A wind power equipment intelligent maintenance scheduling method and system

Abstract

The application belongs to the technical field of wind power generation operation and maintenance, and provides a wind power equipment intelligent maintenance scheduling method and system, which comprises the following steps: obtaining early warning information of a wind turbine generator and cleaning the standardized information, generating a maintenance task package for each early warning record; obtaining a scheduling scheme based on preset constraint conditions and optimization targets, wherein the scheduling scheme comprises execution time and resource allocation of each maintenance task; marking the execution state of each maintenance task, determining tasks in the issued and issued after states as locked tasks, keeping the execution time and resource allocation of the locked tasks as hard constraints, and keeping the rest as unlocked tasks; when a trigger event is monitored, only the unlocked tasks are re-optimized and solved, and are spliced with the locked tasks to obtain an updated scheduling scheme. The application solves the technical problem that existing wind power maintenance scheduling is mostly statically generated, it is difficult to dynamically adjust the plan according to the field situation after the plan is issued, and all tasks need to be recalculated when a sudden fault or resource change occurs, resulting in slow response.

Description

Technical Field

[0001] This application belongs to the field of wind power operation and maintenance technology, specifically involving a method and system for intelligent maintenance scheduling of wind power equipment. Background Technology

[0002] Wind power generation is one of the fastest-growing forms of renewable energy in recent years. With the expansion of installed capacity and the increasing service life of wind turbines, the workload of wind farm maintenance and operation is constantly rising, making maintenance and operation a crucial link in ensuring unit availability and power generation efficiency. Early maintenance primarily employed periodic preventative maintenance, with work scheduled according to fixed cycles. Subsequently, condition-based maintenance, based on the actual operating status of the equipment, has gradually become the mainstream. This involves using online monitoring, condition warnings, and fault diagnosis to assess the health status of the unit, and then scheduling appropriate maintenance when the equipment shows signs of deterioration or is close to failure. Building upon condition-based maintenance, the warning results need to be translated into actionable on-site work arrangements. This process is typically completed by maintenance scheduling, which involves comprehensively considering the urgency of the maintenance task, downtime losses, maintenance costs, and on-site conditions such as weather, personnel, and spare parts to plan the execution time and required resources for the task, generating a maintenance plan and guiding its implementation on-site.

[0003] However, existing maintenance scheduling still has some shortcomings in practical applications. Most schedules are statically generated, making it difficult to adjust them according to on-site conditions after the plan is issued. Once a sudden failure, weather change, or changes in the status of resources such as personnel, tools, and spare parts occur, it is often necessary to recalculate all tasks, resulting in a large amount of calculation and slow response, which is not conducive to dealing with the variable operating conditions at wind power sites. At the same time, actual scheduling needs to meet multiple constraints such as weather operation windows, personnel qualifications, tool and spare parts inventory, and safety procedures. However, existing methods often focus on only one or two of these constraints, lacking a unified approach to constraints such as the matching of personnel qualifications with task requirements and the occupancy relationship between different resources, resulting in a gap between the scheduling results and the actual executable conditions on site. In addition, maintenance scheduling usually involves multiple interdependent objectives such as downtime losses, operation and maintenance costs, and personnel load. Existing methods consider relatively limited objectives and have relatively simplified constraint handling, making it difficult to make trade-offs between safety, cost, and efficiency according to different scenarios. Summary of the Invention

[0004] This application addresses the technical problems of existing wind power maintenance schedules being mostly statically generated, making it difficult to dynamically adjust them according to on-site conditions after the plan is issued, and requiring the recalculation of all tasks in the event of sudden failures or resource changes, resulting in slow response. It provides a method and system for intelligent maintenance scheduling of wind power equipment.

[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, a method for intelligent maintenance scheduling of wind power equipment includes: Acquire early warning information from wind turbines, perform data cleaning and standardization on the early warning information, and generate a maintenance task package for each early warning record in the early warning information, with each maintenance task package corresponding to a maintenance task. Based on preset constraints and optimization objectives, the maintenance task package is optimized and solved to obtain a scheduling scheme, which includes the execution time and resource allocation of each maintenance task. Each maintenance task in the scheduling scheme is marked with an execution status, which includes the status before issuance, the status after issuance, and the status after issuance. Maintenance tasks in the scheduling scheme that are in the status before issuance or the status after issuance are identified as locked tasks, and maintenance tasks that are in the status before issuance are identified as unlocked tasks. The execution time and resource allocation of the locked tasks in the scheduling scheme are kept unchanged as hard constraints. When a triggering event is detected during the execution of the scheduling scheme, the execution time and resource allocation of the locked task are used as hard constraints to optimize the solution of the unlocked task, thereby obtaining the updated execution time and resource allocation of the unlocked task. The updated scheduling scheme is then concatenated with the execution time and resource allocation of the locked task.

[0006] In some implementations, the constraints include at least one of weather window constraints, personnel qualification constraints, tool and spare parts inventory constraints, and safety procedure constraints; the optimization objectives include at least one of downtime losses, operation and maintenance costs, and personnel man-hour balance.

[0007] In some implementations, optimizing the maintenance task package based on preset constraints and optimization objectives includes: The maintenance task package is optimized using a non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set consisting of several scheduling schemes, and a scheduling scheme is determined from the Pareto optimal solution set. The non-dominated sorting genetic algorithm uses a two-layer chromosome encoding, where the upper layer encoding is the sorting sequence of each maintenance task, and the lower layer encoding is the allocation relationship between resources and each maintenance task.

[0008] In some implementations, the execution state includes a pending execution state, a issued state, an executing state, a paused state, a completed state, and a cancelled state; wherein, the pending execution state belongs to the state before the issued state, and the executing state, the paused state, the completed state, and the cancelled state belong to the state after the issued state.

[0009] In some implementations, the triggering event includes at least one of task insertion, task cancellation, resource change, job window change, and execution deviation.

[0010] In some implementations, after concatenating the updated execution time and resource allocation of the unlocked task with the execution time and resource allocation of the locked task to obtain the updated scheduling scheme, the method further includes: The updated scheduling scheme is subjected to conflict detection, which includes resource conflicts where the same resource is assigned to multiple maintenance tasks at the same time. When a resource conflict is detected, the unlocked tasks are re-optimized and solved until the updated scheduling scheme is conflict-free.

[0011] Secondly, a smart maintenance scheduling system for wind power equipment includes: The task package generation module is used to obtain early warning information of wind turbine units, perform data cleaning and standardization on the early warning information, and generate a maintenance task package for each early warning record in the early warning information, with each maintenance task package corresponding to a maintenance task. The scheduling solution module is used to optimize and solve the maintenance task package based on preset constraints and optimization objectives to obtain a scheduling scheme, which includes the execution time and resource allocation of each maintenance task. The status locking module is used to mark the execution status of each maintenance task in the scheduling scheme. The execution status includes the pre-issued status, the issued status, and the issued status. Maintenance tasks in the issued status and the issued status in the scheduling scheme are identified as locked tasks, and maintenance tasks in the pre-issued status are identified as unlocked tasks. The execution time and resource allocation of the locked tasks in the scheduling scheme are kept unchanged as hard constraints. The incremental reordering module is used to optimize the unlocked tasks when a triggering event is detected during the execution of the scheduling scheme, using the execution time and resource allocation of the locked tasks as hard constraints, to obtain the updated execution time and resource allocation of the unlocked tasks, and then concatenate them with the execution time and resource allocation of the locked tasks to obtain the updated scheduling scheme.

[0012] Thirdly, a computer device includes: a processor and a computer-readable storage medium; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the intelligent maintenance scheduling method for wind power equipment as described above.

[0013] Fourthly, a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed as described in the intelligent maintenance scheduling method for wind power equipment.

[0014] Fifthly, a computer program product comprising a computer program that, when executed by a processor, implements the intelligent maintenance scheduling method for wind power equipment as described above.

[0015] Compared with the prior art, this application has the following beneficial effects: This application first processes the early warning information into maintenance task packages and solves for a scheduling scheme. Then, it marks the execution status of each maintenance task, identifying tasks in the "already issued" or later status as locked tasks, with their execution time and resource allocation remaining unchanged as hard constraints. When a trigger event is detected, only the unlocked tasks are re-solved and then combined with the locked tasks. Therefore, when a sudden failure or resource change occurs after the plan is issued, it is not necessary to recalculate all tasks. The already issued and executed parts are preserved as is, and only the unexecuted parts are affected. The scale and scope of recalculation are compressed to the small portion of unlocked tasks. The locking mechanism maintains the stability of the issued plan, and incremental rescheduling allows the scheme to be updated locally as needed during execution, resulting in minimal recalculation and fast response.

[0016] Furthermore, a non-dominated sorting genetic algorithm is used to obtain a set of Pareto optimal solutions, from which a solution is selected. A two-layer chromosome encoding is employed, with the upper layer representing the task sorting sequence and the lower layer representing the resource allocation relationship. Faced with multiple conflicting objectives such as downtime losses, costs, and personnel load, the Pareto solution set provides a set of mutually non-dominated compromise solutions, facilitating selection based on actual needs rather than being locked onto a single solution. The two-layer encoding allows task timing and resource allocation to be determined collaboratively in the same solution process, improving the rationality of the solution and the synergy of resource allocation under multi-objective conflicts.

[0017] Furthermore, after the updated scheduling plan is obtained through splicing, a conflict detection is performed again, focusing on resource conflicts where the same resource is allocated to multiple tasks at the same time. Once a conflict is found, the unlocked tasks are re-solved until the plan is conflict-free, ensuring that the final delivered plan is self-consistent and executable in terms of resource usage, and preventing conflicts generated during splicing from being directly distributed to the site. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart of an intelligent maintenance scheduling method for wind power equipment provided in this application embodiment; Figure 2 A structural diagram of an intelligent maintenance scheduling system for wind power equipment provided in this application embodiment; Figure 3 A detailed architecture diagram of an intelligent maintenance scheduling system for wind power equipment provided in this application embodiment; Figure 4 A computer device diagram provided for an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] In one embodiment of this application, such as Figure 1 As shown, a method for intelligent maintenance scheduling of wind power equipment is provided, including: S1, acquire the early warning information of the wind turbine, perform data cleaning and standardization on the early warning information, and generate a maintenance task package for each early warning record in the early warning information, with each maintenance task package corresponding to a maintenance task; This step provides processing objects for subsequent scheduling. The early warning information reflects the health status or fault risk of the wind turbine. In one embodiment of this application, the early warning information comes from one or more of the following: the wind turbine's Supervisory Control and Data Acquisition (SCADA) system, Condition Monitoring System (CMS) system, fault risk scoring results, and maintenance decision results. For example, early warnings of operating parameters exceeding limits output by the SCADA system, vibration or temperature anomaly warnings output by the CMS system, fault risk scores output by the risk assessment model, and maintenance decision results output by the maintenance decision model can all be used as the early warning information.

[0022] Because the early warning information from different sources often differs in data format, field naming, units, and time base, data cleaning and standardization are performed on the early warning information. In one embodiment of this application, the data cleaning includes removing at least one of duplicate records, missing values, and obvious outliers, and the standardization includes unifying the early warning information from different sources into a preset field format and encoding rules. Those skilled in the art will understand that the specific methods of data cleaning and standardization do not constitute a limitation of this application.

[0023] After cleaning and standardization, each early warning record in the early warning information is generated into a maintenance task package. Each maintenance task package corresponds to one maintenance task, that is, one early warning record corresponds to one maintenance task that needs to be performed. The maintenance task package includes at least one of the following: fault early warning information corresponding to the maintenance task, risk score level, maintenance task priority, estimated maintenance task time, list of resources required for the maintenance task, and standard operating procedure. The maintenance task priority is used to characterize the urgency of the task, for example, it is divided into 1 to 5 levels, with level 1 being the highest; the resource list required for the maintenance task includes at least one of the personnel, vehicles and tools required for the maintenance task. The specific information in the above maintenance task package provides a basis for optimization solutions, resource allocation and status marking in subsequent steps, but is not intended to limit this application.

[0024] Taking a specific application scenario as an example, the accessed early warning information includes: SCADA system early warning information, which states that the temperature of the non-drive end bearing of the generator is abnormal, the early warning level is 2, the wind farm is A, and the wind turbine number is #05; risk assessment results, including severity S=4, maintenance difficulty D=3, evolution risk E=2, and economic impact L=3; and maintenance decision results, which state that the bearing will be replaced, with an estimated time of 18 hours and required resources of 1 technician, 2 maintenance workers, 1 engineering vehicle, and 1 set of spare bearings. After cleaning and standardizing the above early warning information, a maintenance task package corresponding to the bearing replacement maintenance task is generated. This maintenance task package contains information such as the estimated time and the list of required resources.

[0025] S2, Based on preset constraints and optimization objectives, optimize and solve the maintenance task package to obtain a scheduling scheme, which includes the execution time and resource allocation of each maintenance task; This step solves the maintenance task package obtained in S1 to obtain an initial scheduling scheme. The constraints are the limitations that the maintenance tasks must meet when they are executed, and the optimization objective is the indicator that the scheduling scheme aims to optimize.

[0026] The constraints include at least one of the following: weather window constraints, personnel qualification constraints, tool and spare parts inventory constraints, and safety procedure constraints.

[0027] The weather window constraint is used to limit maintenance tasks to be performed only during periods that meet the operating conditions, such as wind speed not exceeding the safe operating threshold and no severe weather such as thunderstorms or heavy rain. When high-altitude operations are involved, humidity and icing risk can be further limited. The data can be obtained from wind speed, precipitation, and temperature forecasts provided by the weather service interface.

[0028] For each time slot t Define meteorological feasibility function W(t) , W(t) ∈{0,1}, W(t) =1 indicates that the time slot is available for operation. W(t) =0 indicates that the time slot cannot be used.

[0029] Personnel qualification constraints are used to limit maintenance tasks to personnel with the appropriate roles and qualifications, and the same person can only be assigned to one maintenance task at a time; the data can be derived from a personnel status database that includes skill certificates, tasks in transit, and accumulated work hours. For maintenance tasks... j The required set of roles is R j When personnel p Having a role r Required Qualifications q p,r =1, otherwise record q p,r =0, and people at the same time p It can only be assigned to one maintenance task.

[0030] The tool and spare parts inventory constraint limits the number of vehicles, tools, and spare parts required for maintenance tasks to no more than the currently available inventory, and may take into account the return time of tools after use; the data may be obtained from a warehouse management system. In one embodiment of this application, for tools... k Maintenance task j The number occupied is n j,k Then each maintenance task requires tools k The total number of tools occupied does not exceed k Inventory S k .

[0031] Safety regulations are used to restrict the execution of maintenance tasks to comply with the requirements of standard operating procedures. For example, specific tasks must be performed during the day, hot work must be applied for in advance, and multi-person collaborative tasks must ensure that personnel arrive at the same time. The data can be obtained from the standard operating procedure library.

[0032] The optimization objectives include at least one of the following: downtime losses, operation and maintenance costs, and manpower hour balance.

[0033] The downtime losses In one embodiment of this application, the power generation benefits lost by the wind turbine during maintenance shutdown are represented as follows:

[0034] in, J A collection of maintenance tasks. In order to carry out maintenance tasks jThe product of wind turbine power and electricity price corresponding to the wind speed during the shutdown period, wherein the wind turbine power is determined by the power-wind speed characteristic curve of the wind turbine unit. For wind turbine d During maintenance tasks j The downtime in the system.

[0035] The operation and maintenance costs Including at least one of the following: labor costs and vehicle usage fees, as expressed in one embodiment of this application:

[0036] in Hourly rates for employees For personnel p In the mission j The time spent in the process; For vehicle usage rates, For vehicles v In the mission j The time spent in the process.

[0037] The balance of employee hours In one embodiment of this application, the standard deviation of the cumulative working hours of each person is used to characterize the balance of their cumulative working hours.

[0038] in, Total number of personnel For personnel p Total working hours The average total working hours for all personnel; The smaller the value, the more balanced the workload of the personnel.

[0039] The optimization objective also includes maintenance task priority. Accordingly, a four-dimensional objective function is constructed, comprising task priority, downtime loss, maintenance cost, and load balancing dimensions, with weight coefficients assigned to each dimension. Wherein: The objective function for the task priority dimension is:

[0040] in For maintenance tasks j priority, For maintenance tasks j Start time; The objective function for the downtime loss dimension is:

[0041] The objective function for the operation and maintenance cost dimension is:

[0042] The objective function for load balancing is:

[0043] The overall optimization objective is to minimize the vector:

[0044] in , , , These are the weighting coefficients for the four dimensions mentioned above, which are determined by the wind farm based on its actual situation and its own needs, in order to achieve different preferences for safety, cost, and efficiency.

[0045] In one embodiment of this application, the specific method for optimizing the maintenance task package includes: using a non-dominated sorting genetic algorithm to optimize the maintenance task package, obtaining a Pareto optimal solution set consisting of several scheduling schemes, and determining a scheduling scheme from the Pareto optimal solution set. The non-dominated sorting genetic algorithm employs a two-layer chromosome encoding, where the upper layer encoding is the sorting sequence of each maintenance task, and the lower layer encoding is the allocation relationship between resources and each maintenance task.

[0046] In one embodiment of this application, the execution process of the non-dominated sorting genetic algorithm includes: generating a partial initial solution using heuristic rules and randomly generating the remaining individuals to form an initial population, wherein the heuristic rules are, for example, the highest priority priority or the minimum shutdown loss priority; performing non-dominated sorting of individuals based on Pareto dominance, and using a dynamic reference point mechanism to adaptively adjust the position of the reference point according to the target value distribution of the current population; performing genetic operations of selection, crossover, and mutation on individuals, wherein selection uses tournament selection combined with reference point association selection, crossover uses partial mapping crossover for the upper-level task sorting sequence and uniform crossover for the lower-level resource allocation relationship, and mutation includes task sequence exchange mutation and resource redistribution mutation; adding a penalty term to the objective function for individuals that violate the constraints; terminating the iteration and outputting the Pareto optimal solution set when the maximum number of iterations (e.g., 500 generations) is reached or the Pareto front has not improved within a consecutive preset number of generations (e.g., 50 consecutive generations). Those skilled in the art will understand that the above-mentioned initial solution generation method, genetic operation method, constraint processing method, termination condition and its example values ​​do not constitute a limitation on this application, and other methods that can obtain a scheduling scheme based on the above-mentioned constraints and optimization objectives are also within the protection scope of this application.

[0047] In one embodiment of this application, each scheduling scheme in the Pareto optimal solution set includes: an execution time sequence for each maintenance task, the execution time sequence including the start and end times of each maintenance task; an allocation scheme for personnel, vehicles, and tools; and the four-dimensional objective function value corresponding to the scheduling scheme. The scheduling scheme includes the execution time and resource allocation for each maintenance task, the execution time including the start and end times of the maintenance task, and the resource allocation including at least one of personnel, vehicles, and tools allocated to each maintenance task.

[0048] Following the aforementioned bearing replacement application scenario, this step automatically analyzes weather data to determine that the wind speed will not exceed 8 m / s for the next 3 days, satisfying the weather window constraint; it queries the personnel status database to identify that the technician has no on-the-go tasks for the day and the corresponding maintenance worker can be released in the afternoon, satisfying the personnel qualification constraint; it queries the spare parts database to confirm that there is sufficient stock of the same model of bearing, satisfying the tool and spare parts inventory constraint; based on this, the non-dominated sorting genetic algorithm is used for optimization to obtain the scheduling scheme for the bearing replacement maintenance task. The scheduling scheme includes the execution time of the maintenance task and the technicians, maintenance workers, and engineering vehicles assigned to it.

[0049] S3, mark the execution status of each maintenance task in the scheduling scheme. The execution status includes the state before issuance, the state after issuance, and the state after issuance. The maintenance tasks in the scheduling scheme that are in the state after issuance and the state after issuance are identified as locked tasks, and the maintenance tasks that are in the state before issuance are identified as unlocked tasks. The execution time and resource allocation of the locked tasks in the scheduling scheme are kept unchanged as hard constraints. This step marks the execution status of each maintenance task in the scheduling scheme obtained in S2, and divides the tasks into locked and unlocked tasks accordingly, so as to determine which tasks can be adjusted and which tasks must be fixed for the incremental rescheduling in S4.

[0050] The execution status is used to characterize the stage of the maintenance task during its issuance and execution. In one embodiment of this application, the execution status includes a pending execution status, an issued status, an executing status, a paused status, a completed status, and a cancelled status; wherein, the pending execution status indicates that the maintenance task has not yet been issued, the issued status indicates that the maintenance task has been issued but has not yet started execution, the executing status indicates that the maintenance task is being executed, the paused status indicates that the maintenance task has been suspended due to sudden weather changes or other reasons, the completed status indicates that the maintenance task has been completed, and the cancelled status indicates that the maintenance task has been cancelled. The pending execution status belongs to the status before the issued status; the executing status, paused status, completed status, and cancelled status belong to the status after the issued status.

[0051] Maintenance tasks in the scheduling scheme that are in the "issued" state or later are identified as locked tasks, while maintenance tasks that are in the "pre-issued" state are identified as unlocked tasks. That is, once a maintenance task enters the "issued" state or later, it is identified as a locked task; maintenance tasks that are still in the "pre-issued" state are unlocked tasks.

[0052] The execution time and resource allocation of the locked task in the scheduling scheme are kept unchanged as hard constraints. That is, the execution time and resource allocation of the locked task, which have been determined in the scheduling scheme obtained in S2, remain fixed in the subsequent optimization process and will not be reallocated. In this way, maintenance tasks that have been issued or are being executed will not be changed due to subsequent rescheduling, ensuring the stability of the issued plan.

[0053] Taking the aforementioned bearing replacement application scenario as an example, the scheduling scheme obtained from S2 is pushed to the display platform, and the task bar of the maintenance task is bound to the assigned engineering vehicle on the Gantt chart. After the maintenance task is issued, its execution status changes from pending execution to issued, and it is correspondingly identified as a locked task. Its execution time in the scheduling scheme and the assigned technician, maintenance worker, and engineering vehicle remain unchanged as hard constraints.

[0054] S4. When a triggering event is detected during the execution of the scheduling scheme, the execution time and resource allocation of the locked task are used as hard constraints to optimize the solution of the unlocked task, so as to obtain the updated execution time and resource allocation of the unlocked task, and then concatenate it with the execution time and resource allocation of the locked task to obtain the updated scheduling scheme.

[0055] In this step, during the execution of the scheduling plan, the unlocked tasks are incrementally rearranged based on changes in the field to obtain an updated scheduling plan, thereby achieving dynamic adjustment of the plan without recalculating all maintenance tasks.

[0056] The triggering event is a field event that causes the scheduling scheme to need adjustment. In one embodiment of this application, the triggering event includes at least one of task insertion, task cancellation, resource change, job window change, and execution deviation; wherein, task insertion is the addition of a new maintenance task, such as a new maintenance task generated by a sudden failure; task cancellation is the cancellation of an existing maintenance task, such as a fault self-recovery or a decision change; resource change is a change in the availability of resources, such as personnel taking leave, tool damage, or spare parts shortage; job window change is a change in the feasible time period for the job, such as a sudden weather change or a power grid dispatch requiring a shutdown; and execution deviation is the deviation of the actual execution of the maintenance task from the scheduling scheme, such as the actual time taken for the maintenance task exceeding the expected time. In one embodiment of this application, the above-mentioned triggering events are monitored in real time through an application programming interface.

[0057] When the triggering event is detected, the execution time and resource allocation of the locked task are used as hard constraints to optimize the solution for the unlocked task. In one embodiment of this application, the execution time and resources occupied by the locked task are marked as unavailable, and optimization is performed only on the unlocked task to obtain the updated execution time and resource allocation of the unlocked task. The optimization solution can use the same non-dominated sorting genetic algorithm as S2, and the number of iterations can be shortened (e.g., 100 generations) to speed up the response. The updated execution time and resource allocation of the unlocked task are concatenated with the execution time and resource allocation of the locked task in the scheduling scheme to obtain the updated scheduling scheme. Since the locked task remains unchanged and only the unlocked task participates in the recalculation, the scope and computational amount of the recalculation are limited to the unlocked task, resulting in a faster response compared to recalculating all maintenance tasks.

[0058] The input to the incremental reordering includes a locked task set, an unlocked task set, and the current resource status, which includes personnel occupancy and tool return times. The output of the incremental reordering is an updated scheduling scheme. Specifically, the incremental reordering includes the following steps: marking the execution time and resources occupied by each maintenance task in the locked task set as unavailable; using the marked resource status as a constraint, optimizing the unlocked task set using the non-dominated sorting genetic algorithm and shortening the iteration number of the non-dominated sorting genetic algorithm to obtain the updated execution time and resource allocation of the unlocked tasks; concatenating the updated execution time and resource allocation of the unlocked tasks with the execution time and resource allocation of the locked tasks; performing conflict detection on the concatenated scheduling scheme, and backtracking adjustment when a conflict is detected, re-optimizing the unlocked tasks; and outputting a conflict-free updated scheduling scheme.

[0059] Furthermore, in one embodiment of this application, after obtaining the updated scheduling scheme, the method further includes: performing conflict detection on the updated scheduling scheme, and re-optimizing the solution for the unlocked tasks when a conflict is detected, until the updated scheduling scheme is conflict-free. The conflict detection includes resource conflict detection, where the same resource is assigned to multiple maintenance tasks within the same time period; in one embodiment of this application, the conflict detection also includes time conflict detection, where the same wind turbine is assigned to multiple maintenance tasks within the same time period. This avoids conflicts arising from overlapping resource usage or time periods when unlocked and locked tasks are combined, ensuring that the updated scheduling scheme is self-consistent and executable in terms of resource usage and time period arrangement.

[0060] Furthermore, in one embodiment of this application, each time an updated scheduling plan is obtained, the update result is written to the database in a transactional manner. If the write fails, it is rolled back to avoid data corruption. The uniqueness of the identifier of the newly generated maintenance task is verified, for example, by using the snowflake algorithm to generate a globally unique 64-bit identifier to avoid conflicts between the identifier and historical identifiers. In one embodiment of this application, the database includes three relational databases for storing task flow information, scheduling segment information, and personnel status information, respectively.

[0061] Furthermore, in one embodiment of this application, the method further includes parsing the scheduling scheme or the updated scheduling scheme into maintenance work orders and outputting them, and visually displaying the scheduling scheme or the updated scheduling scheme in the form of a Gantt chart. The maintenance work order includes at least one of the following: basic information of the corresponding maintenance task, resource allocation details, standard operating procedures, and safety measures, and may include a QR code or barcode for on-site scanning confirmation. The maintenance work order can be output in JSON, PDF, or Excel format; wherein, the Excel format report may include resource load statistics and task completion rate indicators. The Gantt chart uses time as the horizontal axis and wind turbine units as the vertical axis, with each maintenance task displayed as a bar. The color of the bar can be used to represent the priority of the maintenance task, and supports switching between monthly, weekly, and daily views. In one embodiment of this application, the visualization also includes a load view, which includes a bar chart showing the percentage of working hours for each person within a preset time period, and a timeline showing the occupancy of key tools.

[0062] In one embodiment of this application, on-site feedback on the maintenance task is also received, and the execution status of the maintenance task is updated based on the feedback. The feedback methods include at least one of the following: reporting the start, end, actual working hours, and consumed spare parts of the maintenance task via mobile terminal scanning; automatically triggering RFID scanning and updating inventory status when tools are returned via IoT; and manually adjusting the status of the maintenance task. The corresponding processing logic includes: if the maintenance task is completed on time, its execution status is updated to "completed" and the resources it occupies are released; if the actual execution of the maintenance task deviates from the scheduling plan, for example, the actual time exceeds the expected time, it is used as a trigger event to trigger incremental rescheduling in this step; if the maintenance task fails midway, an anomaly report is generated and manual intervention is performed. Thus, the on-site execution status is transmitted back in real time to drive scheduling updates, forming a closed loop between scheduling and execution.

[0063] Taking the aforementioned bearing replacement application scenario as an example, after the maintenance task is issued, it is monitored in real time during its execution. When a tool failure causes a delay, this execution deviation is used as the trigger event to trigger incremental rescheduling. While keeping the locked tasks unchanged, only the remaining unlocked tasks are re-optimized and rescheduled to subsequent idle periods, resulting in an updated scheduling scheme. The final output includes: the maintenance task is the replacement of the generator bearing of wind turbine #05 in wind farm A; the operation time is the corresponding scheduled period; the resource allocation is the corresponding technicians, maintenance workers, and engineering vehicles; and the safety measures are the corresponding number of standard operating procedure lists.

[0064] Through the above S1 to S4, this embodiment does not require recalculating all maintenance tasks when encountering sudden failures or resource changes after the plan is issued. Instead, it locks the issued and subsequent maintenance tasks as hard constraints and only performs incremental rescheduling on the unlocked maintenance tasks. This achieves dynamic adjustment of the scheduling scheme while maintaining the stability of the issued plan. The recalculation is small and the response is rapid. It solves the technical problems of existing wind power maintenance schedules being mostly statically generated, difficult to dynamically adjust according to the site conditions after the plan is issued, and requiring recalculation of all tasks in the event of sudden failures or resource changes, resulting in slow response.

[0065] In one embodiment of this application, such as Figure 2 As shown, a smart maintenance scheduling system for wind power equipment is provided, including: The task package generation module is used to obtain early warning information of wind turbine units, perform data cleaning and standardization on the early warning information, and generate a maintenance task package for each early warning record in the early warning information, with each maintenance task package corresponding to a maintenance task. The scheduling solution module is used to optimize and solve the maintenance task package based on preset constraints and optimization objectives to obtain a scheduling scheme, which includes the execution time and resource allocation of each maintenance task. The status locking module is used to mark the execution status of each maintenance task in the scheduling scheme. The execution status includes the pre-issued status, the issued status, and the issued status. Maintenance tasks in the issued status and the issued status in the scheduling scheme are identified as locked tasks, and maintenance tasks in the pre-issued status are identified as unlocked tasks. The execution time and resource allocation of the locked tasks in the scheduling scheme are kept unchanged as hard constraints. The incremental reordering module is used to optimize the unlocked tasks when a triggering event is detected during the execution of the scheduling scheme, using the execution time and resource allocation of the locked tasks as hard constraints, to obtain the updated execution time and resource allocation of the unlocked tasks, and then concatenate them with the execution time and resource allocation of the locked tasks to obtain the updated scheduling scheme.

[0066] In one embodiment of this application, such as Figure 3 As shown, the overall system architecture on which the above-mentioned intelligent maintenance scheduling method for wind power equipment is based is provided. The architecture includes, from bottom to top, a data source layer, a data access and processing layer, a core engine layer, a data middleware layer, and an output and display layer.

[0067] The data source layer provides various basic data required for the aforementioned early warning information and scheduling, including SCADA early warning information, CMS early warning information, risk assessment data, maintenance decision data, weather service API, warehouse management system, and personnel qualification database. Specifically, SCADA early warning information, CMS early warning information, risk assessment data, and maintenance decision data constitute the early warning information sources in the aforementioned method; the weather service API provides data for weather window constraints; the warehouse management system provides data for tool and spare parts inventory constraints; and the personnel qualification database provides data for personnel qualification constraints.

[0068] The data access and processing layer is located above the data source layer and is used to access and process the data provided by the data source layer. It includes a multi-source heterogeneous data access interface, a data cleaning and standardization module, and a maintenance task package generator. The multi-source heterogeneous data access interface uses MQTT or RESTful methods to access various types of data from the data source layer. The data cleaning and standardization module cleans and standardizes the accessed early warning information. The maintenance task package generator generates a maintenance task package from each cleaned and standardized early warning record. This layer implements the function of S1 in the aforementioned method.

[0069] The core engine layer, located above the data access and processing layer, is used to complete constraint parsing, scheduling, and state tracking. It includes a multi-constraint parsing module, an intelligent scheduling engine, and a dynamic state tracking module. The multi-constraint parsing module parses constraints, including weather window parsing, personnel qualification verification, inventory checks, and safety procedure mapping. The intelligent scheduling engine incorporates an improved NSGA-III algorithm, including sub-modules such as encoders, fitness calculations, non-dominated sorting, and genetic operations. It optimizes maintenance task packages based on the constraints and optimization objectives to obtain a scheduling scheme. The dynamic state tracking module includes state marking, a locking mechanism, and incremental reordering triggers. It marks the execution status of maintenance tasks, locks maintenance tasks in the issued or subsequent states, and triggers incremental reordering of unlocked tasks when a trigger event is detected. This layer implements the functions of S2, S3, and S4 in the aforementioned method.

[0070] The data platform layer, located above the core engine layer, provides data storage and status synchronization for each layer. It includes three relational databases using SQLite: a task flow database, a scheduling fragment database, and a personnel status database. The task flow database stores the flow information of maintenance tasks, the scheduling fragment database stores scheduling schemes and their update results, and the personnel status database stores personnel status information.

[0071] The output and display layer is located above the data platform layer and is used to output and display scheduling results and provide external interfaces, including a visualization platform, a work order generator, and an API interface. The visualization platform is used to generate Gantt charts and load views; the work order generator is used to generate maintenance work orders in JSON, PDF, or Excel format; and the API interface is available for use by mobile apps and third-party systems.

[0072] In the data flow direction, the data from the data source layer is used by the data access and processing layer to generate maintenance task packages. The core engine layer completes constraint parsing, scheduling solution and status tracking, and the data is stored and synchronized by the data middle platform layer. Then, the output and display layer outputs maintenance work orders, Gantt charts, etc. and provides interfaces to the outside world. At the same time, the execution status on site is returned to the dynamic status tracking module of the core engine layer through the output and display layer to trigger incremental rescheduling of unlocked tasks, thereby forming a closed loop between scheduling and execution in the system.

[0073] For specific limitations regarding the intelligent maintenance scheduling system for wind power equipment, please refer to the limitations of the intelligent maintenance scheduling method for wind power equipment described above. The corresponding technical effects can be obtained equally and will not be repeated here. Each module in the aforementioned intelligent maintenance scheduling system for wind power equipment can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0074] Figure 4 An internal structural diagram of a computer device is shown in one embodiment. This computer device may specifically be a terminal or a server. Figure 4 As shown, the computer device includes a processor, memory, network interface, display, camera, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for intelligent maintenance scheduling of wind power equipment. The display screen can be an LCD screen or an e-ink display screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0075] As will be understood by those skilled in the art, computer equipment Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. Specific computing devices may include more or fewer components than those shown in the figure, or combine certain components, or have the same component arrangement.

[0076] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0077] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0078] In summary, the intelligent maintenance scheduling method, system, computer equipment, and storage medium for wind power equipment provided in this application obtain early warning information from wind turbines and generate maintenance task packages after cleaning and standardization. Based on preset constraints and optimization objectives, the maintenance task packages are optimized to obtain a scheduling scheme. Then, each maintenance task is marked with an execution status, and maintenance tasks in the issued or subsequent issued states are identified as locked tasks, with their execution time and resource allocation kept unchanged as hard constraints. When a trigger event is detected, only the unlocked tasks are optimized and solved, and then combined with the locked tasks to obtain an updated scheduling scheme. In this way, when a sudden failure or resource change occurs after the plan is issued, it is not necessary to recalculate all maintenance tasks. Instead, while maintaining the stability of the issued plan, only the unlocked tasks are incrementally rearranged, thereby realizing dynamic adjustment of the scheduling scheme. The recalculation is small and the response is rapid. This solves the technical problems of existing wind power maintenance schedules being mostly statically generated, difficult to dynamically adjust according to the site conditions after the plan is issued, and requiring recalculation of all tasks in the event of a sudden failure or resource change, resulting in slow response.

[0079] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0080] The above-described embodiments are merely preferred embodiments of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. A method for intelligent maintenance scheduling of wind power equipment, characterized in that, include: Acquire early warning information from wind turbines, perform data cleaning and standardization on the early warning information, and generate a maintenance task package for each early warning record in the early warning information, with each maintenance task package corresponding to a maintenance task. Based on preset constraints and optimization objectives, the maintenance task package is optimized and solved to obtain a scheduling scheme, which includes the execution time and resource allocation of each maintenance task. Each maintenance task in the scheduling scheme is marked with an execution status, which includes the status before issuance, the status after issuance, and the status after issuance. Maintenance tasks in the scheduling scheme that are in the status before issuance or the status after issuance are identified as locked tasks, and maintenance tasks that are in the status before issuance are identified as unlocked tasks. The execution time and resource allocation of the locked tasks in the scheduling scheme are kept unchanged as hard constraints. When a triggering event is detected during the execution of the scheduling scheme, the execution time and resource allocation of the locked task are used as hard constraints to optimize the solution of the unlocked task, thereby obtaining the updated execution time and resource allocation of the unlocked task. The updated scheduling scheme is then concatenated with the execution time and resource allocation of the locked task.

2. The intelligent maintenance scheduling method for wind power equipment according to claim 1, characterized in that, The constraints include at least one of the following: weather window constraints, personnel qualification constraints, tool and spare parts inventory constraints, and safety procedure constraints; the optimization objectives include at least one of the following: downtime losses, operation and maintenance costs, and personnel man-hour balance.

3. The intelligent maintenance scheduling method for wind power equipment according to claim 1, characterized in that, The optimization solution for the maintenance task package based on preset constraints and optimization objectives includes: The maintenance task package is optimized using a non-dominated sorting genetic algorithm to obtain a Pareto optimal solution set consisting of several scheduling schemes, and a scheduling scheme is determined from the Pareto optimal solution set. The non-dominated sorting genetic algorithm uses a two-layer chromosome encoding, where the upper layer encoding is the sorting sequence of each maintenance task, and the lower layer encoding is the allocation relationship between resources and each maintenance task.

4. The intelligent maintenance scheduling method for wind power equipment according to claim 1, characterized in that, The execution status includes pending execution status, issued status, executing status, paused status, completed status, and canceled status; wherein, the pending execution status belongs to the status before issued status, and the executing status, paused status, completed status, and canceled status belong to the status after issued status.

5. The intelligent maintenance scheduling method for wind power equipment according to claim 1, characterized in that, The triggering events include at least one of the following: task insertion, task cancellation, resource change, job window change, and execution deviation.

6. The intelligent maintenance scheduling method for wind power equipment according to claim 1, characterized in that, After concatenating the updated execution time and resource allocation of the unlocked tasks with the updated execution time and resource allocation of the locked tasks to obtain the updated scheduling scheme, the process further includes: The updated scheduling scheme is subjected to conflict detection, which includes resource conflicts where the same resource is assigned to multiple maintenance tasks at the same time. When a resource conflict is detected, the unlocked tasks are re-optimized and solved until the updated scheduling scheme is conflict-free.

7. A smart maintenance scheduling system for wind power equipment, characterized in that, include: The task package generation module is used to obtain early warning information of wind turbine units, perform data cleaning and standardization on the early warning information, and generate a maintenance task package for each early warning record in the early warning information, with each maintenance task package corresponding to a maintenance task. The scheduling solution module is used to optimize and solve the maintenance task package based on preset constraints and optimization objectives to obtain a scheduling scheme, which includes the execution time and resource allocation of each maintenance task. The status locking module is used to mark the execution status of each maintenance task in the scheduling scheme. The execution status includes the pre-issued status, the issued status, and the issued status. Maintenance tasks in the issued status and the issued status in the scheduling scheme are identified as locked tasks, and maintenance tasks in the pre-issued status are identified as unlocked tasks. The execution time and resource allocation of the locked tasks in the scheduling scheme are kept unchanged as hard constraints. The incremental reordering module is used to optimize the unlocked tasks when a triggering event is detected during the execution of the scheduling scheme, using the execution time and resource allocation of the locked tasks as hard constraints, to obtain the updated execution time and resource allocation of the unlocked tasks, and then concatenate them with the execution time and resource allocation of the locked tasks to obtain the updated scheduling scheme.

8. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements a wind power equipment intelligent maintenance scheduling method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 6, a method for intelligent maintenance scheduling of wind power equipment.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements a wind power equipment intelligent maintenance scheduling method as described in any one of claims 1 to 6.

Images