Multi-cycle offshore wind power operation and maintenance optimization method considering combination of child ship and mother ship

Abstract

The invention belongs to the technical field of power systems, and particularly provides a multi-cycle offshore wind power operation and maintenance scheduling optimization method based on combination of a primary ship and a secondary ship, and the method comprises the steps: constructing a PSO-KMEANS fan clustering method, and dividing a to-be-maintained fan according to the cycle of a maintenance task; with the lowest comprehensive operation and maintenance cost as the target, an open-sea wind power operation and maintenance integer programming model based on combination of the child ship and the mother ship is constructed and put forward; the method comprises the following steps: based on a Dantzigzag-Wolfe decomposition principle, decomposing an open-sea wind power operation and maintenance integer programming model based on combination of a child ship and a mother ship into a path-based main problem and a pricing child problem to form a main problem model and a child problem model; and solving the main problem model and the sub-problem model based on a branch pricing method, and outputting an integer solution. According to the method, the offshore wind power operation and maintenance technology under the combined operation and maintenance mode of the child ship and the mother ship is optimized, the operation and maintenance cost of the offshore wind power plant is effectively reduced while efficient scheduling is achieved, and theoretical reference and decision support are provided for planning and scheduling personnel in the field of wind power operation and maintenance.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power systems, and in particular, relates to a multi-period offshore wind power operation and maintenance scheduling optimization method based on a combination of a mother ship and a daughter ship. BACKGROUND

[0002] During offshore wind power operation and maintenance, operation and maintenance ships frequently go back and forth to the base for supplies and personnel transfer, greatly increasing the navigation cost and reducing the operation and maintenance efficiency, and possibly affecting the timeliness and safety of the operation and maintenance task. Especially in the case where wind farms are widely distributed and far away from the shore base, the conventional operation and maintenance mode faces high economic cost and operation difficulty. Therefore, in order to improve the efficiency and economy of wind power operation and maintenance, the concept of mother-daughter ship operation and maintenance is gradually put forward and becomes one of the feasible solutions to solve this problem.

[0003] The offshore wind power operation and maintenance mother ship is a large ship that can accommodate personnel and store spare parts. It has a self-sustaining capacity of more than one month, excellent berthing capacity, and can carry operation and maintenance daughter ships to transfer technical maintenance personnel and equipment to the wind turbine to be maintained. Since the ship does not need to frequently go back and forth to the base for supplies, it is suitable for offshore wind power operation and maintenance. In September 2023, the first batch of operation and maintenance mother ships of Shanghai Electric Wind Power Group was launched, marking an important step in China's offshore wind power operation and maintenance field and filling the gap of domestic offshore wind power operation and maintenance special ships. However, although the use of operation and maintenance mother ships has been preliminarily applied in China, in the actual application of offshore wind power operation and maintenance by mother-daughter ships, the current technical solution still has many shortcomings: ① how to effectively coordinate the actions of the mother ship and the daughter ship, fully utilize their respective advantages, and maximize the overall operation and maintenance efficiency is still a problem to be solved; ② offshore wind power operation and maintenance tasks have obvious periodicity and uncertainty, such as the random occurrence of wind turbine failures and the limitation of weather windows. The existing operation and maintenance scheme may not fully consider these factors, leading to unreasonable resource scheduling, high operation and maintenance cost, and even affecting the timeliness and safety of the operation and maintenance task; ③ how to realize the optimal allocation and efficient scheduling of resources based on the consideration of the coordinated operation of the mother ship and the daughter ship, the periodicity and uncertainty of the operation and maintenance task is a difficult problem currently faced. The existing resource scheduling and operation and maintenance strategy may not be able to meet the complex and variable operation and maintenance needs of offshore wind farms, and a balance needs to be found between efficiency, cost, and safety. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a multi-period offshore wind power operation and maintenance scheduling optimization method based on a combination of a mother ship and a daughter ship, which optimizes the offshore wind power operation and maintenance technology under the joint operation and maintenance mode of the mother ship and the daughter ship, and effectively reduces the offshore wind farm operation and maintenance cost while realizing efficient scheduling.

[0005] To solve the above technical problems, the technical scheme adopted by the present application is: a multi-cycle offshore wind power operation and maintenance scheduling optimization method based on the combination of a mother ship and a child ship, comprising the following steps: Step one, build a PSO-KMEANS wind turbine clustering method, and divide the wind turbines to be maintained according to the maintenance task cycle; Step two, taking the lowest comprehensive operation and maintenance cost as the target, an offshore wind power operation and maintenance integer programming model based on the combination of a mother ship and a child ship is constructed; Step three, based on the Dantzig-Wolfe decomposition principle, the offshore wind power operation and maintenance integer programming model based on the combination of a mother ship and a child ship is decomposed into a path-based main problem and a pricing sub-problem, forming a main problem model and a sub-problem model; Step four, based on the branch and price method, the main problem model and the sub-problem model are solved, and an integer solution is output.

[0006] In the preferred scheme, in step one, the PSO-KMEANS wind turbine clustering method comprises the following steps: S1.1, initialize the particle swarm and the clustering center; S1.2, update the speed and position of the particles, and according to the new particle position, assign the wind turbines to the nearest clustering center, while ensuring that the total maintenance time of each cluster does not exceed the window period; S1.3, calculate the fitness function of the current particle, and the fitness function of the current particle is the sum of the distances between all wind turbines, and the goal is to minimize the fitness function value; S1.4, update the individual optimal solution and the global optimal solution: if the fitness of the current particle is better than its historical optimal fitness, update the individual optimal solution; if the fitness of the current particle is better than the global optimal fitness, update the global optimal solution; S1.5, move each particle according to the updated speed and position; S1.6, repeat steps S1.2-S1.5 until the maximum number of iterations is reached; S1.7, output the corresponding result according to the group historical optimal solution.

[0007] In the preferred scheme, in step two, the objective function of the offshore wind power operation and maintenance integer programming model based on the combination of a mother ship and a child ship is: (1); In the formula: C represents the comprehensive operation and maintenance cost; T represents the operation and maintenance cycle, ; S represents the set of stopover positions of the operation and maintenance mother ship, ; V represents the set of all nodes accessible by the operation and maintenance child ship, ; represents a set of each type of maintenance technician, represents a node to a node ; represents a distance from a node to a node ; represents the fuel cost of the mother ship per unit distance of sailing; represents the fuel cost of the daughter ship per unit distance of sailing; represents the cost required for maintaining a wind turbine ; represents the cost of the loss of power generation per hour due to the downtime of a wind turbine; represents the salary that a technician needs to be paid per hour; represents the number of technicians required for maintaining a wind turbine ; represents the time for the daughter ship to arrive at a node ; represents the time required for maintaining a wind turbine ; represents the time for transferring spare parts when a wind turbine is being maintained; represents that when the daughter ship is in a period from a node to a node , then , otherwise ; represents that when the mother ship is in a period from a node to a node , then , otherwise .

[0008] In a preferred solution, in step two, the constraint conditions of the offshore wind power operation integer programming model based on the combination of the mother ship and the daughter ship are: 1) The mother ship can only choose one stop point in each period: (2); 2) Each wind turbine can only be maintained once: (3); 3) The daughter ship departs from the mother ship, completes the operation task, and returns to the mother ship: (4); (5); ​4) The traffic balance constraint is: (6); (7); 5) The spare parts for wind turbine maintenance shall not exceed the maximum spare parts load of the daughter ship: (8); 6) The number of technicians carried by the daughter vessel does not exceed its maximum carrying capacity: (9); 7) The resources required for operation and maintenance tasks shall not exceed the total amount of spare parts of the mother ship: (10); (11); 8) Continuity constraints on operation and maintenance time: (12); (13); (14); 9) Continuity of maintenance equipment; (15); 10) Operation and maintenance time does not exceed the time window period: (16); 11) Integer constraints on decision variables: (17); Where: Represents the set of all wind turbine nodes to be repaired, ; Indicates maintenance fan Required spare parts the number of Indicates the technical personnel carried by the operation and maintenance mother ship the number of Indicates the maintenance spare parts carried by the operation and maintenance mother ship the number of Indicates that the wind farm is in period The time window period; Indicates that the operation and maintenance mother ship has arrived at the docking node time; Indicates the time for transferring personnel and spare parts between the operation and maintenance mother ship and the daughter ship; Indicates the navigation speed of the maintenance sub-vessel; Indicates the spare parts load of the operation and maintenance sub-vessel; Indicates the technician carrying capacity of the operation and maintenance sub-vessel; Represents the maintenance wind turbine The weight of spare parts needed; Represents the weight of spare parts when the maintenance sub-ship leaves the wind turbine .

[0009] In the preferred scheme, in step three, the main problem model divides the wind turbines to be maintained into different sets through the method of set covering; The objective function of the main problem model is to minimize the total maintenance cost; the constraint conditions are: the feasible path is consistent with the position of the mother ship, one sub-ship cannot perform multiple maintenance tasks at the same time, each wind turbine can only be maintained once, and the integer constraint of the decision variable.

[0010] In the preferred scheme, the objective function expression of the main problem model is: (18); The constraint conditions of the main problem model are: 1) The feasible path is consistent with the position of the mother ship: (19); 2) One sub-ship cannot perform multiple maintenance tasks at the same time: (20); 3) Each wind turbine can only be maintained once: (21); 4) Integer constraint of the decision variable (22); In the formula: represents the total maintenance cost; represents the set of all feasible paths of the maintenance sub-ship when the mother ship is parked at node in the cycle ; represents all the costs of the maintenance sub-ship in path ; represents that when path is selected, then , otherwise ; represents that when the maintenance sub-ship maintains the wind turbine in path , then , otherwise .

[0011] In the preferred scheme, in step three, the main problem model is relaxed, and the operation steps are: 1) Some feasible paths are included in the initialization step to form a restrictive main problem; 2) Relax the integer constraint of decision variable, the restrictive master problem is converted into a restrictive relaxed master problem, all feasible paths are changed into partial feasible paths integer variables are changed into continuous variables: (23); 3) Solve the restrictive relaxed master problem by the optimizer, and obtain the dual variables for the sub-problem model.

[0012] 8. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the combination of the mother ship and the child ship according to claim 7, wherein in the step three, the objective function of the pricing sub-problem is: (24); wherein: represents the set of wind turbine nodes that need to be maintained in the period ; represents the set of wind turbine nodes that need to be maintained in the period , wherein the starting node of the operation and maintenance child ship is consistent with the mother ship docking node; represents the set of wind turbine nodes that need to be maintained in the period , wherein the return node of the operation and maintenance child ship is consistent with the mother ship docking node; represents the set of wind turbine nodes that need to be maintained in the period ; wherein the return node of the operation and maintenance child ship is consistent with the mother ship docking node; represents all the costs of the operation and maintenance child ship from node to node in the path ; represents that if the operation and maintenance child ship is from node to node in the path , then , otherwise ; represents that if the wind turbine to be maintained is maintained by the child ship in the path , then , otherwise ; wherein and are the dual variables of constraints (20) and (21) respectively. are changed into partial feasible paths

[0013] In the preferred scheme, the constraint condition of the pricing sub-problem is: 1) The operation and maintenance child ship departs from the mother ship, completes the operation and maintenance task, and returns to the mother ship: ​​ (25); (26); 2) Each wind turbine is only maintained once: (27); 3) The time at which the service sub-ship leaves the mother ship in the path: (28); 4) The time at which the service sub-ship returns to the mother ship in the path: (29); 5) The path must satisfy the capacity constraints of the service sub-ship (30); (31); 6) The path must satisfy the capacity constraints of the mother ship: (32); (33); 7) Continuity of service time (34); (35); (36); 8) Continuity of maintenance equipment: (37); 9) The path must satisfy the time window period (38) 10) The value range of all decision variables (39); In the formula: denotes the time at which the service sub-ship leaves the mother ship stop point in the path ; denotes the time at which the service sub-ship returns to the mother ship stop point in the path ; denotes the time at which the service sub-ship arrives at the node in the path ; denotes the time at which the mother ship arrives at the stop node in the path ; denotes the sea time window period.

[0014] In the preferred scheme, the solving process of step four is: S4.1. Use the greedy algorithm to obtain the initial path of the main problem; S4.2. Initialize the search tree, add the root node, and select some nodes to construct the restricted main problem model; S4.3. Introduce relaxation constraints to relax the restricted master problem model into a restricted relaxed master problem model; S4.4. Solve the restricted relaxation master problem model to obtain the dual variables for the subproblem model; S4.5. Use dual variables to construct sub-problem models and find solutions that satisfy The corresponding solution, ; S4.6. During each iteration of the main problem and sub-problem, one or more columns that can minimize the sub-problem objective function are generated and incorporated into the restricted relaxation main problem model. New columns are continuously added until no more columns can be used. When the column appears, the optimal solution of the original problem can be obtained; S4.7. Use the branch and bound method to process the output non-integer solution, convert it into an integer solution and then output it.

[0015] The present invention provides a multi-period offshore wind power operation and maintenance scheduling optimization method based on the combination of mother and daughter ships, which has the following beneficial effects: 1. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the combination of mother and daughter ships proposed in this paper realizes efficient clustering of wind turbines by combining the PSO-KMEANS method, and comprehensively considers multi-dimensional characteristics such as wind turbine distance and weather window, thereby improving the rationality of task division and resource utilization.

[0016] 2. We constructed a multi-period integer programming model for offshore wind turbine operation and maintenance based on a coordinated operation model between mother and daughter vessels, and applied a branch-and-price algorithm framework to this complex operation and maintenance scenario for the first time. Using the Dantzig-Wolfe decomposition principle, we decomposed the original problem into a master problem and subproblems. We then applied a column generation algorithm and a label setting algorithm with an acceleration strategy to efficiently iterate between the two problems. Finally, we achieved an integer optimal solution using an arc-based branching strategy.

[0017] 3. This invention designs a collaborative operation and maintenance model for offshore wind farms. By using a mother vessel (SOV) to carry a daughter vessel (CTV) to perform maintenance tasks, this model leverages the characteristics of the mother vessel to improve the efficiency of deep-sea maintenance and overcome the high costs associated with distance limitations in traditional models. Case studies demonstrate that this model effectively reduces offshore wind farm operation and maintenance costs and enables efficient scheduling within a short time window.

[0018] 4. In terms of constraint processing, this invention solves the low efficiency of integer programming in traditional methods by relaxing the integer constraints of the main problem variables and introducing a branching strategy for solving the problem. It significantly improves the accuracy and feasibility of operation and maintenance scheduling optimization, and ultimately achieves efficient allocation of multi-cycle tasks. 5. In view of the shortcomings of the existing methods, the present invention first divides the tasks of the wind turbines to be repaired according to the maintenance cycle based on the PSO-KMEANS method by comprehensively considering multiple characteristic data such as wind turbine distance and offshore time window; then, a mother-and-daughter ship combined with multi-cycle offshore wind power operation and maintenance integer programming model is established with the minimum comprehensive operation and maintenance cost as the objective function; considering the complexity of the model, the problem is decomposed into main and sub-problems through the Dantzig-Wolfe decomposition method; and the relaxation constraint is introduced to relax the main problem; then the branch and price method is used to solve the decomposed model to avoid the relaxation model from being unable to output integer solutions; finally, in order to verify the feasibility and necessity of the proposed model and method, a series of verifications are carried out using a certain offshore wind farm as an example. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 It is the technical roadmap of the present invention; Figure 2 This is the flow chart of the PSO-KMEANS method; Figure 3 Flowchart of the branch pricing method; Figure 4 This is the flow chart of the K-means method; Figure 5 This is a diagram of the location and status of wind turbines in an offshore wind farm; Figure 6 Provide a maintenance area division diagram for each period of offshore wind farms; Figure 7 This is the maintenance route map for the first day of the operation and maintenance ship in Plan 1; Figure 8 This is the maintenance route map for the first day of the operation and maintenance ship in Option 2; Figure 9 This is the maintenance route map for the first day of the operation and maintenance ship in Plan 3; Figure 10 This is the maintenance route map for the operation and maintenance vessel on the second day of Plan 1; Figure 11 This is the maintenance route map for the operation and maintenance vessel on the second day of Plan 2; Figure 12 This is the maintenance route map for the operation and maintenance vessel on the second day of Plan 3; Figure 13 This is the maintenance route diagram for the operation and maintenance vessel on the third day of Plan 1; Figure 14A maintenance path diagram for the third day of the scheme 2 maintenance ship; Figure 15 A maintenance path diagram for the third day of the scheme 3 maintenance ship. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0021] Example 1: A multi-period offshore wind power maintenance scheduling optimization method based on the combination of a mother ship and a child ship, as shown in FIG. 1, includes the following steps: Figure 1 Step one, a wind turbine clustering method based on particle swarm optimization (PSO) algorithm and K-means clustering is constructed, and the wind turbines to be maintained are divided according to the maintenance task period. When performing offshore wind power maintenance, the maintenance task cannot be completed within the day and is divided into multiple periods for completion. The maintenance mother ship departs from the base and selectively stops at a certain wind turbine to be maintained in each period, and then lowers the maintenance child ship carried to perform the maintenance task in the period. Therefore, in order to improve the maintenance efficiency, it is necessary to reasonably allocate the maintenance task in each period.

[0022] Considering the offshore wind turbine maintenance problem combined with the child ship and the vehicle and drone joint distribution problem (Vehicle Routing Problem with Drone, VRPD), it is similar to the vehicle problem and belongs to the NP-hard problem, and the global optimal solution cannot be obtained. In the vehicle problem, the idea of "clustering first, path second" is proposed, the essence of which is to cluster nodes to different partitions by distance as a feature, and then realize the optimal path in the partition through the nearest neighbor algorithm. Inspired by the vehicle problem, the present application divides the wind turbines to be maintained according to the maintenance task period through the clustering method. On the one hand, since the wind turbine will incur a penalty cost due to the loss of power generation when it is not maintained, the clustering method can be used to allocate the wind turbines with close distances to the same maintenance period as much as possible under the premise of meeting the time window, thereby reducing the penalty cost, and the clustering can also reduce the complexity of the overall maintenance problem. On the other hand, since the total period time of the wind turbine maintenance task can be determined in advance, the cluster number of the clustering method can be directly determined by dividing the maintenance task according to the period time, so that a more accurate clustering result can be obtained.

[0023]

[0024] ​The core idea of ​​the traditional K-MEANS method is mainly the two stages of cluster initialization and iterative optimization. In the initialization stage, the method divides the data set into K groups, and randomly selects or selects K objects from each group according to specific criteria as the initial cluster centers, and then assigns the data points to the nearest cluster centers through continuous iteration. However, this method has certain limitations: (1) The method is very sensitive to the selection of initial values, and different initial points may lead to completely different clustering results. If the initial cluster center is not selected well, it may lead to incorrect clustering or reduce the performance of the K-MEANS method; (2) If the method repeatedly calculates the distance from each data point to the cluster center, it will greatly increase the running time of the method. The process of the traditional K-MEANS method is shown in the attached figure. Figure 4 As shown: (1) Randomly select K objects in the data set as the initial cluster centers ; (2) Each sample in the dataset is recorded as , calculate the distance between the sample and the randomly selected K cluster centers according to the Euclidean distance, and calculate the average error based on it, and then divide; (3) For each category of samples , recalculate the location of each cluster center ( represents the total number of samples, represents the centroid of the sample of this category) (4) Repeat steps 2 and 3 until the maximum number of iterations is reached, and compare whether the average errors of the two iterations are consistent. If so, it indicates that the position of the cluster center has stabilized.

[0025] The particle swarm optimization (PSO) algorithm is a swarm intelligence optimization algorithm. Its key feature is that each particle in the swarm represents a potential solution to a problem. Through the individual behaviors of particles and information exchange within the swarm, the algorithm intelligently approaches the optimal solution to a problem.

[0026] Assume that the particle swarm consists of particles, in The optimal position of searching particles in dimensional space can be expressed as As shown in formula (1): (1); For a For particles at a certain position, the fitness value of the position can be calculated by the objective function. The fitness value determines the quality of the optimization result, and the optimal position reached by the particle in each optimization is recorded. The optimal position is recorded as , as shown in formula (2): (2); When all particles in the population reach the optimal position, i.e. the optimal position of the population, the optimal position of the population is recorded as As shown in equation (3): (3); Let the velocity of the i-th particle be When the particle reaches the same position each time, the particle velocity and position are updated, as shown in equation (4).

[0027] (4); The updating method of the particle velocity and position is as shown in equation (5): (5); In the equation, is an inertia factor, and the greater the inertia factor, the farther the particle jumps, i.e. the wider the search range; , is the individual optimal position (local optimum) and the population optimal position (global optimum) of the particle, respectively; , is an acceleration constant; , represents a random number between 0 and 1, represents a constraint factor.

[0028] The particle swarm algorithm flow is as follows: (1) A number of particles are randomly generated, and the position vector and velocity vector of each particle are defined. The position vector represents the position of the particle in the current search space, and the velocity vector represents the current moving speed of the particle; (2) The fitness of the particle is calculated based on the objective function of the current problem; (3) The particle with the highest fitness is selected, and its position vector is set as the global optimal solution; (4) The velocity and position of the particle are updated according to the current velocity, position and global optimal solution of the particle.

[0029] (5) After updating the velocity and position of the particle, the fitness of the particle at the new position is calculated, and the local optimal solution and global optimal solution of the particle are updated accordingly.

[0030] (6) Repeat steps (4) and (5) until the iteration condition is reached.

[0031] ​The fan clustering method of the PSO-KMEANS method can initialize the clustering center by using the global optimization ability of the PSO algorithm, reduce the dependence on the initial clustering center, and improve the stability and robustness of the method. Meanwhile, the method can also consider multiple feature data such as fan distance and operation and maintenance time window period to divide the to-be-maintained fans, and arrange as many fans as possible to the same maintenance period, so as to reduce the penalty cost of the to-be-maintained fans due to loss of power generation.

[0032] The process of the PSO-KMEANS method is shown in FIG. 1. Figure 2 (1) initialize the particle swarm, clustering center and other parameters; (2) update the speed and position of the particles, and according to the new particle position (i.e. the new clustering center), distribute the fans to the nearest clustering center, while ensuring that the total maintenance time of each cluster does not exceed the window period; (3) calculate the fitness of the current particle (i.e. the total distance between all fans). The goal is to minimize this function value, i.e. to make the distance between the fans in the same cluster as close as possible to reduce the maintenance time and cost; (4) update the individual optimal solution and the global optimal solution: if the fitness of the current particle is better than its historical optimal fitness, update the individual optimal solution; if the fitness of the current particle is better than the global optimal fitness, update the global optimal solution; (5) move each particle according to the updated speed and position; (6) repeat steps 2-5 until the maximum number of iterations is reached; (7) output the corresponding result according to the group historical optimal solution.

[0033] Step two, taking the minimum comprehensive operation and maintenance cost as the goal, an integer programming model for offshore wind power operation and maintenance based on the combination of mother ships and daughter ships is constructed.

[0034] The sailing speed of the operation and maintenance mother ship in the offshore wind farm is very slow, especially in the dynamic positioning mode, at about 10-20 knots, so the mother ship needs a long time to sail between wind turbines. In order to complete the maintenance task of the wind turbine within a limited time window, the mother ship will stop at a to-be-maintained wind turbine in each cycle, and the daughter ship will continuously go back and forth between the mother ship to complete the maintenance task in the cycle. Therefore, in each maintenance cycle, we can regard the mother ship as an offshore operation and maintenance base, and optimize the stopping position of the mother ship in each cycle to ensure the economic optimization.

[0035] ​The multi-cycle operation and maintenance integer programming model for the daughter and mother ship established in this invention is a deterministic optimization model. Model parameters include various information about the wind turbine to be repaired and the required operation and maintenance resources, such as the operation and maintenance vessel, spare parts, and technicians. Information about the wind turbine to be repaired includes: the coordinates of the wind turbine to be repaired; and operation and maintenance requirements, including the turbine power loss cost, turbine maintenance cost, turbine maintenance time, transfer time for technicians and spare parts, type and quantity of spare parts, and type and number of technicians. Operation and maintenance vessel information includes the maximum spare parts load capacity, maximum number of passengers, navigation speed, and fuel cost per unit distance of navigation for the daughter and mother ship. Weather window information includes the time range for the daughter ship to depart and return to the mother ship, meaning that the operation and maintenance work time must not exceed the time window, and the daughter ship cannot stay at sea overnight and must return to the mother ship on the same day. The mother ship departs from the base and returns to the base after completing the maintenance task in the last cycle. Furthermore, this invention assumes that the mother ship has sufficient resources and does not participate in wind turbine maintenance tasks.

[0036] 1. Integer programming model objective function The multi-period operation and maintenance integer programming model considering the combination of daughter and mother ships takes the minimum comprehensive operation and maintenance cost as the objective function (including the daughter ship navigation cost, the mother ship navigation cost, the wind turbine maintenance cost, the power generation loss cost and the technician cost), which is expressed as follows: (6); Where: Indicates the operation and maintenance cycle, ; Represents the set of locations where the operation and maintenance mother ship can dock. ; Represents the set of all nodes accessible to the operation and maintenance sub-ship, ; Represents a collection of various types of maintenance technicians, ; Representation node To Node the distance between them; Represents a slave node To Node the distance between them; Indicates the fuel cost per unit distance traveled by the operation and maintenance mother ship; Indicates the fuel cost per unit distance of the operation and maintenance sub-vessel; Indicates maintenance fan Required expenses (spare parts purchase cost + maintenance cost); It represents the cost of power generation loss caused by wind turbine downtime per hour; Indicates technical personnel The hourly wage to be paid; Indicates maintenance fan Required technical personnel the number of Indicates that the maintenance sub-ship has arrived at the node time; Indicates maintenance fan the time required; Indicates maintenance fan Time to transfer spare parts; Indicates that when the maintenance sub-ship is in the cycle Slave Node To Node ,but ,otherwise ; Indicates that when the operation and maintenance mother ship is in the cycle Slave Node To Node ,but ,otherwise .

[0037] 2. Integer programming model constraints The constraints of the model are: (7); (8); (9); (10); (11); (12); (13); (14); (15); (16); (17); (18); (19); (20); (twenty one); (twenty two); Where: Represents the set of all wind turbine nodes to be repaired, ; Indicates maintenance fan Required spare parts the number of Indicates the technical personnel carried by the operation and maintenance mother ship the number of Indicates the maintenance spare parts carried by the operation and maintenance mother ship the number of Indicates that the wind farm is in period The time window period; Indicates that the operation and maintenance mother ship has arrived at the docking node time; Indicates the time for transferring personnel and spare parts between the operation and maintenance mother ship and the daughter ship; Indicates the navigation speed of the maintenance sub-vessel; Indicates the spare parts load of the operation and maintenance sub-vessel; Indicates the technician carrying capacity of the operation and maintenance sub-vessel; Indicates maintenance fan The weight of the spare parts required; Indicates that the maintenance vessel has left the wind turbine Spare parts weight at the time of use.

[0038] Among them, Equation (7) indicates that the mother ship can only choose one docking point in each cycle; Equation (8) indicates that each wind turbine can only be maintained once; Equations (9) and (10) indicate that the operation and maintenance daughter ship departs from the mother ship and returns to the mother ship after completing the operation and maintenance task; Equations (11) and (12) are flow balance constraints; Equation (13) indicates that the wind turbine maintenance spare parts do not exceed the maximum spare parts load of the daughter ship; Equation (14) indicates that the number of technicians carried by the daughter ship does not exceed its maximum carrying capacity; Equations (15) and (16) indicate that the resources required for the operation and maintenance task do not exceed the total number of spare parts of the mother ship; Equations (17), (18) and (19) indicate the continuity of the operation and maintenance time; Equation (20) indicates the continuity of the maintenance equipment; Equation (21) indicates that the operation and maintenance time does not exceed the time window period; Equation (22) indicates the integer constraint of the decision variable.

[0039] Step 3: Based on the Dantzig-Wolfe decomposition principle, the offshore wind power operation and maintenance integer programming model based on the mother-daughter ship combination is decomposed into a path-based main problem and a pricing sub-problem, forming a main problem model and a sub-problem model.

[0040] Since the example in step 2 is large in scale, a decomposition strategy of the main problem and subproblems based on Dantzig-Wolfe decomposition is proposed to simplify and decompose the problem.

[0041] The integer programming model established in step two can be solved directly by a commercial optimization solver, which can solve limited scale of examples, and the application simplifies the initial problem based on Dantzig-Wolfe decomposition principle, and decomposes the initial problem into a path-based main problem and a pricing sub-problem. The main problem after decomposition is solved by a GUROBI solver, and the sub-problem is solved by a label setting algorithm.

[0042] 1. Main problem model The to-be-repaired wind turbines are divided into different sets by the set covering method, and it is necessary to define the single trip condition: 1) To meet the sub-ship capacity constraint; 2) The time of the wind turbines repaired in the trip does not exceed the time window period of the current maintenance period; 3) The path starts from the mother ship and finally returns to the mother ship, and each path starts from the mother ship only once.

[0043] The single trip set covering model is represented as follows: (23); (24); (25); (26); (27); In the formula: represents the set of all feasible paths of the sub-ship when the operation and maintenance mother ship stops at the node ; represents all the costs of the operation and maintenance sub-ship in the path ; represents that when the path is selected, then , otherwise ; ; represents that when the operation and maintenance sub-ship repairs the wind turbine in the path , then , otherwise .

[0044] In the main problem model, formula (23) is the objective function, which represents the lowest total operation and maintenance cost (mother ship navigation cost + other costs); formula (24) represents that the selected feasible path is consistent with the position of the mother ship stop; formula (25) represents that a sub-ship cannot perform multiple maintenance tasks at the same time; formula (26) represents that each wind turbine can be maintained only once; and formula (27) represents the integer constraint of the decision variable.

[0045] To accelerate the solution speed, the main problem is relaxed. First, the partially feasible path The restricted master problem is formed in the initialization step, and then the integer constraints of the decision variables are relaxed, and the restricted master problem becomes the relaxed restricted master problem (RRMP): (28); (29); (30); (31); (32); The relaxed restricted master problem can be solved by the GUROBI optimizer, and the dual variables for the sub-models are obtained. The model constructed by the partially feasible path and constraints (24)-(27) is the restricted master problem, and the result obtained by the relaxed restricted master problem is the lower bound of the restricted master problem.

[0046] 2. Pricing sub-problem For the sub-problem, the stopping points of the operation and maintenance mother ship in the cycle are known, so the navigation cost of the mother ship is known. Let and be the dual variables of constraints (30) and (31) respectively, then the check number can be expressed as: (33); Because the objective function is a minimization problem, the column of can be found and added to the restricted relaxed master problem for iteration until no column can be found to make the check number less than 0. The objective function of the sub-problem is expressed as: (34); In the formula: represents the set of nodes that need to be repaired in the cycle ; represents the set of nodes that need to be repaired in the cycle , where the starting node of the operation and maintenance sub-ship is consistent with the stopping node of the mother ship; represents the set of nodes that need to be repaired in the cycle , where the return node of the operation and maintenance sub-ship is consistent with the stopping node of the mother ship; represents the set of nodes that need to be repaired in the cycle ; where the starting node of the operation and maintenance sub-ship is consistent with the stopping node of the mother ship; represents the set of nodes that need to be repaired in the cycle ; where the return node of the operation and maintenance sub-ship is consistent with the stopping node of the mother ship; represents the path of the operation and maintenance sub-ship from node to node all costs; denotes if the service sub-ship is in path from node to node , then , otherwise ; denotes if the wind turbine to be serviced is serviced by the sub-ship in path , then , otherwise .

[0047] The pricing sub-problem constraints are expressed as: (35); (36); (37); (38); (39) (40); (41); (42); (43); (44); (45); (46); (47); (48); (49); where: denotes the time the service sub-ship leaves the mother ship docking point in path ; denotes the time the service sub-ship returns to the mother ship docking point in path ; denotes the time the service sub-ship arrives at node in path ; denotes the time the service mother ship arrives at docking node in path ; denotes the sea time window period.

[0048] Among them, equations (35) and (36) indicate that the maintenance sub-vessel departs from the mother ship and returns to the mother ship after completing the maintenance task; equation (37) indicates that each wind turbine can only be maintained once; equation (38) records the time when the maintenance sub-vessel leaves the mother ship in the path; equation (39) records the time when the maintenance sub-vessel returns to the mother ship in the path; equations (40) and (41) indicate that the path must meet the capacity constraint of the maintenance sub-vessel; equations (42) and (43) indicate that the path must meet the capacity constraint of the maintenance mother ship; equations (44), (45) and (46) indicate the continuity of maintenance time; equation (47) indicates the continuity of maintenance equipment; equation (48) indicates that the path must meet the time window period; equation (49) indicates the value range of all decision variables.

[0049] Step 4: Solve the main problem model and sub-problem model based on the branch-and-price method and output integer solutions.

[0050] The branch-and-price method combines the column generation algorithm with the branch-and-bound method to solve the problem. The branch-and-price method consists of two layers: the outer layer performs the branch-and-bound method to find integer solutions to the original problem; the inner layer uses the column generation method to find the linear relaxation solution to the original problem. In addition, the present invention also redefines the "column" in the column generation algorithm, where each column represents the operation and maintenance mother ship in the period Docked at the node The process design of the branch pricing method is shown in the attached figure. Figure 3 As shown: S4.1. Use the greedy algorithm to obtain the initial path of the main problem; S4.2. Initialize the search tree, add the root node, and select some nodes to construct the restricted main problem model; S4.3. Introduce relaxation constraints to relax the restricted master problem model into a restricted relaxed master problem model; S4.4. Solve the restricted relaxation master problem model to obtain the dual variables for the subproblem model; S4.5. Use dual variables to construct sub-problem models and find solutions that satisfy The corresponding solution, ; S4.6. During each iteration of the main problem and sub-problem, one or more columns that can minimize the sub-problem objective function are generated and incorporated into the restricted relaxation main problem model. New columns are continuously added until no more columns can be used. When the column appears, the optimal solution of the original problem can be obtained; S4.7. Use the branch and bound method to process the output non-integer solution, convert it into an integer solution and then output it.

[0051] In this embodiment, the sub-problem is solved by using an accelerated label algorithm, which is a dynamic programming method commonly used to solve the shortest path problem. In the label algorithm, the label represents the partial path state of the maintenance ship from the starting point to a certain node, and records the consumption of all resources on the path, and ensures that it does not exceed the upper limit of the resource constraint. Each label can be expanded to a set of new labels through feasible arcs until the resources are exhausted. In order to speed up the solution, a bidirectional search strategy (forward expansion and backward expansion) and a pruning strategy are introduced.

[0052] In the forward label expansion setting, the label represents a feasible path access point state. For the label terminated at point , if point satisfies , , and , the label can be expanded to a new label terminated at point through arc , and the expansion equation of the new label can be expressed as: (50); (51); (52); (53); (54); (55); wherein: represents the inspection number of the maintenance sub-ship reaching node on the path; represents the number of technical personnel required by the maintenance sub-ship to reach node on the path; represents the weight of the maintenance spare parts required by the maintenance sub-ship to reach node on the path; represents the time of the maintenance sub-ship reaching node on the path; represents the number of nodes visited by the maintenance sub-ship reaching node on the path; represents the set of nodes that can be reached by the maintenance sub-ship on the path.

[0053] Although all feasible labels can be generated by expanding equations, the number of labels will increase exponentially in the face of large-scale problems. To speed up the solution, some dominant rules can be made to effectively screen and eliminate some invalid labels, i.e. perform the dominant criterion on a bunch of labels that terminate at the same node, effectively reducing the number of labels.

[0054] Given two labels that both terminate at point , and , , they can be considered dominant when the following conditions are met: (56); (57); (58); (59); (60); Since can expand the solution generated by , will continue to expand in the next iteration, while is deleted.

[0055] In backward label setting, like forward label, define backward label as , which represents the state of the partial path from point to the source point. If point satisfies , , and , it can be expanded to a new label that terminates at point through arc . The expansion equation of new label can be expressed as: (61); (62); (63); (64); (65); (66); Similarly, the dominance is given two labels that both terminate at point , and , they can be considered dominant when the following conditions are met:When the following conditions are met, the label dominates : (67); (68); (69); (70); (71); In the process of bidirectional search, the label obtained by forward extension and the label obtained by backward extension need to be connected to form a complete path. For the forward label and the backward label both of which terminate at point , when the following conditions are met, the two labels can be connected to form a complete path: (72); (73); (74); (75); (76); Finally, the paths that do not meet the conditions are pruned, and the main pruning basis is the time window pruning strategy and the capacity constraint pruning: if there is or , it indicates that the time when the wind turbine to be maintained is reached exceeds the time window period, so the part of the route that exceeds the time window is not feasible and is pruned; if there is , , or , it indicates that the capacity of the operation and maintenance ship is exceeded, so the part of the route that exceeds the capacity is not feasible and is pruned.

[0056] Branching strategy is a method for accelerating the convergence of an algorithm and obtaining an integer solution of a problem in the branch and price framework. At each node of the branch and price tree, the method solves the main problem through a column generation process to obtain a relaxed solution of the main problem, which is the lower bound of the node. If the lower bound is not less than the upper bound value of the current solution, the node is pruned and deleted; otherwise, branching operation is performed on the node. If the obtained solution is feasible and better than the current upper bound, the current upper bound is updated.

[0057] For a sub-problem, branching on the column variable will make the solution of the sub-problem more complex. Therefore, the method should prefer branching strategies that do not change the structure of the sub-problem. The branching bound algorithm of the present application will use an arc branching strategy: let is a 0-1 variable, and when the variable is 1, it represents an arc is a path is accessed. Let be the arc closest to 0.5 is branched. At one node, let be the arc that is forbidden to be accessed by the ship can be removed from the tree graph to achieve. At another node, let be the arc that is forced to be accessed by the ship .

[0058] Example 2 To verify the necessity of optimization of the present application and the feasibility of the method, a certain offshore wind farm project is taken as an example for simulation verification. Taking a certain offshore wind farm as an example, the wind turbine T1 of the offshore wind farm is about 70 km away from the base, the distance between wind turbines is 3 km, 60 3MW wind turbine generators are installed in the wind farm, the annual effective utilization hours are 2200h, and the wind power price is taken as 0.302 yuan / kWh according to the onshore wind power price in China. The position and state of the offshore wind farm are shown in Figure 5 .

[0059] There are 16 wind turbines to be maintained in the offshore wind farm, and it is set that the maintenance task needs to be completed within 3 days, and the time window periods are 22h, 18h and 20h respectively. The specific information of the wind turbines to be maintained is shown in Table 1:

[0060] The average repair time of important components of the unit is shown in Table 2:

[0061] The classification of technicians and their costs are shown in Table 3:

[0062] The parameter information of the operation and maintenance mother ship and the sub-ships carried by it is shown in Table 4:

[0063] Through the PSO-KMEANS clustering method, the wind turbines to be maintained are divided according to periodic tasks by comprehensively considering multiple characteristic data such as the distance of the wind turbine and the operation and maintenance time window period. The division of each periodic maintenance area is shown in Figure 6 .

[0064] In order to verify the superiority of the scheme of the present application in the operation and maintenance of offshore wind power, three different schemes are designed for comparative analysis by using the traditional nearshore operation and maintenance method, the operation and maintenance method using only the mother ship and the operation and maintenance method combining the mother ship and the daughter ship. Scheme 1 is the traditional nearshore operation and maintenance scheme, which is to send an operation and maintenance ship from the coastal base to complete the operation and maintenance task by constantly going back and forth to the wind farm (assuming that the base has only one type of operation and maintenance ship, and the technical personnel and maintenance spare parts are sufficient). Scheme 2 is the operation and maintenance scheme using only the mother ship, which uses the mother ship to go to each wind turbine that needs to be maintained to complete the maintenance task. Scheme 3 is the multi-cycle offshore operation and maintenance scheme combining the mother ship and the daughter ship proposed by the present application, which uses the mother ship to go to each maintenance cycle docking point, and then the daughter ship constantly goes back and forth to the mother ship to complete each cycle of maintenance task. Among them, schemes 1 and 2 use a single operation and maintenance base-single wind farm model for solution, and scheme 3 uses the model established by the present application for solution.

[0065] The operation and maintenance path and navigation cost of the three schemes are compared in Table 5.

[0066]

[0067] Figures 7-9 The operation and maintenance path of the operation and maintenance ship of the three schemes in the first maintenance cycle is shown in Table 6.

[0068] Figures 10-12 The operation and maintenance path of the operation and maintenance ship of the three schemes in the second maintenance cycle is shown in Table 7.

[0069] Figures 13-15 The operation and maintenance path of the operation and maintenance ship of the three schemes in the second maintenance cycle is shown in Table 7.

[0070] As can be seen from Table 5, the navigation cost of scheme 1 is the highest, especially in the second maintenance cycle, the navigation cost of scheme 1 is much higher than the other two schemes. This is because scheme 1 relies on the CTV sent from the coastal base to complete the operation and maintenance task of the wind farm. In this mode, the CTV directly departs from the base, goes to each wind turbine location for maintenance, and then returns to the base. Since the offshore wind farm is generally tens of kilometers offshore, in order to complete the operation and maintenance task within a limited time window, multiple CTVs need to be sent at one time, which results in a large increase in navigation cost. In schemes 2 and 3, since the SOV has a large capacity, has the ability to store spare parts, and has a large oil, water and living material storage cabin and a long self-sustaining force, it does not need to go back and forth to the coastal base for replenishment, so a lot of navigation cost can be saved.

[0071] In addition, it can be clearly observed from the data comparison that, although scheme 2 uses SOV for operation and maintenance, the navigation cost in three different periods is 10.21%, 39% and 19.94% higher than that of scheme 3. This shows that there is a certain limitation in relying on SOV operation and maintenance only. Although it has the advantages of large carrying capacity, its relatively slow speed and high navigation cost cannot be ignored. On the contrary, scheme 3 combines the long-distance navigation ability of SOV and the flexibility of CTV, not only improves the efficiency of operation and maintenance, but also significantly reduces the overall navigation cost.

[0072]

[0073] Table 6 shows the cost comparison of the three schemes. Since the three schemes use the same example, the crew repair cost and personnel cost are the same. In terms of penalty cost, scheme 1 is 46.39% lower than scheme 2, and scheme 3 is 31.51% lower than scheme 2. This is because the wind turbine will lose power generation due to downtime before maintenance is completed, resulting in a penalty cost. In scheme 1, since multiple CTVs can be dispatched from the coastal base to perform maintenance tasks at one time, simultaneous maintenance of multiple wind turbines is realized, thereby effectively shortening the overall maintenance time and reducing the power generation loss due to downtime. On the contrary, scheme 2 is limited by the number and speed of SOV, and cannot perform maintenance on multiple wind turbines at the same time, resulting in an extended maintenance period and an increase in penalty cost due to downtime. In scheme 3, CTV can assist SOV in wind turbine maintenance, thereby effectively reducing the penalty cost. In terms of total cost, scheme 3 is 10.87% and 4.07% lower than scheme 1 and scheme 2, respectively. Therefore, it is proved that the scheme of the present application has superiority in improving the efficiency and economy of offshore operation and maintenance.

[0074] To sum up, the present application firstly divides the wind turbine to be maintained into maintenance periods according to the PSO-KMEANS method by comprehensively considering multiple feature data such as the distance of the wind turbine and the offshore time window; then, an integer programming model of multi-period offshore wind power operation and maintenance is established by taking the minimum comprehensive operation and maintenance cost as the objective function; then, in the framework of branch and price, the model is decomposed into master and sub-problems by the Dantzig-Wolfe decomposition principle; the column generation algorithm and the label setting method with acceleration strategy are used to iterate between the master and sub-problems to obtain the optimal solution of the linear relaxation of the master problem; finally, the arc-based branching strategy is used to obtain the integer solution. In order to verify the applicability of the proposed model, the traditional nearshore operation and maintenance scheme and the operation and maintenance scheme using only the mother ship are compared. The final result shows that the total cost of the proposed scheme is 10.87% and 4.07% lower than that of the traditional nearshore operation and maintenance scheme and the operation and maintenance scheme using only the mother ship respectively. It is proved that the proposed model not only improves the efficiency of offshore operation and maintenance, but also significantly reduces the overall operation and maintenance cost.

[0075] Those skilled in the art will easily understand that the above description is only the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A multi-period offshore wind power operation and maintenance scheduling optimization method based on the combination of mother and daughter ships, characterized in that: The following steps are involved: Step 1: Construct the PSO-KMEANS wind turbine clustering method and divide the wind turbines to be repaired according to the maintenance task cycle; Step 2: With the goal of minimizing the comprehensive operation and maintenance cost, an integer programming model for offshore wind power operation and maintenance based on the combination of mother and daughter ships is proposed; Step 3: Based on the Danziger-Wolff decomposition principle, the offshore wind power operation and maintenance integer programming model based on the mother-daughter ship combination is decomposed into a path-based main problem and a pricing sub-problem, forming a main problem model and a sub-problem model; Step 4: Solve the main problem model and sub-problem model based on the branch-and-price method and output integer solutions.

2. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the combination of mother and daughter ships according to claim 1 is characterized in that: In step 1, the PSO-KMEANS wind turbine clustering method includes the following steps: S1.1, initialize particle swarm and cluster centers; S1.

2. Update the particle's velocity and position. Based on the new particle position, assign the wind turbine to the nearest cluster center while ensuring that the total maintenance time of each cluster does not exceed the window period. S1.

3. Calculate the fitness function of the current particle. The fitness function of the current particle is the sum of the distances between all wind turbines. The goal is to minimize the fitness function value. S1.

4. Update the individual optimal solution and the global optimal solution: If the fitness of the current particle is better than its historical optimal fitness, then update the individual optimal solution; if the fitness of the current particle is better than the global optimal fitness, then update the global optimal solution; S1.

5. Move each particle according to the updated velocity and position. S1.

6. Repeat steps S1.2 to S1.5 until the maximum number of iterations is reached; S1.

7. Output the corresponding result based on the historical optimal solution of the group.

3. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-daughter ship combination according to claim 1 is characterized in that: In step 2, the objective function of the offshore wind power operation and maintenance integer programming model based on the mother-daughter ship combination is: （1）； Where: represents the comprehensive operation and maintenance cost; Indicates the operation and maintenance cycle, ; Represents the set of locations where the operation and maintenance mother ship can dock. ; Represents the set of all nodes accessible to the operation and maintenance sub-ship, ; Represents a collection of various types of maintenance technicians, ; Representation node To Node the distance between them; Represents a slave node To Node the distance between them; Indicates the fuel cost per unit distance traveled by the operation and maintenance mother ship; Indicates the fuel cost per unit distance of the operation and maintenance sub-vessel; Indicates maintenance fan the expenses required; It represents the cost of power generation loss caused by wind turbine downtime per hour; Indicates technical personnel The hourly wage to be paid; Indicates maintenance fan Required technical personnel the number of Indicates that the maintenance sub-ship has arrived at the node time; Indicates maintenance fan the time required; Indicates maintenance fan Time to transfer spare parts; Indicates that when the maintenance sub-ship is in the cycle Slave Node To Node ,but ,otherwise ; Indicates that when the operation and maintenance mother ship is in the cycle Slave Node To Node ,but ,otherwise .

4. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-daughter ship combination according to claim 3 is characterized in that: In step 2, the constraints of the offshore wind power operation and maintenance integer programming model based on the mother-daughter ship combination are: 1) The mother ship can only choose one docking point in each cycle: （2）； 2) Each fan can only be maintained once: （3）； 3) The maintenance sub-vessel departs from the mother ship and returns to the mother ship after completing the maintenance task: （4）； （5）； 4) The traffic balance constraint is: （6）； （7）； 5) The spare parts for wind turbine maintenance shall not exceed the maximum spare parts load of the daughter ship: （8）； 6) The number of technicians carried by the daughter vessel does not exceed its maximum carrying capacity: （9）； 7) The resources required for operation and maintenance tasks shall not exceed the total amount of spare parts of the mother ship: （10）； （11）； 8) Continuity constraints on operation and maintenance time: （12）； （13）； （14）； 9) Continuity of maintenance equipment; （15）； 10) Operation and maintenance time does not exceed the time window period: （16）； 11) Integer constraints on decision variables: （17）； Where: Represents the set of all wind turbine nodes to be repaired, ; Indicates maintenance fan Required spare parts the number of Indicates the technical personnel carried by the operation and maintenance mother ship the number of Indicates the maintenance spare parts carried by the operation and maintenance mother ship the number of Indicates that the wind farm is in period The time window period; Indicates that the operation and maintenance mother ship has arrived at the docking node time; Indicates the time for transferring personnel and spare parts between the operation and maintenance mother ship and the daughter ship; Indicates the navigation speed of the maintenance sub-vessel; Indicates the spare parts load of the operation and maintenance sub-vessel; Indicates the technician carrying capacity of the operation and maintenance sub-vessel; Indicates maintenance fan The weight of the spare parts required; Indicates that the maintenance vessel has left the wind turbine Spare parts weight at the time of use.

5. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-daughter ship combination according to claim 1 is characterized in that: In step 3, the main problem model divides the wind turbines to be repaired into different sets by using a set covering method; The objective function of the main problem model is to minimize the total operation and maintenance cost; the constraints are: the feasible path is consistent with the docking location of the mother ship, a daughter ship cannot perform multiple maintenance tasks at the same time, each wind turbine can only be maintained once, and the policy variable is an integer constraint.

6. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-daughter ship combination according to claim 5 is characterized in that: The objective function expression of the main problem model is: （18）； The constraints of the main problem model are: 1) The feasible path is consistent with the mother ship's docking location: （19）； 2) A sub-ship cannot perform multiple maintenance tasks at the same time: （20）； 3) Each fan can only be maintained once: （21）； 4) Integer constraints on decision variables （22）； Where: represents the total operation and maintenance cost; Indicates that the mother ship is in the cycle Docked at the node The set of all feasible paths for the sub-ship; Indicates that the maintenance sub-vessel is on the path All expenses incurred; Indicates that the path When selected, ,otherwise ; Indicates that when the maintenance sub-vessel is on the path Medium maintenance fan When ,otherwise .

7. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the combination of mother and daughter ships according to claim 5 is characterized in that: In step 3, the main problem model is relaxed, and the operation steps are as follows: 1) Partially feasible paths Incorporate into the initialization step to form a restricted master problem; 2) Relax the integer constraints of decision variables, transform the restricted master problem into a restricted relaxed master problem, and convert all feasible paths Change to partially feasible path , integer variables become continuous variables: （23）； 3) Solve the restricted relaxed master problem through the optimizer to obtain the dual variables used for the subproblem model.

8. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-daughter ship combination according to claim 7 is characterized in that: In step 3, the objective function of the pricing sub-problem is: （24）； Where: Represents a period The collection of wind turbine nodes that need to be repaired, ; Indicates the starting node and cycle of the maintenance sub-ship The collection of wind turbine nodes that need to be repaired, ,Among them, the starting node of the operation and maintenance daughter ship is consistent with the docking node of the mother ship; Indicates the return node and cycle of the maintenance sub-ship The collection of wind turbine nodes that need to be repaired, ; Among them, the return node of the operation and maintenance daughter ship is consistent with the docking node of the mother ship; Indicates that the maintenance ship is on the path Middle slave node To Node All expenses; If the maintenance sub-vessel is on the path Middle slave node To Node ,but ,otherwise ; If the fan is to be repaired On the path During the maintenance of the quilt boat, ,otherwise ; in, and Constraints (20) and (21) for all feasible paths Change to partially feasible path The dual variable of .

9. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-and-daughter combination according to claim 8 is characterized in that: The constraints of the pricing sub-problem are: 1) The operation and maintenance sub-vessel departs from the mother ship and returns to the mother ship after completing the operation and maintenance task: （25）； （26）； 2) Each fan can only be maintained once: （27）； 3) Time when the maintenance sub-vessel leaves the mother ship: （28）； 4) Time for the maintenance sub-vessel to return to the mother ship: （29）； 5) The path must meet the capacity constraints of the operation and maintenance sub-vessel （30）； （31）； 6) The path must meet the capacity constraints of the operation and maintenance mother ship: （32）； （33）； 7) Continuity of operation and maintenance time （34）； （35）； （36）； 8) Continuity of maintenance equipment: （37）； 9) The path must meet the time window （38） 10) The value range of all decision variables （39）； Where: Indicates that the maintenance sub-vessel is on the path The time of departure from the mother ship's docking point; Indicates that the maintenance sub-vessel is on the path The time to return to the mother ship's docking point; Indicates that the maintenance sub-vessel is on the path Arrival Node time; Indicates that the operation and maintenance mother ship is on the path Arrival at the docking point time; Indicates the maritime time window.

10. The multi-period offshore wind power operation and maintenance scheduling optimization method based on the mother-and-daughter combination according to claim 1 is characterized in that: The solution process of step 4 is: S4.

1. Use the greedy algorithm to obtain the initial path of the main problem; S4.

2. Initialize the search tree, add the root node, and select some nodes to construct the restricted main problem model; S4.

3. Introduce relaxation constraints to relax the restricted master problem model into a restricted relaxed master problem model; S4.

4. Solve the restricted relaxation master problem model to obtain the dual variables for the subproblem model; S4.

5. Use dual variables to construct sub-problem models and find solutions that satisfy The corresponding solution, ; S4.

6. During each iteration of the main problem and sub-problem, one or more columns that can minimize the sub-problem objective function are generated and incorporated into the restricted relaxation main problem model. New columns are continuously added until no more columns can be used. When the column appears, the optimal solution of the original problem can be obtained; S4.

7. Use the branch and bound method to process the output non-integer solution, convert it into an integer solution and then output it.

Images