A maintenance scheduling method and system based on operations research optimization model

Abstract

This invention provides a maintenance scheduling method and system based on an operations research optimization model, relating to the field of computer technology. The method includes: acquiring multiple maintenance tasks and candidate time windows corresponding to each maintenance task; using an operations research optimization model, with the goal of minimizing the total loss associated with executing each maintenance task, determining a target time window from the multiple candidate time windows; the operations research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delays, and a fourth loss related to resource usage costs; generating a maintenance scheduling result based on the target time window corresponding to each maintenance task. This implementation can reduce revenue losses caused by maintenance and reduce scheduling conflicts, improving the executability of the target time window.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a maintenance scheduling method and system based on an operations research optimization model. Background Technology

[0002] New energy power plants (such as wind farms and photovoltaic power plants) require various types of planned maintenance during operation, including daily, monthly, quarterly, and annual inspections. However, not all maintenance work can be carried out during normal operation of the power plant, and the losses to the power plant vary depending on when maintenance is carried out. Therefore, determining when to carry out maintenance has become an urgent technical problem to be solved.

[0003] In existing technologies, maintenance scheduling typically relies on the experience of maintenance personnel, fixed-cycle plans, or simple single-task window assessments, lacking a joint scheduling optimization mechanism for multiple maintenance tasks. Furthermore, the assessment of single-task time windows generally relies on predicted electricity prices and predicted power to calculate the downtime revenue loss, using minimizing this loss as the objective for determining the maintenance window. This fails to comprehensively consider factors commonly encountered in the actual production and operation of renewable energy power plants, such as contractual risks, the impact of maintenance delays, resource usage costs, concurrent downtime restrictions, mutual exclusion relationships, and task dependencies. Consequently, it is difficult to generate an overall maintenance scheduling result that matches the actual production and operation scenario. Therefore, the maintenance windows determined by existing methods often fail to meet the actual production needs of renewable energy power plants, leaving the problem of maintenance being impossible to achieve. Summary of the Invention

[0004] This invention provides a maintenance scheduling method and system based on an operations research optimization model to address the problems in existing technologies, such as the lack of joint scheduling optimization for multiple maintenance tasks and the inability to obtain executable and implementable maintenance scheduling results. Specifically, by considering the first loss related to power generation revenue, the second loss related to contractual agreements, the third loss related to maintenance delays, and the fourth loss related to resource usage costs in the total loss calculation, and combining these with preset constraints, the invention reduces revenue losses caused by maintenance, minimizes scheduling conflicts, and improves the executability of the target time window.

[0005] This invention provides a maintenance scheduling method based on an operations research optimization model, applied to new energy power plants, comprising the following steps: obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task; wherein, the candidate time windows are selected from preset time windows based on a maintenance suitability index; using the operations research optimization model, with the goal of minimizing the total loss corresponding to the maintenance task, determining the target time window corresponding to each maintenance task from the multiple candidate time windows; wherein, the operations research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delay, and a fourth loss related to resource usage costs; generating maintenance scheduling results based on the target time window corresponding to each maintenance task.

[0006] Optionally, the first loss is determined based on the predicted electricity price, predicted electricity volume, maintenance impact coefficient, and coverage variable corresponding to all the candidate time windows; wherein, the coverage variable indicates whether the maintenance task is in the execution state within the candidate time window; And / or, The second loss is determined based on the under-sending deviation corresponding to all of the candidate time windows; And / or, The third loss is determined based on the time difference between all the candidate time windows and the earliest start time of each of the maintenance tasks; And / or, The fourth loss is determined based on the total resources occupied by the tasks currently being executed corresponding to all the candidate time windows.

[0007] Optionally, obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task includes: obtaining multiple preset time windows; for each maintenance task, filtering the multiple time windows according to the maintenance suitability index to obtain a candidate time window corresponding to the maintenance task.

[0008] Optionally, the step of filtering multiple time windows based on the maintenance suitability index to obtain candidate time windows corresponding to the maintenance task includes: obtaining the average predicted electricity price, average predicted electricity volume, contractual risk, and feasibility coefficient for each time window; wherein, the contractual risk is used to characterize the contract performance risk caused by under-distribution deviation under the time window; calculating the maintenance suitability score corresponding to each time window based on the average predicted electricity price, the average predicted electricity volume, the contractual risk, and the feasibility coefficient; and filtering candidate time windows from multiple time windows based on the maintenance suitability scores corresponding to each time window.

[0009] Optionally, the step of selecting candidate time windows from multiple time windows based on the maintenance suitability scores corresponding to each of the time windows includes: The time window in which the maintenance suitability score is greater than the preset score is used as the candidate time window; or, The time windows are sorted according to the maintenance suitability score; based on the sorting results, a preset number of the time windows with the highest sorting scores are selected as candidate time windows.

[0010] Optionally, the constraints include at least one of the following: unique task scheduling constraint, resource capacity constraint, maximum concurrent device downtime constraint, mutual exclusion constraint, no-stop time period constraint, and logical relationship and dependency constraint.

[0011] Optionally, the operations research optimization model is a mixed-integer linear programming model.

[0012] Optionally, according to a preset prediction period, the step of obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task is performed to generate maintenance scheduling results corresponding to each prediction period.

[0013] Optionally, obtaining multiple maintenance tasks includes: for the current prediction period: obtaining historical maintenance scheduling results generated in historical prediction periods prior to the current prediction period; and using historical maintenance tasks that have not been executed in the historical maintenance scheduling results and whose target time window is more than a preset time away from the current time as maintenance tasks for the current prediction period.

[0014] This invention also provides a maintenance scheduling system based on an operations research optimization model, applicable to new energy power plants, comprising the following modules: The data acquisition module is used to acquire multiple maintenance tasks and candidate time windows corresponding to each maintenance task; wherein, the candidate time windows are obtained by filtering from preset time windows based on the maintenance suitability index; The determination module is used to use an operations research optimization model to determine the target time window for executing each of the maintenance tasks from multiple candidate time windows, with the goal of minimizing the total loss corresponding to the execution of the maintenance task. The operation research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delays, and a fourth loss related to resource usage costs. The result output module is used to generate maintenance scheduling results based on the target time window corresponding to each of the maintenance tasks.

[0015] The present invention also provides an electronic device, 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 maintenance scheduling method based on the operations research optimization model as described above.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the maintenance scheduling method based on the operations research optimization model as described above.

[0017] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the maintenance scheduling method based on the operations research optimization model as described above.

[0018] This invention provides a maintenance scheduling method and system based on an operations research optimization model to address the problems in existing technologies, such as the lack of joint scheduling optimization for multiple maintenance tasks and the inability to obtain executable and implementable maintenance scheduling results. Specifically, by considering the first loss related to power generation revenue, the second loss related to contractual obligations, the third loss related to maintenance delays, and the fourth loss related to resource usage costs in the total loss, and solving the problem in conjunction with preset constraints, the invention reduces revenue losses caused by maintenance, minimizes scheduling conflicts, and improves the executability of the target time window. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the maintenance scheduling method based on the operations research optimization model provided by the present invention.

[0021] Figure 2 This is a flowchart illustrating the process of determining candidate time windows for each maintenance task provided by the present invention.

[0022] Figure 3 This is a schematic diagram of the scheduling results of a specific embodiment provided by the present invention.

[0023] Figure 4 This is a schematic diagram of the maintenance scheduling device based on the operations research optimization model provided by the present invention.

[0024] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0026] Figure 1 This is one of the flowcharts illustrating the maintenance scheduling method based on an operations research optimization model provided by this invention, such as... Figure 1 As shown, the method includes the following: Step 101: Obtain multiple maintenance tasks and corresponding candidate time windows for each maintenance task; the candidate time windows are selected from preset time windows based on the maintenance suitability index.

[0027] In the maintenance scenarios of new energy power stations, maintenance types are generally divided into four categories: daily inspection, periodic maintenance, emergency repair, and preventive testing. Periodic maintenance refers to the regular maintenance of specific hardware equipment, which requires interruption of normal operation. Therefore, reasonable maintenance time windows need to be pre-determined for these hardware maintenance tasks to minimize the impact on the daily operation of the new energy power station during the equipment downtime. Specifically, there are usually multiple maintenance tasks depending on the equipment and the maintenance project, each with a specific maintenance duration. The present invention provides a method for determining target time windows for each maintenance task while meeting preset constraints. It should be noted that the maintenance scheduling method based on an operations research optimization model provided in this invention can simultaneously determine target time windows for multiple maintenance tasks and fully consider the relationships between them during the solution process, providing a holistic scheduling result.

[0028] For each maintenance task, the requirements for candidate time windows vary depending on the maintenance requirements and the hardware equipment to be maintained. Therefore, in an optional embodiment, step 101 may specifically include: obtaining multiple preset time windows; for each maintenance task, filtering the multiple time windows according to the maintenance suitability index to obtain a candidate time window corresponding to the maintenance task.

[0029] For scheduled maintenance, the typical planning period H = 168 hours, i.e., the next 7 days, with each hour serving as a time window. Therefore, in this embodiment of the invention, 168 time windows are preset. For each maintenance task, time windows need to be filtered based on the maintenance suitability index to obtain candidate time windows for each task. In one optional embodiment, such as... Figure 2 As shown, the process of determining candidate time windows for each maintenance task includes: Step 201: Obtain the average predicted electricity price, average predicted electricity volume, contractual risk, and feasibility coefficient for each time window; Among them, contractual risk is used to characterize the risk of under-generation deviation or contract performance risk caused by the decrease in available electricity due to maintenance during the corresponding time window.

[0030] Step 202: Calculate the maintenance suitability score for each time window based on the average predicted electricity price, average predicted electricity volume, contractual risk, and feasibility coefficient. Step 203: Based on the maintenance suitability score corresponding to each time window, select candidate time windows from multiple time windows.

[0031] The maintenance suitability score is used to characterize the overall suitability of the maintenance task within the corresponding time window. The higher the score, the higher the overall suitability of the task within the corresponding time window, and vice versa.

[0032] For example, for each maintenance task Maintenance suitability score at each time window t The following formula (1) is used to calculate: Equation (1) in, Indicates the feasibility coefficient. This is used to filter out infeasible windows. This indicates normalization processing. This indicates the maintenance task. During this time window The average predicted electricity price is as follows. This represents the weighting factor for the average predicted electricity price. This indicates the maintenance task. During this time window The average predicted power consumption is as follows. This represents the weighting coefficient for the average predicted electricity consumption. This indicates the maintenance task. During this time window Contractual risks under the following terms This represents the weighting coefficient for contractually agreed risks. Contractually agreed risks can be calculated based on underpayment deviations, deviation penalty unit prices, and / or contracted electricity volume deviations within the corresponding time window.

[0033] In a further optional embodiment, the process of filtering time windows based on maintenance suitability scores in step 203 can include various implementation methods. For example, time windows with maintenance suitability scores greater than a preset score can be used as candidate time windows. By setting a preset score, candidate time windows can be quickly selected from multiple time windows. Alternatively, multiple time windows can be sorted according to maintenance suitability scores, and a preset number of time windows at the top of the sorting can be selected as candidate time windows. Specifically, the top N time windows in the sorting result can be used as candidate time windows, and the loss decomposition information and / or candidate sorting information corresponding to the candidate time windows can be output simultaneously. This invention does not specifically limit this approach.

[0034] Through steps 201 to 203 above, a pre-screening of time windows can be performed for each maintenance task. First, a screening is performed on a large number of time windows based on whether they are suitable for performing the maintenance task, and candidate time windows corresponding to each maintenance task are obtained. This allows the target time window to be calculated only based on the candidate time windows, which can significantly reduce the computational scale and improve computational efficiency.

[0035] Step 102: Using an operations research optimization model, with the goal of minimizing the total loss corresponding to the maintenance task, determine the target time window corresponding to each maintenance task from multiple candidate time windows; wherein, the operations research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delay, and a fourth loss related to resource usage costs.

[0036] In one optional embodiment, the operations research optimization model is a mixed integer linear programming (MILP) model. The objective function in MILP is to minimize the total loss. To solve the objective function, variables need to be constructed based on each task to be inspected, ensuring that each task is indivisible, starts only once, and is executed continuously.

[0037] For example, the following decision variables are introduced: The starting variable x(j,t) determines from which time window to start the maintenance task j; where x(j,t)∈ {0,1}, x(j,t)=1 indicates that the maintenance task j starts execution in the candidate time window t, and x(j,t)=0 indicates that the maintenance task j does not start in the candidate time window t.

[0038] Covering variable y(j,t): indicates that the maintenance task j is being executed in the candidate time window t; where y(j,t)∈ {0,1}, y(j,t)=1 indicates that the maintenance task j is being executed in the candidate time window t (within the continuous interval), and y(j,t)=0 indicates that the maintenance task j is not being executed in the candidate time window t.

[0039] The relationship between the coverage variable and the start variable can be expressed as: y(j,t) = Σ x(j,τ); τ = max(ES(j), td(j)+1), …, min(t, LS(j)). This relationship between the coverage variable and the start variable restricts each maintenance task to only one start within the planning period (168h).

[0040] The objective function is shown in equation (2) below: Equation (2) in, This represents the first loss related to power generation revenue. By setting this loss, maintenance tasks can be performed to avoid high-price periods, thereby maximizing the revenue of the new energy power plant. Indicates the weight of the first loss; This indicates a second loss related to the contractual agreement. By setting this loss, deviations in performance evaluation caused by underpayment can be avoided, thus minimizing the impact on performance evaluation. Indicates the weight of the second loss; This represents the third loss associated with maintenance delays. By setting this loss, we can prevent maintenance tasks from being delayed indefinitely and ensure that they are completed on schedule. Indicates the weight of the third loss; This represents the fourth loss related to resource usage costs. By setting this loss, resource costs can be reduced or resource congestion can be avoided, ensuring the scheduling feasibility of maintenance tasks. This indicates the weight of the fourth loss.

[0041] The first loss, second loss, third loss, and fourth loss set in the embodiments of the present invention will be described in detail below.

[0042] In one optional embodiment, the first loss is determined based on the predicted electricity price, predicted electricity volume, maintenance impact coefficient, and coverage variable corresponding to all candidate time windows; wherein, the coverage variable indicates whether the maintenance task is in the execution state within the candidate time window.

[0043] For example, the first loss It can be calculated using the following formula (3): Equation (3) in, Candidate time window The predicted electricity price is expressed in yuan / MWh. For equipment In the candidate time window The predicted electricity consumption is in kWh. Let be the downtime impact coefficient of the maintenance task j on the equipment, where , This indicates a complete shutdown for maintenance; during the maintenance period, the equipment will lose all its power. This indicates an incomplete shutdown for maintenance; the equipment's power is partially available during the maintenance period. For example, this could be a reduced-load maintenance operation. Indicates a covered variable.

[0044] In one alternative embodiment, the second loss is determined based on the under-sending deviations corresponding to all candidate time windows.

[0045] For example, the second loss It can be calculated using the following formula (4): Equation (4) in, It is the deviation penalty unit price (yuan / MWh), which can be a fixed value (such as deviation assessment based on a fixed rate) or vary with time period (such as some mechanisms related to electricity price / peak and valley). This indicates underpayment deviation; specifically, underpayment deviation... It can be calculated using the following equations (5) to (7): Δ(t) = max( 0 , E con (t) - E avail (t) ) Equation (5) Equation (6) Equation (7) Wherein, E in equation (5) con (t) represents the contracted electricity volume. This indicates the available power after the maintenance is affected, specifically... The total power generation of the station is calculated using equation (6), and the predicted power generation of the station in equation (6) is... Then it can be calculated by equation (7).

[0046] In one alternative embodiment, the third loss is determined based on the time difference between all candidate time windows and the earliest start time of each maintenance task.

[0047] For example, the third loss It can be calculated using the following formula (8): Equation (8) in, The value can be selected based on the urgency of the task to be repaired; specifically, the more urgent the task, the larger the value. Indicates the earliest start time of the maintenance task. Indicates the starting variable.

[0048] In one alternative embodiment, the fourth loss is determined based on the total resources used by the tasks being executed across all candidate time windows.

[0049] For example, the fourth loss It can be calculated using the following formula (9): Equation (9) in, This indicates the total resources used by the tasks currently executing within the candidate time window. This represents the unit cost of resource usage.

[0050] By using the total loss, which includes the first, second, third, and fourth losses mentioned above, as the objective function and solving it in conjunction with preset constraints, we can reduce the revenue loss caused by maintenance, reduce scheduling conflicts, and improve the feasibility of the target time window.

[0051] In a further optional embodiment, constraints need to be set to further limit the process of solving the objective function. Specifically, the constraints include at least one of the following: unique task scheduling constraint, resource capacity constraint, maximum concurrent device downtime constraint, mutual exclusion constraint, no-stopping period constraint, and logical relationship and dependency constraint.

[0052] The following provides a detailed explanation of the aforementioned constraints: (i) Task unique constraint arrangement means that each maintenance task must be arranged and can only be arranged once within each scheduled maintenance cycle. This can effectively avoid the problems of missed scheduling and duplicate scheduling. For example, it can be expressed by the following formula (10): Equation (10) (ii) Resource capacity constraint refers to the fact that the resource capacity of each new energy power station is often limited. Whether it is personnel, hoisting equipment, transport vehicles, or equipment parts, the resource occupation in any candidate time window cannot exceed the upper limit of the station's resource capacity. Therefore, it is necessary to constrain the resource capacity. For example, it can be expressed by the following formula (11): Equation (11) in, Indicates tasks to be inspected. Resource consumption; Indicates the candidate time window Each type of resource The available capacity of resources, where each type of resource It belongs to the resource type set R and can be at least one of personnel teams, hoisting equipment, transport vehicles, and equipment parts.

[0053] (iii) The upper limit constraint on concurrent shutdown of equipment refers to the upper limit of the number or capacity of equipment that can be shut down at the same time within the same candidate time window, thereby ensuring the stability of power generation or grid connection.

[0054] For example, the number of devices that can be shut down simultaneously within the same candidate time window is limited as shown in equations (12) and (13) below: Equation (12) Equation (13) in, Tasks awaiting maintenance Number of devices affected This is the maximum number of devices that can be shut down simultaneously within the same candidate time window.

[0055] For example, the capacity that can be shut down simultaneously within the same candidate time window is limited as shown in equations (14) and (15): Equation (14) Equation (15) in, Indicates tasks to be inspected. The capacity affected; Indicates candidate time window The predicted power.

[0056] (iv) Mutual exclusion constraint refers to the fact that certain maintenance tasks cannot be performed simultaneously due to safety or process factors. For example, two sections of the same collector line cannot be stopped at the same time, and the tasks served by the same crane cannot overlap.

[0057] For example, a mutual exclusion constraint can be represented by the following equation (16): Equation (16) Where J(f) represents the set of maintenance tasks in the mutual exclusion group f.

[0058] (v) No-stopping period constraints refer to the fact that maintenance tasks cannot be performed during certain time windows when considering scheduling instructions or strategies.

[0059] For example, the no-parking period constraint can be represented by the following equation (17): y(j,t) = 0, j, t∈T ban Equation (17) Among them, T ban This represents a pre-defined set of no-parking periods.

[0060] (vi) Logical relationships and dependencies refer to the logical or sequential dependencies between certain tasks to be inspected. For example, when performing task j to be inspected... B The maintenance task must be executed first. A For example, logical relationships and dependencies can be represented by equations (18) and (19) as follows: Equation (18) start(j B ) ≥ start(j A ) + d(j AEquation (19) Where d(j) A () indicates the task to be repaired. A The construction period.

[0061] In summary, the embodiments of the present invention constrain the work from multiple perspectives, including the task to be repaired, resource capacity, and time window limitations, by using the above-mentioned multiple constraints. This approach integrates various factors in practical applications and effectively improves the executability and practicality of the target time window obtained from the solution.

[0062] Step 103: Generate maintenance scheduling results based on the target time window corresponding to each maintenance task.

[0063] Specifically, a maintenance schedule can be generated based on the target time window corresponding to each maintenance task, and the execution time arrangement of each maintenance task within a forecast period can be displayed intuitively in a table format to guide the implementation of the maintenance tasks.

[0064] In a further optional embodiment, the maintenance scheduling method based on an operations research optimization model provided by this invention further includes: according to a preset prediction period, executing step 101 to obtain multiple maintenance tasks and candidate time windows corresponding to each maintenance task, so as to generate maintenance scheduling results corresponding to each prediction period. It is understood that the accuracy of predictions for data such as electricity prices and electricity consumption generally decreases the further away from the current time. Therefore, in this invention, a preset prediction period can be set to perform periodic execution with a shorter prediction period, thereby ensuring that maintenance scheduling can be performed with relatively accurate predicted electricity prices and electricity consumption within each prediction period.

[0065] In a further optional embodiment, the process of obtaining multiple maintenance tasks in step 101 further includes: for the current prediction period: obtaining the historical maintenance scheduling results generated in previous prediction periods; and using historical maintenance tasks that have not been executed in the historical maintenance scheduling results and whose target time window is more than a preset duration as the maintenance tasks for the current prediction period. That is, the prediction period is usually shorter than the planning period (168h), meaning that maintenance tasks within a planning period need to be scheduled before they are fully executed. Therefore, within each current prediction period, it is necessary to first obtain the historical maintenance scheduling results generated in previous prediction periods and find the historical maintenance tasks that have not been executed in the historical maintenance scheduling results and whose target time window is more than a preset duration. In other words, historical maintenance tasks that have been executed or are being executed within the historical prediction period do not need to be scheduled; only historical maintenance tasks that have not yet been executed need to be scheduled in a new round. Specifically, the preset duration can be 2h, 3h, 4h, 5h, 6h, etc. As can be seen, when generating maintenance schedule results periodically, the embodiments of the present invention can avoid frequently changing the target time window of maintenance tasks that have already started or are about to be executed, thus taking into account both the optimality and executability of the generated maintenance schedule results.

[0066] The following specific example illustrates the maintenance scheduling method based on the operations research optimization model provided in this embodiment of the invention: The tasks to be inspected are shown in Table 1 below: Table 1 List of tasks to be repaired

[0067] The basic information of the scene is as follows: (1) Site: A wind farm with 10 wind turbines (T01~T10) and a rated capacity of 2 MW per unit.

[0068] (2) Planning period: the next 72 hours, discrete time granularity Δ=1 hour, time index t=0,1,…,71.

[0069] (3) Predicted electricity price p(t) (yuan / MWh): Repeated in 24-hour daily pattern: 00-05:280, 06-11:380, 12-17:520, 18-21:760, 22-23:420.

[0070] (4) Predicted power per unit P(t) (MW / unit): 00-05: 1.2, 06-11: 1.0, 12-17: 0.8, 18-21: 0.9, 22-23: 1.1.

[0071] (5) Calculation of predicted power consumption:

[0072] Where E(t) is the predicted electricity consumption (MWh) for time period t, P(t) is the predicted power per unit (MW) for time period t, and Δ is the time granularity (h). In this embodiment, Δ = 1h, then E(t) = P(t).

[0073] (6) Resource capacity: Personnel team Cap_man(t)=2; hoisting Cap_crane(t)=1; spare parts Cap_spare(t)=1 (all calculated in hours).

[0074] (7) Mutual exclusion: Feeder F1={T01,T02,T03}; Feeder F2={T04,T05}; Each mutual exclusion group can stop at most 1 unit at any time.

[0075] Based on the above list of tasks to be repaired and their basic information, a target maintenance schedule is created for the five tasks to be repaired, and the output is as follows: Figure 3 The results are shown. Figure 3 The results not only show the final selected target time window but also the parallel execution relationship between multiple maintenance tasks. This demonstrates that the joint scheduling results of this invention do not merely output a few selectable time windows, but rather can output the correspondence and timing relationship between each maintenance task and the target time window. The scheduling results for the five maintenance tasks are as follows: Task J1 to be inspected (T01, 6h, hoisting + spare parts): t=0~5 (Day 1 00:00-06:00); Maintenance task J2 (T02, 8h, hoisting + spare parts): t=24~31 (Day 2 00:00-08:00); Maintenance task J3 (T03, 4h): t=48~51 (Day 3 00:00-04:00); Maintenance task J4 (T04, 4h): t=24~27 (Day 2 00:00-04:00) (parallel to J2, personnel=2, hoisting=1, spare parts=1, all not exceeding). Maintenance task J5 (T05, 6h, spare parts): t=48~53 (Day 3 00:00-06:00) (runs in parallel with J3, personnel=2, spare parts=1, not exceeding).

[0076] In summary, the maintenance scheduling method based on the operations research optimization model provided in this embodiment of the invention reduces revenue losses caused by maintenance and decreases scheduling conflicts by considering the first loss related to power generation revenue, the second loss related to contractual agreements, the third loss related to maintenance delays, and the fourth loss related to resource usage costs in the total loss, and solves the problem in combination with preset constraints, thereby improving the feasibility of the target time window.

[0077] The maintenance scheduling device based on the operations research optimization model provided by the present invention will be described below. The maintenance scheduling device based on the operations research optimization model described below and the maintenance scheduling method based on the operations research optimization model described above can be referred to in correspondence.

[0078] like Figure 4 As shown, the maintenance scheduling system 400 based on the operations research optimization model provided by this invention is applied to new energy power plants, including: The data acquisition module 401 is used to acquire multiple maintenance tasks and candidate time windows corresponding to each maintenance task; wherein, the candidate time windows are obtained by filtering from preset time windows based on the maintenance suitability index; The determination module 402 is used to use an operations research optimization model to determine the target time window for executing each of the maintenance tasks from multiple candidate time windows, with the goal of minimizing the total loss corresponding to the execution of the maintenance task. The operation research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delays, and a fourth loss related to resource usage costs. The result output module 403 is used to generate maintenance scheduling results based on the target time window corresponding to each of the maintenance tasks.

[0079] In an optional embodiment of the present invention, the first loss is determined based on the predicted electricity price, predicted electricity volume, maintenance impact coefficient, and coverage variable corresponding to all the candidate time windows; wherein the coverage variable indicates whether the maintenance task is in execution state within the candidate time window; and / or, the second loss is determined based on the under-generation deviation corresponding to all the candidate time windows; and / or, the third loss is determined based on the time difference between all the candidate time windows and the earliest start time of each maintenance task; and / or, the fourth loss is determined based on the total resources occupied by the currently executing tasks corresponding to all the candidate time windows.

[0080] In an optional embodiment of the present invention, the data acquisition module 401 is further configured to acquire a plurality of preset time windows; and for each of the maintenance tasks, filter the plurality of time windows according to the maintenance suitability index to obtain a candidate time window corresponding to the maintenance task.

[0081] In an optional embodiment of the present invention, the data acquisition module 401 is further configured to acquire, respectively, the average predicted electricity price, average predicted electricity volume, contractual risk, and feasibility coefficient corresponding to each time window; wherein, the contractual risk is used to characterize the contract performance risk caused by under-distribution deviation under the time window; calculate the maintenance suitability score corresponding to each time window based on the average predicted electricity price, the average predicted electricity volume, the contractual risk, and the feasibility coefficient; and select candidate time windows from multiple time windows based on the maintenance suitability scores corresponding to each time window.

[0082] In an optional embodiment of the present invention, the data acquisition module 401 is further configured to: use the time window in which the maintenance suitability score is greater than a preset score as a candidate time window; or, sort the multiple time windows according to the maintenance suitability score; and select a preset number of the time windows that are ranked higher as candidate time windows according to the sorting result.

[0083] In an optional embodiment of the present invention, the constraints include at least one of the following: unique task scheduling constraint, resource capacity constraint, maximum concurrent device downtime constraint, mutual exclusion constraint, no-stopping period constraint, and logical relationship and dependency constraint.

[0084] In an optional embodiment of the present invention, the operations research optimization model is a mixed integer linear programming model.

[0085] In an optional embodiment of the present invention, the maintenance scheduling system 400 based on the operations research optimization model provided by the present invention further includes: a rolling update module, used to execute the step of obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task according to a preset prediction period, so as to generate maintenance scheduling results corresponding to each prediction period.

[0086] In an optional embodiment of the present invention, the data acquisition module 401 is further configured to, for the current prediction period: acquire the historical maintenance scheduling results generated in the historical prediction periods prior to the current prediction period; and use the historical maintenance tasks that have not been executed in the historical maintenance scheduling results and whose target time window is more than a preset time away from the current time as the maintenance tasks to be performed in the current prediction period.

[0087] In summary, the maintenance scheduling system based on an operations research optimization model provided by this invention solves the problems in existing technologies, such as the lack of joint scheduling optimization for multiple maintenance tasks and the inability to obtain executable maintenance scheduling results. Specifically, by considering the first loss related to power generation revenue, the second loss related to contractual agreements, the third loss related to maintenance delays, and the fourth loss related to resource usage costs in the total loss, the system avoids scheduling conflicts while minimizing revenue losses caused by maintenance when determining the target time window, thus improving the feasibility of the target maintenance window. Furthermore, it ensures that the target time window is more closely aligned with the actual situation, thereby improving the executability of the maintenance tasks.

[0088] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 550, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 550. The processor 510 can call logical instructions in the memory 530 to execute a maintenance scheduling method based on an operations research optimization model. The method includes: obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task; using the operations research optimization model, with the goal of minimizing the total loss corresponding to the maintenance tasks, determining a target time window corresponding to each maintenance task from the multiple candidate time windows; wherein the operations research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delay, and a fourth loss related to resource usage costs; and generating a maintenance scheduling result based on the target time window corresponding to each maintenance task.

[0089] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0090] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the maintenance scheduling method based on the operations research optimization model provided by the above methods. The method includes: obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task; wherein the candidate time windows are obtained by filtering from preset time windows based on maintenance suitability index; using the operations research optimization model, with the goal of minimizing the total loss corresponding to the maintenance task, determining the target time window corresponding to each maintenance task from the multiple candidate time windows; wherein the operations research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delay, and a fourth loss related to resource usage costs; and generating maintenance scheduling results based on the target time window corresponding to each maintenance task.

[0091] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a maintenance scheduling method based on an operations research optimization model provided by the above methods. This method includes: acquiring multiple maintenance tasks and candidate time windows corresponding to each maintenance task; wherein the candidate time windows are selected from preset time windows based on a maintenance suitability index; using an operations research optimization model, with the goal of minimizing the total loss corresponding to the maintenance task, determining a target time window corresponding to each maintenance task from the multiple candidate time windows; wherein the operations research optimization model is configured with preset constraints, and the total loss includes at least one of the following: a first loss related to power generation revenue, a second loss related to contractual agreements, a third loss related to maintenance delays, and a fourth loss related to resource usage costs; and generating a maintenance scheduling result based on the target time window corresponding to each maintenance task.

[0092] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0093] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A maintenance scheduling method based on an operations research optimization model, characterized in that, Applications in new energy power plants include: Multiple maintenance tasks and candidate time windows corresponding to each maintenance task are obtained; wherein, the candidate time windows are obtained by filtering from preset time windows based on the maintenance suitability index; Using an operations research optimization model, with the goal of minimizing the total loss corresponding to the task to be repaired, a target time window for each task to be repaired is determined from multiple candidate time windows. The operations research optimization model is configured with preset constraints. The total loss includes: a first loss related to power generation revenue, a second loss related to contractual obligations, a third loss related to maintenance delays, and a fourth loss related to resource usage costs. The first loss is determined based on the predicted electricity price, predicted power generation, maintenance impact coefficient, and coverage variable corresponding to all candidate time windows. The coverage variable indicates whether the maintenance task is in execution within the candidate time window. The second loss is determined based on the under-generation deviation corresponding to all candidate time windows. The third loss is determined based on the time difference between all candidate time windows and the earliest start time of each maintenance task. The fourth loss is determined based on the total resources occupied by the currently executing tasks corresponding to all candidate time windows. A maintenance schedule is generated based on the target time window corresponding to each of the maintenance tasks.

2. The maintenance scheduling method based on the operations research optimization model according to claim 1, characterized in that, The step of obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task includes: Get multiple preset time windows; For each maintenance task, multiple time windows are filtered according to the maintenance suitability index to obtain a candidate time window corresponding to the maintenance task.

3. The maintenance scheduling method based on the operations research optimization model according to claim 2, characterized in that, The step of filtering multiple time windows based on the maintenance suitability index to obtain candidate time windows corresponding to the maintenance task includes: The average predicted electricity price, average predicted electricity volume, contractual risk, and feasibility coefficient are obtained for each time window; wherein, the contractual risk is used to characterize the contract performance risk caused by the under-generation deviation under the time window; Based on the average predicted electricity price, the average predicted electricity volume, the contractual risk, and the feasibility coefficient, the maintenance suitability score corresponding to each time window is calculated. Candidate time windows are obtained by filtering from multiple time windows based on the maintenance suitability scores corresponding to each time window.

4. The maintenance scheduling method based on the operations research optimization model according to claim 3, characterized in that, The step of selecting candidate time windows from multiple time windows based on the maintenance suitability scores corresponding to each time window includes: The time window in which the maintenance suitability score is greater than the preset score is used as the candidate time window; or, The time windows are sorted according to the maintenance suitability score. Based on the sorting results, a preset number of time windows that rank highly are selected as candidate time windows.

5. The maintenance scheduling method based on the operations research optimization model according to claim 1, characterized in that, The constraints include at least one of the following: unique task scheduling constraint, resource capacity constraint, maximum concurrent device downtime constraint, mutual exclusion constraint, no-stopping time period constraint, and logical relationship and dependency constraint. And / or, The operations research optimization model is a mixed-integer linear programming model.

6. The maintenance scheduling method based on the operations research optimization model according to claim 1, characterized in that, According to the preset prediction cycle, the steps of obtaining multiple maintenance tasks and candidate time windows corresponding to each maintenance task are executed to generate maintenance scheduling results corresponding to each prediction cycle.

7. The maintenance scheduling method based on the operations research optimization model according to claim 6, characterized in that, The acquisition of multiple maintenance tasks includes: For the current forecast period: Obtain the historical maintenance scheduling results generated in the historical forecast periods prior to the current forecast period; Historical maintenance tasks that have not been executed in the historical maintenance scheduling results and whose target time window is more than a preset time away from the current time are taken as maintenance tasks for the current prediction cycle.

8. A maintenance scheduling system based on an operations research optimization model, characterized in that, Applications in new energy power plants include: The data acquisition module is used to acquire multiple maintenance tasks and candidate time windows corresponding to each maintenance task; wherein, the candidate time windows are obtained by filtering from preset time windows based on the maintenance suitability index; The determination module is used to use an operations research optimization model to determine the target time window for executing each of the maintenance tasks from multiple candidate time windows, with the goal of minimizing the total loss corresponding to the execution of the maintenance task. The operations research optimization model is configured with preset constraints. The total loss includes: a first loss related to power generation revenue, a second loss related to contractual obligations, a third loss related to maintenance delays, and a fourth loss related to resource usage costs. The first loss is determined based on the predicted electricity price, predicted power generation, maintenance impact coefficient, and coverage variable corresponding to all candidate time windows. The coverage variable indicates whether the maintenance task is in execution within the candidate time window. The second loss is determined based on the under-generation deviation corresponding to all candidate time windows. The third loss is determined based on the time difference between all candidate time windows and the earliest start time of each maintenance task. The fourth loss is determined based on the total resources occupied by the currently executing tasks corresponding to all candidate time windows. The result output module is used to generate maintenance scheduling results based on the target time window corresponding to each of the maintenance tasks.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the maintenance scheduling method based on the operations research optimization model as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the maintenance scheduling method based on the operations research optimization model as described in any one of claims 1 to 7.

Images