An offshore maintenance resource collaborative scheduling optimization method considering the influence of wind wave

Abstract

This invention belongs to the field of marine maintenance resource decision-making technology, and particularly relates to a collaborative scheduling optimization method for marine maintenance resources that considers the impact of wind and waves. This collaborative scheduling optimization method quantifies wind and wave levels, making the generated scheduling scheme more consistent with actual sea conditions, thus improving the reliability and executability of the maintenance scheme. Furthermore, by constructing a topology network including transfer nodes, it expands the original single "point-to-point" scheduling mode into a flexible "point-transfer-point" collaborative scheduling mode. The collaborative scheduling optimization method includes: Step S1: Constructing a marine maintenance resource topology network and determining the matching rules of the topology network; Step S2: Calculating the dynamic path weights that incorporate the impact of wind and waves; Step S3: Quantifying and calculating the total collaborative scheduling time including waiting time; Step S4: Constructing a multi-objective evaluation model; Step S5: Using an improved particle swarm optimization algorithm to optimize and solve the collaborative scheduling scheme for marine maintenance resources.

Description

Technical Field

[0001] This invention belongs to the field of marine maintenance resource decision-making technology, and particularly relates to a method for optimizing the collaborative scheduling of marine maintenance resources that takes into account the impact of wind and waves. Background Technology

[0002] Offshore equipment is exposed to harsh marine environments characterized by high salinity, high humidity, and strong winds and waves. Its stable operation is highly dependent on a rapid-response maintenance and support system. If critical equipment malfunctions and maintenance resources are not deployed in a timely manner, it can lead not only to production interruptions and reduced capacity but also to safety accidents, significant economic losses, and even severe ecological disasters.

[0003] However, further research revealed significant technical flaws in existing offshore maintenance resource scheduling technologies, severely restricting the emergency response capabilities and resource utilization efficiency of maintenance support systems. Specifically, existing offshore maintenance resource scheduling models typically involve independent decision-making on a single platform or regional basis, resulting in the inability to coordinate and allocate resources such as maintenance vessels, technical personnel, spare parts inventory, and underwater operation equipment across platforms and regions. When a sudden multi-point failure occurs in a localized area, a structural imbalance arises, with resources idle on one side and in urgent need on the other, greatly weakening the overall support capability of maintenance resources. Furthermore, existing offshore maintenance resource scheduling processes are mostly limited to direct scheduling methods from "resource point to failure point," lacking exploration of the collaborative potential of intermediate transfer nodes (such as maintenance mother ships, supply platforms, and transfer bases). Faced with complex scenarios such as remote offshore operations, large equipment transportation, or insufficient resource matching, the inability to fully utilize transfer relay mechanisms leads to an artificially compressed feasible solution space, resulting in a severe lack of scheduling flexibility and robustness.

[0004] Therefore, it is imperative for those skilled in the art to design a new collaborative scheduling and optimization method for maintenance resources that can deeply integrate the dynamic marine environment and interconnect resources across the entire domain, thereby fundamentally improving the resilience and intelligence level of the offshore maintenance system and meeting the urgent needs of modern offshore equipment for highly reliable, low-cost, and rapid-response maintenance support. Summary of the Invention

[0005] This invention provides a method for optimizing the collaborative scheduling of marine maintenance resources, taking into account the impact of wind and waves. This method quantifies wind and wave levels, making the generated scheduling schemes more consistent with actual sea conditions and improving the reliability and feasibility of maintenance plans. Furthermore, by constructing a topology network including transfer nodes, it expands the original single "point-to-point" scheduling mode into a flexible "point-transfer-point" collaborative scheduling mode, providing a feasible and economical alternative when resources are unavailable or direct costs are too high.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] A collaborative scheduling optimization method for marine maintenance resources that takes into account the impact of wind and waves includes the following steps:

[0008] Step S1: Construct a marine maintenance resource topology network and determine the matching rules for the topology network;

[0009] Step S2: Calculate the dynamic path weights that incorporate the effects of wind and waves;

[0010] Step S3: Quantify and calculate the total collaborative scheduling time, including waiting time;

[0011] Step S4: Construct a multi-objective evaluation model;

[0012] Step S5: Use an improved particle swarm optimization algorithm to optimize and solve the collaborative scheduling scheme for marine maintenance resources.

[0013] Preferably, the marine maintenance resource topology network constructed in step S1 satisfies: G=(V,E,W); where V represents the set of nodes, E represents the set of edges, and W represents the path weight.

[0014] A set of nodes V that satisfies: ;

[0015] Among them, V M For a set of maintenance resource nodes, satisfying: (1); V T Let be a set of transit nodes, satisfying: (2); V O Let the set of job nodes satisfy: (3);

[0016] in, The i-th maintenance node in the set of maintenance resource nodes; It is the kth transit node in the set of transit nodes; Let j be the j-th job node in the set of job nodes;

[0017] The edge set E is determined by the directly scheduled edge set. and transfer scheduling edge set constitute;

[0018] And, nodes With nodes The distance between them satisfies: (4);

[0019] When a transfer occurs, the distance between the two nodes satisfies: (5);

[0020] Where e is an edge in the edge set E; This is an indicator function that takes the value 1 when the event is true and 0 when the event is false; that is, an indicator function. ,satisfy: (6).

[0021] Preferably, in step S2, the path weight W is determined by the directly scheduled edge weight. and the weight of the transfer scheduling edge The path weights are determined together; that is, the path weights W satisfy: (7);

[0022] Among them, direct scheduling edge weight ,satisfy: (8);

[0023] In equation (8), The weighting coefficients are determined by experts, prioritizing cost and the urgency of operational support. For work nodes The risk level weight, with a value ranging from 1 to 5; ω max The highest risk level among all work nodes is fixed at level 5; d max This represents the maximum distance between all scheduled paths, including direct edges and transmission edges.

[0024] C i For maintenance resources The unit time cost satisfies: (9);

[0025] In equation (9), f i For maintenance resources fuel consumption per unit distance; C f The price per unit of fuel; v i,max For maintenance resources Maximum sailing speed;

[0026] Path reliability coefficient ,satisfy: Its functional relationship with the wave level ξ satisfies: (10);

[0027] In equation (10), The critical wind and wave level at which path reliability begins to decrease; This is the attenuation coefficient of path reliability as the wind and wave level decreases, used to control the steepness of the reliability decline;

[0028] And, the weight of the transfer scheduling edge. ,satisfy: (11);

[0029] In equation (11), C k For transit node T k The unit time cost; For transit node T k One-time costs;

[0030] Furthermore, by combining equations (8) and (11), the maximum cost is obtained. ,satisfy: .

[0031] Preferably, the process of quantifying and calculating the total coordinated scheduling time including the waiting time in step S3 specifically includes the following steps:

[0032] Step S31: Quantify the total coordinated scheduling time, including waiting time. The components of;

[0033] Among them, the total coordinated scheduling time ,satisfy: (13);

[0034] Step S32: Calculate the sailing time And determine how the speed decreases as the wave level ξ increases. ;

[0035] Among them, sailing time ,satisfy: (14);

[0036] The decrease in speed as the sea level ξ increases ,satisfy: (15);

[0037] In equation (15), For maintenance resources The maximum speed of the ship is negatively correlated with the wave level ξ. The wind and wave sensitivity coefficient;

[0038] Step S33: Calculate preparation time;

[0039] The preparation time consists of the maintenance preparation time. and transfer preparation time constitute;

[0040] Step S34: Calculate the waiting time ;

[0041] Among them, waiting time ,satisfy: (16);

[0042] In equation (16), As the basic unit of time; It is a growth factor used to control waiting time as the load rate increases. The rate of increase; This is the critical value for the load rate of the transfer node; For the load factor, the following conditions must be met: (17);

[0043] In equation (17), μ is the maintenance resource category index; Q jμ For transit node T k The demand for the μth type of resource; For transmission node T k capacity; x ikj These are execution variables based on parameters i, k, and j.

[0044] Preferably, the total scheduling cost CS in the multi-objective evaluation model constructed in step S4 satisfies: (18);

[0045] in, For direct scheduling cost, the following condition must be met: (19);

[0046] In equation (19), x ij These are execution variables based on parameters i and j;

[0047] To transfer scheduling costs, the following must be satisfied: (20);

[0048] In equation (20), f i f k Repair resources and transfer resources T k fuel consumption per unit distance; C f It is the unit price of fuel;

[0049] It is a fixed cost that occurs only at the start of the scheduled task and satisfies the following: (twenty one);

[0050] In equation (21), x ij These are execution variables based on parameters i and j;

[0051] Downtime cost refers to the direct economic loss caused by the interruption of operations at a work node. It is positively correlated with recovery time and loss per unit time, satisfying the following: (twenty two);

[0052] In equation (22), For work nodes In time period Downtime losses within the facility; For work nodes Recovery time, which reflects the total time from a node's failure to its full recovery, satisfies: (twenty three);

[0053] In formula (23) For work nodes Maintenance time.

[0054] Preferably, the job assurance rate OS in the multi-objective evaluation model constructed in step S4 satisfies: (twenty four);

[0055] in, For the importance of node j, Let be the importance weight of node j. A higher importance weight indicates that the interruption of the node has a greater impact on the job availability rate. The length of the time window covered by the scheduling scheme.

[0056] Preferably, step S5 specifically includes the following steps:

[0057] Step S51: Construct a multi-objective optimization model with the objectives of minimizing total scheduling cost (CS) and maximizing job availability (OS);

[0058] The constructed multi-objective optimization model satisfies: (25);

[0059] in, The weighting coefficient for the total scheduling cost CS. The maximum total scheduling cost CS for all particles in the s-th iteration is given by [the value of CS]. The weighting coefficient for the job assurance rate (OS). The time-constrained penalty factor; For work nodes Maximum interruption tolerance time

[0060] Step S52: Constrain resource capacity;

[0061] The constrained resource capacity satisfies the following: (26);

[0062] In equation (26), For maintenance resources The demand for the μth resource; x ij For execution variables; where, Indicates activation from maintenance resources To the work node The scheduling task for allocating resources. Indicates that it will not start;

[0063] Step S53: Constrain the capacity of the transfer nodes;

[0064] The constrained capacity of the transfer nodes satisfies: (27);

[0065] In equation (27), x ij These are execution variables based on parameters i and j;

[0066] Step S54: Perform two-level discrete coding on the decision variables of the collaborative scheduling scheme for offshore maintenance resources;

[0067] The first layer of discrete coding is used to solve the resource allocation problem, satisfying: (28); Indicates assignment to job nodes Maintenance resource number;

[0068] The second-layer discrete coding is used to solve the path selection problem, satisfying: (29); Indicates the work node The scheduling method used;

[0069] Step S55: Initialize and set the algorithm parameters of the multi-objective optimization model, and generate a random initial particle swarm;

[0070] Update the individual optimal pbest and the global optimal gbest, and update the particle velocity and position;

[0071] The process continues iteratively until the optimal solution for the decision variables of the collaborative scheduling scheme for offshore maintenance resources is obtained.

[0072] This invention provides an optimization method for the collaborative scheduling of marine maintenance resources, taking into account the impact of wind and waves. The method includes the following steps: Step S1: Constructing a marine maintenance resource topology network and determining the matching rules for the topology network; Step S2: Calculating the dynamic path weights that incorporate the impact of wind and waves; Step S3: Quantifying and calculating the total collaborative scheduling time, including waiting time; Step S4: Constructing a multi-objective evaluation model; Step S5: Using an improved particle swarm optimization algorithm to optimize and solve the collaborative scheduling scheme for marine maintenance resources.

[0073] The collaborative scheduling optimization method for marine maintenance resources that takes into account the impact of wind and waves, and which features the above-mentioned steps, has at least the following technical advantages compared to existing technologies:

[0074] (1) The present invention provides a method for optimizing the collaborative scheduling of marine maintenance resources. By systematically quantifying the key environmental variable of wind and wave level into its impact on path safety and navigation efficiency, the generated scheduling scheme is more in line with the actual sea conditions, thereby improving the reliability and executability of the scheduling scheme. Furthermore, a topology network containing transfer nodes is constructed, and strict matching rules are defined, which expands the single "point-to-point" scheduling into a flexible "point-transfer-point" collaborative scheduling network, enriching the scheduling strategy and enabling feasible and economical alternatives to be found when resources are unavailable or direct costs are too high.

[0075] (2) The method for collaborative scheduling optimization of marine maintenance resources provided by the present invention truly reflects the queuing and congestion situation that maintenance resources may encounter at transfer nodes, avoids the disconnection of scheduling plans caused by underestimating waiting time, and significantly improves the accuracy of time planning. In addition, the designed dual-layer discrete coding PSO algorithm greatly compresses the solution space and can quickly generate high-quality scheduling schemes for large-scale complex scenarios. Attached Figure Description

[0076] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the following drawings:

[0077] Figure 1 A flowchart illustrating an optimization method for collaborative scheduling of marine maintenance resources that takes into account the impact of wind and waves, provided by this invention.

[0078] Figure 2 This is a schematic diagram of the architecture of the improved particle swarm optimization (PSO) algorithm used in this invention. Detailed Implementation

[0079] This invention provides a method for optimizing the collaborative scheduling of marine maintenance resources, taking into account the impact of wind and waves. This method quantifies wind and wave levels, making the generated scheduling schemes more consistent with actual sea conditions and improving the reliability and feasibility of maintenance plans. Furthermore, by constructing a topology network including transfer nodes, it expands the original single "point-to-point" scheduling mode into a flexible "point-transfer-point" collaborative scheduling mode, providing a feasible and economical alternative when resources are unavailable or direct costs are too high.

[0080] This invention provides a method for optimizing the collaborative scheduling of marine maintenance resources, taking into account the impact of wind and waves. Figure 1 As shown, it includes the following steps:

[0081] Step S1: Construct a marine maintenance resource topology network and determine the matching rules for the topology network.

[0082] In a preferred embodiment of the present invention, the marine maintenance resource topology network constructed in step S1 satisfies: G=(V,E,W); where V represents the set of nodes, E represents the set of edges, and W represents the path weight.

[0083] A set of nodes V that satisfies: .

[0084] Among them, V M For a set of maintenance resource nodes, satisfying: (1); V T Let be a set of transit nodes, satisfying: (2); V O Let the set of job nodes satisfy: (3);

[0085] in, The i-th maintenance node in the set of maintenance resource nodes; It is the kth transit node in the set of transit nodes; Let j be the j-th job node in the set of job nodes.

[0086] The edge set E is determined by the directly scheduled edge set. and transfer scheduling edge set constitute.

[0087] And, nodes With nodes The distance between them satisfies: (4). It is worth noting that for nodes With nodes In other words, Let x and y be the x and y coordinates of node A, respectively. These are the x and y coordinates of node B, respectively.

[0088] When a transfer occurs, the distance between the two nodes satisfies: (5).

[0089] Where e is an edge in the edge set E; This is an indicator function that takes the value 1 when the event is true and 0 when the event is false; that is, an indicator function. ,satisfy: (6).

[0090] To facilitate understanding of this solution by those skilled in the art, a specific embodiment is further provided. Assume a wind farm is located approximately 35-60 kilometers offshore, containing 30 wind turbine generators (hereinafter referred to as "wind turbines") with a single unit capacity of 8MW. For ease of management, the wind turbines are further divided into near-shore zones (O1-O2). 10 ), midfield area (O 11 -O 20 ) and the far-shore area (O 21 -O 30 Three regions. Based on the operational characteristics of the aforementioned offshore maintenance resources, a topology network is defined and constructed:

[0091] First, define node V: Maintenance resource nodes (V_M) include the onshore technical personnel base (M1), onshore spare parts warehouse (M2), onshore UAV control center (M3), and special operations vessel home port (M4) located at the main port. Transfer nodes (V_T) include the offshore maintenance mother ship (SOV, T1) permanently stationed at the wind farm, and an emergency transfer platform (T2) equipped with emergency spare parts and rapid hoisting equipment. Operation nodes (V_O) refer to the 30 wind turbines that have failed (O1 to O...). 30 ).

[0092] Then, define edge E and determine the matching rules of the topology network: determine the effective scheduling path based on the matching relationship between the possible failure modes of the wind turbine (such as blade damage, gearbox failure, structural alarm, electrical system failure) and the reserve resources of each node (such as professional teams, spare parts, tool kits).

[0093] Among them, the set of directly scheduled edges in edge set E This is established when the maintenance resource node's capabilities can directly meet the needs of the faulty wind turbine. For example, if the blade maintenance team (μ1) in the M1 reserve can be directly dispatched to the O1 wind turbine that has suffered blade damage, then edge e_M1O1 is formed.

[0094] Transfer scheduling edge set in edge set E This is established when maintenance resources cannot or are inconvenient to directly reach the faulty wind turbine and must be transferred through a transit node. For example, gearbox spare parts (μ4) in warehouse M2 need to be transported to the T1 maintenance mother ship first, and then from T1 to the O turbine where the gearbox failure occurred. 12 The wind turbine forms edge e_M2T1O12. It should be noted that the validity of this edge depends on the following double-matching condition: the spare parts type of M2 matches the transfer capacity of T1, and the transfer capacity of T1 matches O. 12 Fault requirements.

[0095] Then, network parameters are obtained: based on the actual layout of the wind farm, the maritime navigation distance matrix between all nodes is obtained or calculated. At the same time, the static performance parameters of each maintenance resource node (Mi) and transfer node (Tk) are determined, such as maximum speed v_max, fuel consumption per unit distance f, cost rate C, fixed start-up cost C_fix, and the reserve of various resources.

[0096] Step S2: Calculate the dynamic path weights that incorporate the effects of wind and waves.

[0097] Building upon step S1, step S2 is further implemented. Here, dynamic path weights are used to quantify the priority of dynamic paths. It's important to note that during path planning, the magnitude of the dynamic path weight represents the cost or resistance of traversing that path. Therefore, to obtain the path with the minimum total weight (i.e., prioritize the path with the minimum total weight), the dynamic path weights of different paths need to be calculated.

[0098] In a preferred embodiment of the present invention, the path weight W in step S2 is determined by the directly scheduled edge weight. and the weight of the transfer scheduling edge The path weights are determined together; that is, the path weights W satisfy: (7).

[0099] Among them, direct scheduling edge weight ,satisfy: (8);

[0100] In equation (8), The weighting coefficients are determined by experts, prioritizing cost and the urgency of operational support. For work nodes The risk level weight, with a value ranging from 1 to 5; ω max The highest risk level among all work nodes is fixed at level 5; d max This represents the maximum distance between all scheduled paths, including direct edges and transport edges.

[0101] C i For maintenance resources The unit time cost satisfies: (9);

[0102] In equation (9), f i For maintenance resources fuel consumption per unit distance; C f The price per unit of fuel; v i,max For maintenance resources Maximum sailing speed.

[0103] Path reliability coefficient ,satisfy: Its functional relationship with the wave level ξ satisfies: (10);

[0104] In equation (10), The critical wind and wave level at which path reliability begins to decrease; This is the attenuation coefficient of path reliability as the wind and wave level decreases, used to control the steepness of the reliability decline.

[0105] And, the weight of the transfer scheduling edge. ,satisfy: (11);

[0106] In equation (11), C k For transit node T k The unit time cost; For transit node T k One-time cost.

[0107] It is worth noting that by combining equation (8) and equation (11), the maximum cost is obtained. ,satisfy: .

[0108] One additional point to note is that, to complete the calculation of the dynamic path weights, it is necessary to further define fault scenarios: for example, simulating a typical stress scenario where multiple areas within a wind farm experience simultaneous faults under average wind and wave conditions (e.g., sea state 4). Specifically, blade damage occurs in the nearshore areas O1-O3, and in the mid-field area O... 12 O 13 Gearbox failure occurred, O 15 A structural alarm has been triggered in the offshore area O. 29 O 30 An electrical system failure has occurred.

[0109] Then, input dynamic environmental parameters to obtain the current or predicted wind and wave levels. Based on the model fitted using historical data, the impact coefficient of wind and waves on path reliability is determined. Critical wave levels at which reliability begins to decline And the critical wave levels that affect sailing speed. .

[0110] Then, the dynamic weights are calculated: for each valid scheduling edge (e.g., e_M1O1), according to the formula... Calculate the path reliability coefficient under the current wind and wave conditions. And according to the formula Estimate the actual sailing speed of each maintenance resource under the current wind and wave conditions. After calculating the dynamic weights of the direct scheduling edge and the transfer scheduling edge, normalize them.

[0111] Step S3: Quantify and calculate the total collaborative scheduling time, including waiting time.

[0112] Based on step S2, step S3 is further implemented. In a preferred embodiment of the present invention, step S3, which quantifies and calculates the total coordinated scheduling time including the waiting time, specifically includes the following steps:

[0113] Step S31: Quantify the total coordinated scheduling time, including waiting time. It consists of various parts.

[0114] Among them, the total coordinated scheduling time ,satisfy: (13).

[0115] Step S32: Calculate the sailing time And determine how the speed decreases as the wave level ξ increases. .

[0116] Among them, sailing time ,satisfy: (14).

[0117] The decrease in speed as the sea level ξ increases ,satisfy: (15);

[0118] In equation (15), For maintenance resources The maximum speed of the ship is negatively correlated with the wave level ξ. Let ξ be the wave sensitivity coefficient. It should be noted that, based on the basic principles of fluid mechanics, wave resistance is proportional to the square of the wave height and the wave height increases linearly with the wave level. When ξ is large, its attenuation rate increases significantly, exhibiting the characteristic of weak influence in low waves and rapid deterioration in high waves. Thus, a quadratic exponential velocity attenuation model is constructed.

[0119] Step S33: Calculate the preparation time.

[0120] The preparation time consists of the maintenance preparation time. and transfer preparation time constitute.

[0121] Step S34: Calculate the waiting time .

[0122] Among them, waiting time ,satisfy: (16);

[0123] In equation (16), As the basic unit of time; It is a growth factor used to control waiting time as the load rate increases. The rate of increase; This is the critical value for the load rate of the transfer node; For the load factor, the following conditions must be met: (17).

[0124] In equation (17), μ is the maintenance resource category index; Q jμ For transit node T k The demand for the μth type of resource; For transmission node T k capacity; x ikj These are execution variables based on parameters i, k, and j.

[0125] It should be noted that, for example, the calculation process for the total scheduling time of each possible scheduling task (such as "send M1 to O2 via T1") can be referred to as follows:

[0126] sailing time Using the calculated actual speed v(ξ), the travel time for each segment is calculated and summed.

[0127] Preparation time: Time spent on maintenance and preparation before departure (e.g., personnel assembly, spare parts issuance) and transfer preparation time at transfer nodes (e.g., during hoisting preparation) perform summation.

[0128] Waiting time Assess the real-time load rate of transit nodes (such as T1). .like Below the preset threshold If the waiting time is zero, then the waiting time is recorded as zero; otherwise, it follows the exponential model. This involves calculating the queuing time. and The parameters are based on historical data and will not be elaborated on here.

[0129] Step S4: Construct a multi-objective evaluation model.

[0130] Building upon step S3, step S4 is further implemented. It should be noted that for the constructed multi-objective evaluation model, the engineering value of the offshore maintenance resource collaborative scheduling scheme is measured from the following two dimensions: First, cost, which specifically includes navigation fuel costs, fixed start-up costs, and losses due to downtime. Second, operational availability rate, used to measure the offshore maintenance resource collaborative scheduling scheme's ability to ensure production continuity, calculating and considering the percentage of actual available time weighted by node importance.

[0131] In a preferred embodiment of the present invention, the total scheduling cost CS in the multi-objective evaluation model constructed in step S4 satisfies: (18).

[0132] in, For direct scheduling cost, the following condition must be met: (19).

[0133] In equation (19), x ij These are execution variables based on parameters i and j;

[0134] To transfer scheduling costs, the following must be satisfied: (20);

[0135] In equation (20), f i f k Repair resources and transfer resources T k fuel consumption per unit distance; C f That is the unit price of fuel.

[0136] It is a fixed cost that occurs only at the start of the scheduled task and satisfies the following: (twenty one).

[0137] In equation (21), x ij These are execution variables based on parameters i and j;

[0138] Downtime cost refers to the direct economic loss caused by the interruption of operations at a work node. It is positively correlated with recovery time and loss per unit time, satisfying the following: (twenty two).

[0139] In equation (22), For work nodes In time period Downtime losses within the facility; For work nodes Recovery time, which reflects the total time from a node's failure to its full recovery, satisfies: (twenty three).

[0140] In formula (23) For work nodes Maintenance time.

[0141] In a preferred embodiment of the present invention, the job assurance rate OS in the multi-objective evaluation model constructed in step S4 satisfies: (twenty four).

[0142] in, For the importance of node j, Let be the importance weight of node j. A higher importance weight indicates that the interruption of the node has a greater impact on the job availability rate. The length of the time window covered by the scheduling scheme.

[0143] Step S5: Use an improved particle swarm optimization algorithm to optimize and solve the collaborative scheduling scheme for marine maintenance resources.

[0144] Based on completing step S4, further implement step S5. It is worth noting that step S4 shows that the offshore maintenance resource collaborative scheduling scheme (optimization objective) should include at least two objective fitnesss: 1) minimizing total cost CS (including navigation fuel cost, fixed cost of node start-up, and loss cost due to power outage); 2) maximizing operational availability OS (measuring operational continuity).

[0145] Based on this, the optimization solution for the collaborative scheduling scheme of marine maintenance resources is further obtained through decision variables and constraints. The decision variables refer to a complete collaborative scheduling scheme of marine maintenance resources represented by two layers of discrete encoding: "resource allocation" and "path selection." The constraints mainly include: the total amount of resources dispatched by each maintenance resource node must not exceed its inventory capacity, and the total amount of resources processed by each transfer node must not exceed its carrying capacity.

[0146] In a preferred embodiment of the present invention, step S5 specifically includes the following steps:

[0147] Step S51: Construct a multi-objective optimization model with the objectives of minimizing total scheduling cost (CS) and maximizing job availability (OS).

[0148] The constructed multi-objective optimization model satisfies: (25).

[0149] in, The weighting coefficient for the total scheduling cost CS. The maximum total scheduling cost CS for all particles in the s-th iteration is given by [the value of CS]. The weighting coefficient for the job assurance rate (OS). The time-constrained penalty factor; For work nodes Maximum interruption tolerance time.

[0150] Step S52: Constrain resource capacity.

[0151] The constrained resource capacity satisfies the following: (26).

[0152] In equation (26), For maintenance resources The demand for the μth resource; x ij For execution variables; where, Indicates activation from maintenance resources To the work node The scheduling task for allocating resources. This indicates that the program will not be started.

[0153] Step S53: Constrain the capacity of the transfer nodes.

[0154] The constrained capacity of the transfer nodes satisfies: (27).

[0155] In equation (27), x ij These are execution variables based on parameters i and j.

[0156] Step S54: Perform two-level discrete coding on the decision variables of the collaborative scheduling scheme for offshore maintenance resources.

[0157] The first layer of discrete coding is used to solve the resource allocation problem, satisfying: (28); Indicates assignment to job nodes The maintenance resource number.

[0158] The second-layer discrete coding is used to solve the path selection problem, satisfying: (29); Indicates the work node The scheduling method used.

[0159] Step S55: Initialize and set the algorithm parameters of the multi-objective optimization model, and generate a random initial particle swarm.

[0160] Update the individual optimal pbest and the global optimal gbest, and update the particle velocity and position.

[0161] The process continues iteratively until the optimal solution for the decision variables of the collaborative scheduling scheme for offshore maintenance resources is obtained.

[0162] One point that needs further explanation is that step S54, which involves performing a two-level discrete encoding of the decision variables for the collaborative scheduling scheme of marine maintenance resources, can be understood as decomposing the decision variables into the following two levels: the first level is the resource allocation problem, specifically for each work node. Allocate a maintenance resource This determines "who goes." The second is the path selection (second-level) problem, specifically referring to the selection of each job node. Choosing which scheduling mode (direct or transshipment) solves the "how to get there" problem.

[0163] Then, in step S55, after the position update is completed, technicians select particles that may violate resource capacity constraints and transfer node capacity constraints to be processed using a penalty function method.

[0164] For detailed procedures, please refer to the following: Figure 2 As shown: Set the PSO algorithm parameters (such as population size, maximum number of iterations, inertia weight range, and learning factor), and randomly generate an initial particle swarm, where the position of each particle is composed of the aforementioned two-layer discrete encoding. Decode the scheduling scheme represented by each particle, calculate its total cost and job guarantee rate according to the aforementioned steps, and then obtain the fitness value F. Then, record the historical best position (pbest) of each particle and the global best position (gbest) of the entire population.

[0165] Then, for each particle and each layer of discrete decision variables, its continuous velocity value is updated according to the standard PSO formula. Next, the Softmax function is used to convert the velocity values ​​into a probability distribution of selecting various possible discrete values ​​(e.g., which resource to choose, which path to select). Then, random sampling is performed based on this probability distribution to generate a new generation of discrete positions.

[0166] Finally, the iterative calculation is repeated until the maximum number of iterations is reached or the fitness function converges, thus outputting the optimal solution for the decision variables of the collaborative scheduling scheme for offshore maintenance resources. The output scheme explicitly specifies which maintenance resource node will provide service for each faulty turbine, whether each scheduling task will be directly scheduled or transferred through a specific transit node, and estimates the arrival time, total cost, and overall system uptime for each task.

[0167] Further verification revealed that, based on the offshore maintenance resource collaborative scheduling optimization method considering the impact of wind and waves provided by this invention, the operation and maintenance center can quickly generate offshore maintenance resource collaborative scheduling schemes for multiple concurrent faults. This scheme can automatically avoid high-risk wind and wave paths, prioritize high-risk critical units, and balance loads by utilizing transfer nodes to avoid congestion and waiting. Compared with existing scheduling methods that rely on manual experience, this method can significantly improve resource utilization efficiency, shorten overall recovery time, and reduce operation and maintenance costs while ensuring safety.

[0168] This invention provides an optimization method for the collaborative scheduling of marine maintenance resources, taking into account the impact of wind and waves. The method includes the following steps: Step S1: Constructing a marine maintenance resource topology network and determining the matching rules for the topology network; Step S2: Calculating the dynamic path weights that incorporate the impact of wind and waves; Step S3: Quantifying and calculating the total collaborative scheduling time, including waiting time; Step S4: Constructing a multi-objective evaluation model; Step S5: Using an improved particle swarm optimization algorithm to optimize and solve the collaborative scheduling scheme for marine maintenance resources.

[0169] The collaborative scheduling optimization method for marine maintenance resources that takes into account the impact of wind and waves, and which features the above-mentioned steps, has at least the following technical advantages compared to existing technologies:

[0170] (1) The present invention provides a method for optimizing the collaborative scheduling of marine maintenance resources. By systematically quantifying the key environmental variable of wind and wave level into its impact on path safety and navigation efficiency, the generated scheduling scheme is more in line with the actual sea conditions, thereby improving the reliability and executability of the scheduling scheme. Furthermore, a topology network containing transfer nodes is constructed, and strict matching rules are defined, which expands the single "point-to-point" scheduling into a flexible "point-transfer-point" collaborative scheduling network, enriching the scheduling strategy and enabling feasible and economical alternatives to be found when resources are unavailable or direct costs are too high.

[0171] (2) The method for collaborative scheduling optimization of marine maintenance resources provided by the present invention truly reflects the queuing and congestion situation that maintenance resources may encounter at transfer nodes, avoids the disconnection of scheduling plans caused by underestimating waiting time, and significantly improves the accuracy of time planning. In addition, the designed dual-layer discrete coding PSO algorithm greatly compresses the solution space and can quickly generate high-quality scheduling schemes for large-scale complex scenarios.

[0172] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for optimizing the collaborative scheduling of offshore maintenance resources considering the impact of wind waves, characterized in that, The steps include the following: Step S1: Construct a marine maintenance resource topology network and determine the matching rules for the topology network; Step S2: Calculate the dynamic path weights that incorporate the effects of wind and waves; Step S3: Quantify and calculate the total collaborative scheduling time, including waiting time; Step S4: Construct a multi-objective evaluation model; Step S5: Use an improved particle swarm optimization algorithm to optimize and solve the collaborative scheduling scheme for offshore maintenance resources; The marine maintenance resource topology network constructed in step S1 satisfies: G=(V,E,W); where V represents the set of nodes, E represents the set of edges, and W represents the path weight. a set of nodes V, satisfying: ; Among them, V M For a set of maintenance resource nodes, satisfying: (1); V T Let be a set of transit nodes, satisfying: (2); V O Let the set of job nodes satisfy: (3); wherein, is the i-th repair node in the set of repair resource nodes; is the k-th transfer node in the set of transfer nodes; is the j-th job node in the set of job nodes; The edge set E is composed of a direct dispatch edge set and a transfer dispatch edge set ; And, nodes With nodes The distance between them satisfies: (4); When the transfer occurs, the distance between the two nodes satisfies: (5); where e is an edge in the edge set E; is an indicator function that takes the value 1 when the event is true and 0 when the event is false; i.e. the indicator function satisfies: (6). The path weight W in the step S2 is determined by the weight of the direct scheduling edge and the weight of the transit scheduling edge together; that is, the path weight W satisfies: (7) wherein the direct scheduling edge weight satisfies: (8); In equation (8), The weighting coefficients are determined by experts, prioritizing cost and the urgency of operational support. For work nodes The risk level weight, with a value ranging from 1 to 5; ω max The highest risk level among all work nodes is fixed at level 5; d max This represents the maximum distance between all scheduled paths, including direct edges and transmission edges. C i For maintenance resources The unit time cost satisfies: (9); In formula (9), f i is the fuel consumption per unit distance of the maintenance resource ; C f is the fuel unit price; v i,max is the maximum cruising speed of the maintenance resource . Path reliability coefficient , satisfies: ; its functional relationship with the wind wave level ξ satisfies: (10); In formula (10), is a critical sea state level at which the path reliability starts to decrease; is a decay coefficient of the path reliability as the sea state level decreases, to control the steepness of the reliability decrease. and the weight of the transit dispatch edge satisfies: (11)； In formula (11), C k is the cost per unit of time of the transit node T k ; is the one-time cost of the transit node T k ; and, combining equation (8) with equation (11), the maximum cost is satisfied: 。 2. The method of claim 1, wherein, The process of quantifying and calculating the total cooperative scheduling time, including the waiting time, in step S3 specifically includes the following steps: Step S31: quantifying the total coordinated scheduling time including latency constituent parts; Wherein the total coordinated scheduling time satisfies: (13)； Step S32: Calculate the sailing time And determine how the speed decreases as the wave level ξ increases. ; wherein the sailing time satisfies: (14); The speed decreases as the wind and wave level ξ increases , satisfying: (15); In formula (15), to repair resources the maximum speed, and the wind wave level ξ is negatively related; is the wind wave sensitivity coefficient; Step S33: Calculate preparation time; wherein the preparation time is made up of a repair preparation time and a transfer preparation time ; Step S34: Calculate the waiting time ; wherein the waiting time satisfies: (16)； In equation (16), As the basic unit of time; It is a growth factor used to control waiting time as the load rate increases. The rate of increase; This is the critical value for the load rate of the transfer node; For the load factor, the following conditions must be met: (17); In equation (17), μ is the maintenance resource category index; Q jμ For transit node T k The demand for resource type μ; For transmission node T k capacity; x ikj These are execution variables based on parameters i, k, and j.

3. The method of claim 1, wherein, the total scheduling cost in the multi-objective evaluation model constructed in the step S4 , satisfies: (18); wherein, is the direct dispatch cost, satisfying: (19); In equation (19), x ij These are execution variables based on parameters i and j; To transfer the dispatch cost, satisfy: (20)； In equation (20), f i f k Repair resources and transfer resources T k fuel consumption per unit distance; C f It is the unit price of fuel; To fix the cost, only at the beginning of the scheduled task occurs, satisfies: (21)； In formula (21), x ij is an execution variable based on parameters i, j; The downtime cost refers to the direct economic loss caused by the interruption of the job of the job node, and is positively correlated with the recovery time and the loss per unit time, satisfying: (22); In equation (22), For work nodes In time period Downtime losses within the facility; For work nodes Recovery time, which reflects the total time from a node's failure to its full recovery, satisfies: (twenty three); In formula (23) is the maintenance time of the job node is the maintenance time of the job node 4. The method of claim 1, wherein, The job guarantee rate in the multi-target evaluation model constructed in the step S4 , satisfies: (24); wherein, is the importance of node j, is the importance weight of node j, the higher the importance weight, the greater the impact of the interruption of the node on the job protection rate; is the length of the time window covered by the scheduling scheme.

5. The method of claim 1, wherein, Step S5 specifically includes the following steps: Step S51: Minimizing the total scheduling cost Minimizing the job protection rate Maximizing the objective, a multi-objective optimization model is constructed; The constructed multi-objective optimization model satisfies: (25)； in, Total scheduling cost The weighting coefficients, For the first The total scheduling cost of all particles in the next iteration Maximum value To ensure operational reliability The weighting coefficients, The time-constrained penalty factor; For work nodes Maximum interruption tolerance time; Step S52: Constrain resource capacity; wherein the constrained resource capacity satisfies: (26); In equation (26), For maintenance resources The demand for the μth resource; x ij For execution variables; where, Indicates activation from maintenance resources To the work node The scheduling task for allocating resources. Indicates that it will not start; Step S53: Constrain the capacity of the transfer nodes; The constrained capacity of the transfer nodes satisfies: (27)； In formula (27), x ij is an execution variable based on parameters i, j; Step S54: Perform two-level discrete coding on the decision variables of the collaborative scheduling scheme for offshore maintenance resources; The first layer of discrete coding is used to solve the resource allocation problem, satisfying: (28); Indicates assignment to job nodes Maintenance resource number; The second-layer discrete coding is used to solve the path selection problem, satisfying: (29); Indicates the work node The scheduling method used; Step S55: Initialize and set the algorithm parameters of the multi-objective optimization model, and generate a random initial particle swarm; Update the individual optimal pbest and the global optimal gbest, and update the particle velocity and position; The process continues iteratively until the optimal solution for the decision variables of the collaborative scheduling scheme for offshore maintenance resources is obtained.

Images