtained after reaching the preset maximum number of issuances, the execution feedback module unlocks the locked resources and maintains the original effective scheduling version. The execution feedback module continues to receive task and resource status feedback, and based on the effective scheduling scheme, the updated task status, and the resource status, it forms a baseline scheduling snapshot for the next cycle and provides this snapshot to subsequent resource disturbance handling and local repair.
[0082] The scheduling database is implemented using a relational database to associate and store task data, resource status, baseline scheduling snapshots, local rescheduling data, candidate recovery schemes, scheduling versions, and execution feedback data. The baseline scheduling snapshots, candidate recovery schemes, and execution feedback data are respectively associated with the scheduling version and the generation time, enabling each functional module to read data within the same scheduling cycle and keeping the task records and resource usage records before and after the version switch traceable.
[0083] The models and configurations of the aforementioned servers, mobile data terminals, industrial gateways, and project management terminals are used to illustrate a feasible hardware combination; servers, industrial computers, edge gateways, or mobile terminals with corresponding data processing capabilities, storage capabilities, data communication capabilities, and resource control interfaces can all be used to carry the corresponding functional modules.
[0084] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.
Claims
1. A method for intelligent scheduling of project tasks based on limited resources, characterized in that, Includes the following steps: S1. The scheduling server obtains task data and resource status containing duration and task dependencies, forms task constraints and baseline scheduling scheme, and generates a baseline scheduling snapshot that records task execution interval, resource binding and resource consumption based on the baseline scheduling scheme. S2. Receive resource disturbance events containing the target resource identifier, disturbance time, and expected recovery time. Use the period from the disturbance time to the expected recovery time as the disabled time interval, update the available time interval of the target resource, obtain the task status and remaining duration at the disturbance time, identify unfinished tasks bound to the target resource in the baseline scheduling scheme and whose execution interval overlaps with the disabled time interval, as well as unfinished successor tasks determined along the task dependencies, form a local influence domain, fix tasks outside the local influence domain, completed tasks, and historical resource occupancy, and determine the tasks to be repaired and freeze constraints. S3. Remove the execution schedule after the disturbance of the task to be repaired, determine the insertion duration based on the task status, duration and remaining duration, and insert the task to be repaired into the candidate execution interval that satisfies the task constraints, resource available time interval and freeze constraints according to the task dependency, forming the earliest feasible insertion scheme and the minimum offset insertion scheme. S4. Compare the earliest feasible insertion scheme and the smallest offset insertion scheme in the order of the number of tasks to be adjusted, the total offset at the start time, and the increment of the total project duration to determine the local target scheduling scheme. If no local target scheduling scheme is formed, bounded retries are performed according to the updated recovery time. If no local target scheduling scheme is formed after bounded retries, global rescheduling is performed on the unfinished tasks. When a feasible recovery scheme is formed, a target scheduling scheme and target change record are generated. When no feasible recovery scheme is formed, a scheduling exception instruction is generated. S5. When a solution is found, send adjustment instructions and resource control data based on the target scheduling scheme and target change records to the task terminal and resource terminal. After confirmation by the receiving end, switch the scheduling version and control the resource terminal to update the job sequence, task assignment and resource reservation. When no solution is found, send the scheduling exception instruction to the project management terminal, receive execution feedback, update the task and resource status, and form the next cycle baseline scheduling snapshot with the effective scheduling scheme.
2. The intelligent scheduling method for project tasks based on limited resources according to claim 1, characterized in that, Step S1 specifically includes: Obtain the scheduling baseline time from the unified project clock, and obtain the task identifier, estimated duration, direct predecessor tasks, task resource conditions, earliest allowed start time, latest required finish time, and task failure recovery attributes. Establish task dependencies based on direct preceding tasks; Obtain the resource identifier, resource type, available capacity, available time interval, and capability attributes of limited resources. Limited resources include equipment resources, personnel resources, and spare parts resources. Task resource conditions include equipment type and quantity, personnel skills and quantity, and spare parts type and quantity. Hard constraints are formed based on task dependencies, allowed execution time, available resource capacity, available resource time interval, resource capability attributes, and spare parts inventory. Candidate scheduling schemes are generated based on task dependencies. For candidate scheduling schemes that do not meet hard constraints, task timing adjustments, resource rebinding, or search for the next available time interval are performed. A baseline scheduling scheme is determined from the candidate scheduling schemes that meet hard constraints. Extract the task plan start time, plan completion time, and baseline resource binding from the baseline scheduling scheme. Form a baseline resource occupancy record according to the resource identifier. Store the scheduling baseline time, baseline scheduling scheme, task data, task dependencies, resource status, hard constraints, task plan execution interval, baseline resource binding, and baseline resource occupancy record as a baseline scheduling snapshot.
3. The intelligent scheduling method for project tasks based on limited resources according to claim 1, characterized in that, Step S2 specifically includes: Obtain the scheduling baseline time for the current scheduling cycle, limit the target resource to equipment resources, and receive the target equipment identifier, fault code or recovery status code, fault occurrence time, and expected recovery time or actual recovery time through the equipment management interface; Based on the status code, the time of fault occurrence, the time of recovery, the scheduling reference time, and the correlation between the fault events that are not closed, the resource disturbance event is verified, and a valid equipment fault event is formed after the verification is passed. The time interval from the time of the fault occurrence to the time of recovery is defined as the disabled time interval of the target device. The disabled time interval of the target device is excluded from the original available time interval of the target device, and the updated available time interval of the target device is used to form the post-event resource status. Obtain the task execution status and remaining execution time at the time of the failure from the task terminal. The task execution status includes not started, in progress, paused, and completed.
4. The intelligent scheduling method for project tasks based on limited resources according to claim 3, characterized in that, Step S2 also includes: Read the tasks occupying the target device, their corresponding planned execution intervals, and resource bindings from the baseline scheduling scheme. Determine the tasks whose resource bindings include the target device, whose execution status is not started, in progress, or paused, and whose planned execution intervals overlap with the target device's disabled time interval as directly affected tasks. The directly affected task is used as the starting task in the task dependency relationship. The subsequent tasks that have not started, are being executed, or are paused are identified as the subsequent affected tasks. The directly affected task and the subsequent affected task form a local influence domain. Fix the execution records, resource bindings, and resource usage of completed tasks and tasks that partially affect external domains; Read the task failure recovery attributes of tasks within the local impact domain from the task data. Based on the task failure recovery attributes, retain the original resource bindings and historical execution records for running or paused tasks that are not allowed to be migrated and recovered. Determine the recovery execution interval based on the recovery time and remaining execution time. For running or paused tasks that are allowed to be migrated and recovered, fix the historical execution records before the failure occurred and retain the remaining part of the corresponding task in the scope to be repaired. Unfixed tasks within the local impact domain are identified as tasks to be repaired, and freeze constraints are formed based on fixed execution intervals, fixed resource bindings, historical resource usage, restored execution intervals, target device disabled time intervals, and task constraints.
5. The intelligent scheduling method for project tasks based on limited resources according to claim 1, characterized in that, Step S3 specifically includes: The tasks to be repaired are divided into directly affected tasks and subsequently affected tasks; Copy the baseline scheduling scheme and retain the task records, execution intervals and resource bindings of the fixed tasks in step S2. Remove the planned start time, planned completion time and resource occupation after the disturbance time of the directly affected tasks and the subsequent affected tasks. At the same time, retain the task identifier, duration, task dependency, task resource conditions and historical execution records before the disturbance time to form a partial scheduling scheme. The insertion order of tasks to be repaired is determined according to task dependencies, and the source markers of tasks to be repaired belonging to directly affected tasks or subsequent affected tasks are retained. In the resource availability time interval included in the updated resource status in step S2, the resource occupancy limited by the freeze constraint is excluded, and the target resource disabled time interval is kept in an unavailable state, forming the initial resource status for local repair.
6. The intelligent scheduling method for project tasks based on limited resources according to claim 5, characterized in that, Step S3 also includes: Based on the partial scheduling scheme and the initial resource state of local repair, the earliest feasible insertion branch and the minimum offset insertion branch are established respectively, and the tasks are processed one by one according to the insertion order of the tasks to be repaired. For tasks that have not started, the estimated duration is used as the insertion duration; for tasks that are running or paused, the remaining execution duration is used as the insertion duration. The starting boundary of the candidate execution interval is defined based on the disturbance time, the earliest allowed start time of the task, and the completion time of the preceding task, and the duration of the candidate execution interval is made consistent with the insertion duration. Based on task resource conditions, initial resource status of partial repair, task constraints and freeze constraints, candidate execution intervals and resource bindings are selected to ensure that equipment capacity, personnel skills, available resource time intervals and available spare parts inventory meet the task resource conditions. For tasks directly affected, priority is given to searching for candidate execution intervals that are bound to compatible alternative resources. If no candidate results are obtained, candidate execution intervals that are rebound to the original target resource after the recovery time are searched. For subsequent tasks that are affected, priority is given to searching for candidate execution intervals that maintain the binding of the baseline resource. If no candidate results are obtained, candidate execution intervals that are bound to other compatible resources are searched. The earliest feasible insertion branch selects the candidate execution interval with the earliest start time, while the minimum offset insertion branch selects the candidate execution interval with the smallest offset relative to the baseline plan's start time. After each task insertion is completed, the task execution interval, resource binding, resource availability time interval, and the prerequisite completion time of subsequent tasks in the corresponding branch are updated. After all tasks to be repaired have been inserted, a corresponding local recovery candidate solution is formed.
7. The intelligent scheduling method for project tasks based on limited resources according to claim 1, characterized in that, Step S4 specifically includes: For the earliest feasible insertion scheme and the minimum offset insertion scheme, the start time, completion time and resource binding of the task to be repaired are compared with the candidate scheme and the baseline scheduling scheme to determine the number of tasks to be adjusted. The total offset of the start time is determined based on the degree of offset of the candidate start time relative to the reference start time; The increment of the total project duration is determined based on the project completion time of the candidate scheduling scheme relative to the project completion time of the benchmark scheduling scheme. Candidate schemes are compared step-by-step in the order of adjusting the number of tasks, the total offset at the start time, and the increment of the total project duration. The candidate scheme with the highest ranking is determined as the local target scheduling scheme. When all comparison results are the same, the scheme with the smallest offset is determined as the local target scheduling scheme, and the task insertion record and updated resource status corresponding to the local target scheduling scheme are retained. The tasks to be checked are determined based on the task insertion record. The task start time, completion time, and resource binding in the local target scheduling scheme are compared with those in the baseline scheduling scheme to generate a target change record for the local recovery path. The updated resource status is determined as the target resource status of the local recovery path.
8. The intelligent scheduling method for project tasks based on limited resources according to claim 7, characterized in that, Step S4 also includes: When no local target scheduling scheme is formed, the target resource is limited to equipment resources. The latest estimated recovery time of the target equipment is obtained through the equipment status acquisition interface. The resource disturbance event, the target equipment disable time interval and resource status are updated. The remaining execution time and recoverable start time of the affected tasks are recalculated based on the task execution status. The local influence domain and the tasks to be repaired are redefined based on the updated resource disturbance event, resource status and task execution status. The freeze constraints are updated based on the remaining execution time, recoverable start time and the redefined local influence domain. Re-execute local repair and candidate scheme comparison within the maximum number of local retries; When the maximum number of local retries is reached and a local target scheduling scheme has not yet been formed, the execution records of completed tasks and the historical occupancy of tasks that are in execution or paused are fixed, and a global rescheduling is performed on all uncompleted tasks to satisfy the constraints of task dependencies, resource availability time intervals, resource capabilities, personnel skills and spare parts inventory. When a global recovery scheduling plan is formed, a target scheduling plan and a target change record are generated. If a global recovery scheduling plan is not formed within the preset scheduling time range, a scheduling exception instruction is generated.
9. The intelligent scheduling method for project tasks based on limited resources according to claim 1, characterized in that, Step S5 specifically includes: Generate recovery processing identifier; When forming a target scheduling plan, the target resource status is formed based on the task execution interval and resource binding update resource availability time interval and resource occupancy status in the target scheduling plan. A pending scheduling version is established based on the current effective scheduling version. The task execution interval, resource binding and target resource status of the target scheduling plan are written into the pending scheduling version. Based on the target change record, task adjustment instructions, equipment operation queue update data, personnel task assignment data and spare parts reservation data are generated, each carrying a recovery processing identifier, a pending scheduling version and a task identifier. Send the task adjustment instruction to the corresponding task terminal, and send the equipment operation queue update data, personnel task assignment data and spare parts reservation data to the corresponding resource terminal; The receiving end checks the local status based on the task identifier and the scheduling version to be effective. When the adjustment can be performed, it locks the corresponding resource and returns confirmation information carrying the recovery processing identifier, the scheduling version to be effective, the task identifier, and the acceptance status. When the resource status has changed or the instruction version is lower than the locally confirmed version, it returns confirmation information carrying the recovery processing identifier, the scheduling version to be effective, the task identifier, and the rejection status. Within the confirmation waiting period, check the confirmation information of all receivers. If there are receivers that have not returned confirmation information, resend the corresponding instruction only to the receivers that have not returned confirmation information, and continue to wait for confirmation information if the cumulative number of transmissions is less than the maximum number of transmissions. When all the confirmation messages returned by the receivers carry an acceptance status, the pending scheduling version is switched to the effective scheduling version; when there are receivers that return confirmation messages carrying a rejection status, or when the cumulative number of sends reaches the maximum number of sends and there are still receivers that have not returned confirmation messages, the locked resources are released, the task records and resource records in the pending scheduling version are revoked, and the original effective scheduling version continues to be used. When generating a scheduling exception command, send the scheduling exception command to the project management terminal and keep the original effective scheduling version. Receive task execution feedback and resource status feedback carrying recovery processing identifier and effective scheduling version; update task status, actual execution time and remaining execution time according to task execution feedback; update resource availability time interval, capacity validity status and spare parts inventory status according to resource status feedback; generate the next cycle resource disturbance event when resource status feedback indicates equipment abnormality and the equipment management system confirms that the equipment has stopped or restricted operation. The effective scheduling scheme, updated task status, updated resource status, actual execution interval, future planned execution interval, resource binding, and resource occupancy records are used to form a baseline scheduling snapshot for the next cycle. When generating resource disturbance events for the next cycle, the resource disturbance events for the next cycle and the baseline scheduling snapshot for the next cycle are used to determine the local impact domain and repair tasks for the next scheduling cycle.
10. A project task intelligent scheduling system based on limited resources, characterized in that, It includes a scheduling server, a task terminal, a resource terminal, and a project management terminal. The scheduling server communicates with the task terminal, the resource terminal, and the project management terminal through communication interfaces. The scheduling server includes a processor, memory, communication interface and scheduling database. The memory stores the project task scheduling program, and the processor executes the project task scheduling program to implement the baseline scheduling module, disturbance handling module, local repair module, recovery decision module and execution feedback module. The baseline scheduling module is used to acquire task data and resource status, and to form task constraints, baseline scheduling schemes, and baseline scheduling snapshots; The disturbance handling module is used to update the available time interval of the target resource based on the resource disturbance event, and to determine the local impact domain, the task to be repaired, and the freeze constraints. The local repair module is used to remove future execution schedules of tasks to be repaired, and to form the earliest feasible insertion scheme and the minimum offset insertion scheme according to task dependencies; The recovery decision module is used to determine the target scheduling scheme based on the number of tasks to be adjusted, the total offset at the start time, and the increment of the total project duration, and to perform bounded retries or global rescheduling when local recovery fails. The execution feedback module is used to send adjustment instructions and resource control data to the task terminal and resource terminal through the communication interface, switch the scheduling version according to the confirmation result of the receiving end, receive task and resource execution feedback, and form the next cycle baseline scheduling snapshot according to the effective scheduling scheme, task status and resource status. The scheduling database is used to store task data, resource status, baseline scheduling snapshots, local rescheduling contexts, candidate recovery schemes, scheduling versions, and execution feedback snapshots.
Citation Information
Patent Citations
Multimode resource limited project scheduling optimization method adopting two-stage genetic algorithm
CN110991917A
