Topological optimization method for offshore wind farm power collection system considering intensive sea use constraints

Abstract

The application discloses a method for optimizing the topology of a power collection system of an offshore wind farm considering the constraint of intensive sea use. The application construes the laying constraint of a submarine cable into a broken line path along the direction of the row and the direction of the main path belt, generates a candidate topology by adopting a two-level grouping mode of non-terminal point fixed attribution and structured allocation of terminal points, verifies and eliminates infeasible schemes in combination with the cross-row physical constraint, and calculates the loop cost based on a submarine cable cross-section cost mapping table, so that a physically feasible topology meeting the requirement of parallel and non-intersecting submarine cables can be automatically generated under the constraint of intensive sea use, the step cost factor of the submarine cable is considered to realize the optimal construction cost, and the calculation complexity is controlled within an engineering acceptable range by structuring the search space.

Description

Technical Field

[0001] This invention relates to the technical field of power collection system topology optimization, and specifically to a method for optimizing the topology of offshore wind farm power collection systems considering intensive sea use constraints. Background Technology

[0002] Offshore wind farm power collection systems are used to collect the electrical energy generated by each wind turbine generator and transmit it to an offshore substation via submarine cables. The power collection system consists of multiple collection circuits, each of which connects several wind turbines in series to the offshore substation via submarine cables, thus realizing the collection and transmission of electrical energy.

[0003] Optimizing the current collection system involves optimizing the connection method and topology of the submarine cable to minimize construction costs. In offshore wind farm design, the number of wind turbines carried by different cable segments at different locations within a single current collection loop varies, resulting in different current transmission rates and therefore different cable cross-sectional specifications and costs per unit length. Optimizing cable costs cannot solely rely on reducing the total cable length; it must also consider the price differences between cables with different cross-sections.

[0004] In related technologies, the topology optimization scenario of offshore wind farm power collection systems under the constraint of intensive sea use presents a technical problem: the search space for feasible topology solutions is too large, and the evaluation cost of a single solution is too high. This leads to excessively high exhaustive computational complexity, making it impossible to complete topology optimization within an acceptable engineering timeframe while ensuring the physical feasibility of grouped solutions. This technical problem directly affects the computational efficiency and engineering feasibility of power collection system design.

[0005] Specifically, it includes the following three sub-problems:

[0006] 1. Under the constraints of intensive use of the sea, submarine cables can only be laid in a broken line along the row direction of the wind turbine array and the direction of the preset main path. How to establish a cost calculation model that can accurately reflect the actual length of the broken path, so as to take into account the stepped cost factors of submarine cables and achieve the optimal construction cost?

[0007] 2. Given a very large search space for grouping schemes, how can we reduce the complexity of candidate scheme generation and evaluation through structured decomposition and computational reuse, so that the exhaustive search strategy can be completed within an acceptable engineering timeframe?

[0008] 3. In cross-row connection groups, how to formally determine whether the grouping scheme meets the physical constraint of parallel and non-crossing cables under the condition of intensive marine use, so as to ensure the physical feasibility of the topology scheme. Summary of the Invention

[0009] The technical problem this invention aims to solve is that, in related technologies, under the constraint of intensive sea use, the topology optimization of offshore wind farm power collection systems suffers from excessively large search spaces for feasible topology schemes and high evaluation costs for individual schemes. This leads to excessively high exhaustive computational complexity, making it impossible to complete topology optimization within an acceptable engineering timeframe, while simultaneously ensuring the physical feasibility of grouping schemes. The purpose is to provide a topology optimization method for offshore wind farm power collection systems that considers the constraint of intensive sea use, solving the technical problems of excessively high exhaustive computational complexity, inability to complete topology optimization within an acceptable engineering timeframe, and ensuring the physical feasibility of grouping schemes.

[0010] This invention is achieved through the following technical solution:

[0011] In a first aspect, the present invention provides a topology optimization method for offshore wind farm power collection systems considering intensive sea use constraints, comprising:

[0012] Obtain the set of wind turbine coordinates, substation coordinates, maximum number of single-circuit loads, and cost mapping table for submarine cable cross-sections. The cost mapping table records the correspondence between current carrying capacity level and cost per unit length.

[0013] The wind turbine layout is identified based on the set of wind turbine coordinates, and a main path band connecting each row to the substation is constructed.

[0014] The terminal point is determined based on the relative position of each exhaust fan and the main path belt. The non-terminal point fans of each row are fixedly assigned to the row group, and the terminal points are assigned to the corresponding group according to the preset rules, generating a variety of group division schemes.

[0015] The wind turbines in each group are arranged into a linear sequence according to a preset sorting rule, and then continuously divided based on the linear sequence and the maximum number of loads per loop to generate multiple group schemes;

[0016] Feasibility verification was performed on cross-row groups in the group scheme based on preset cross-row physical constraints, and infeasible schemes were eliminated.

[0017] Based on the main path band, the cost of the loop is calculated for each group that has passed the verification by using a broken line path along the row direction and the main path band direction, and the cost of each physical segment is determined by combining the cost mapping table.

[0018] By summing up the circuit costs of each group, the total cost of the entire field is obtained from various group division schemes, and the optimal power collection system topology scheme with the best total cost is output.

[0019] Furthermore, the step of identifying the wind turbine layout based on the wind turbine coordinate set and constructing the main path band connecting each row to the substation includes:

[0020] Based on the longitudinal coordinate distribution of each fan, multiple fan rows are identified, and the fans in each row are sorted according to their lateral coordinates.

[0021] Based on the sorted wind turbine coordinates, the geometric center of adjacent wind turbines in each row is extracted as the candidate feature point of that row, and multiple candidate main paths are generated by combining the candidate feature points of each row.

[0022] Candidate main paths that meet the following conditions are selected as main path bands: the lateral offset between adjacent candidate feature points does not exceed a preset threshold, and the lateral offset between the last candidate feature point and the booster station does not exceed a preset threshold.

[0023] Furthermore, the steps of determining the terminal points based on the relative positions of each exhaust fan and the main path belt, fixing the non-terminal point fans of each row into the same row group, and assigning the terminal points into the corresponding groups according to preset rules to generate multiple group division schemes include:

[0024] For each row of fans, calculate the horizontal distance between each fan in the row and the characteristic point of the main path in the row, and determine the fan with the smallest horizontal distance as the terminal point of the row, and the remaining fans as the non-terminal point fans of the row.

[0025] The non-terminal point fans in each row are fixedly assigned to the corresponding basic group of that row;

[0026] The terminal points of each row are assigned to the corresponding basic group, other basic groups with numbers lower than the row number, or additional groups. Different assignment options correspond to different group division schemes.

[0027] Furthermore, the step of forming a linear sequence of wind turbines within each major group according to a preset sorting rule, and continuously dividing the group based on the linear sequence and the maximum load capacity of a single loop to generate multiple group schemes includes:

[0028] Arrange the fans within the large group in ascending order of row number as the primary sorting key and ascending order of horizontal coordinate as the secondary sorting key to form a linear sequence;

[0029] Based on the maximum number of loads on a single loop, the total number of wind turbines in a large group is decomposed into multiple small group sizes; each combination consists of multiple positive integers not exceeding the maximum load, and the sum of the positive integers is equal to the total number of wind turbines in the large group.

[0030] By performing full permutations for each group size combination, sequences of various lengths are obtained;

[0031] For each length sequence, a subsequence of wind turbines of the corresponding length is extracted from the linear sequence to generate a subgroup scheme for that large group.

[0032] Furthermore, the step of verifying the feasibility of cross-row groups in the group scheme based on preset cross-row physical constraints and eliminating infeasible schemes includes:

[0033] For cross-row groups of multi-row fans, the following checks are performed sequentially: first-row integrity constraint verification, intermediate terminal row single-point constraint verification, and intermediate empty row constraint verification. The first-row integrity constraint verification determines whether all fans in the first row of the cross-row group belong to that group. The intermediate terminal row single-point constraint verification determines whether each intermediate terminal row of the cross-row group contains only one fan. The intermediate empty row constraint verification determines whether the number of fans in the non-terminal rows between the first and last rows of the cross-row group is zero. The intermediate terminal row single-point constraint verification and the intermediate empty row constraint verification are performed based on the overall fan layout information.

[0034] Eliminate the group schemes corresponding to cross-row groups that do not meet any of the above constraint verifications.

[0035] Furthermore, the step of calculating the loop cost for each verified group based on the main path band using a broken line path along the row direction and the main path band direction, and determining the cost of each physical segment in conjunction with the cost mapping table, includes:

[0036] For two adjacent wind turbines within the group, the physical length of the cable connecting the two wind turbines is determined using the following path calculation methods, based on their relative positions in the row and whether they are terminal points:

[0037] If the two wind turbines are in the same row, the straight-line distance between them is used as the physical segment length. If the two wind turbines are both terminal points and are in different rows, the straight-line distance between them is used as the physical segment length. If at least one of the two wind turbines is not a terminal point and is in a different row, the physical segment length is the sum of the alignment segment length from one wind turbine to the feature point of the main path in that row, the main path segment length from the feature point of that row to the feature point of another row along the main path, and the alignment segment length from the other wind turbine to the feature point of the main path in its row. For the last wind turbine in the group, the physical segment length is the sum of the alignment segment length from the wind turbine to the feature point of the main path in its row, the main path segment length from the feature point of that row to the feature point of the last row along the main path, and the distance from the feature point of the last row to the substation.

[0038] Based on the length of each physical segment and the number of wind turbines it supports, the cost per unit length is obtained by querying the cost mapping table. The cost per unit length is then multiplied by the length of the corresponding physical segment to obtain the cost of each physical segment. Finally, the costs of each physical segment are summed to obtain the cost of the group loop.

[0039] Furthermore, before the step of summarizing the costs of each sub-group loop to obtain the total cost of the entire site for various large-group division schemes, the method further includes:

[0040] After sorting the wind turbines within each major group by row number and horizontal coordinate, the sequence of wind turbine numbers included in that major group is extracted as the unique signature of that major group.

[0041] Iterate through all large group partitioning schemes, perform continuous partitioning, feasibility verification and loop cost calculation only once for large groups with the same unique signature, and store the obtained small group schemes and corresponding small group loop costs associated with the unique signature.

[0042] For each group division scheme, the cost of the small loops corresponding to each of the major groups included in the scheme is obtained by querying the associated storage, and the obtained small loop costs are summarized to obtain the total cost of the entire site for the group division scheme.

[0043] Furthermore, the step of summarizing the costs of each group of loops to obtain the total cost of the entire field for various large group division schemes, and outputting the optimal power collection system topology scheme in terms of total cost, includes:

[0044] The total cost of each scheme is ranked, and the scheme with the lowest total cost is selected as the first candidate scheme.

[0045] If the difference between the total cost of multiple schemes and the first candidate scheme is less than a preset ratio threshold, then the spatial aggregation degree index and the intra-row continuity index are calculated for each of the multiple schemes. Based on the spatial aggregation degree index and the intra-row continuity index, the multiple schemes are sorted a second time, and the optimal scheme after the second sorting is taken as the final output power collection system topology scheme.

[0046] Furthermore, the step of outputting the optimal power collection system topology scheme in terms of total cost also includes:

[0047] Lightweight calculations are performed for each main path zone scheme to obtain the total cost of the entire site for each main path zone scheme; wherein, the lightweight calculations include generating multiple large group division schemes, generating multiple small group schemes, performing feasibility verification and calculating loop cost, and do not perform secondary sorting based on spatial aggregation degree index and intra-row continuity index;

[0048] The target main path scheme is determined based on the total cost of the entire site corresponding to each main path scheme.

[0049] Perform a full calculation for the target main path scheme to output the final collector system topology scheme.

[0050] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0051] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0052] This invention constrains submarine cable laying into a broken-line path along the row direction and the main path strip direction by constructing a main path strip. It generates candidate topologies using a two-level grouping method of fixed non-terminal point assignment and structured terminal point allocation. Combined with cross-row physical constraints, it verifies and eliminates infeasible schemes. Based on the submarine cable cross-section cost mapping table, it calculates the loop cost. Thus, it can automatically generate physically feasible topologies that meet the requirements of parallel and non-crossing cables under the constraint of intensive sea use, taking into account the stepped cost factors of submarine cables to achieve optimal construction cost. At the same time, the structured search space controls the computational complexity within an acceptable range for engineering. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0054] Figure 1 A flowchart illustrating a topology optimization method for an offshore wind farm collection system considering intensive sea use constraints, provided in the embodiments of this specification;

[0055] Figure 2 This specification provides a system architecture diagram for a topology optimization method for offshore wind farm power collection systems that considers intensive sea use constraints, as illustrated in the embodiments herein.

[0056] Figure 3 This is a schematic diagram of a submarine cable layout method that does not consider intensive use of the sea, as provided in the embodiments of this specification.

[0057] Figure 4 This is a schematic diagram illustrating a submarine cable arrangement method that considers efficient use of the sea, as provided in the embodiments of this specification.

[0058] Figure 5 This is a block diagram of an electronic device provided in the embodiments of this specification. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0060] In offshore wind farm power collection systems, the electricity generated by multiple wind turbines is typically collected via on-site collection cables to an offshore substation, and then transmitted to the onshore power grid through a transmission system. To optimize the topology of the collection system, it is often necessary to obtain the spatial coordinates of the wind turbines, the location of the substation, the allowable load capacity per circuit, and the cost information of different specifications of submarine cables. Based on the arrangement of the wind turbine array, the cable connection relationships between the wind turbines and between the wind turbines and the substation must be determined. In engineering implementation, wind turbines in offshore wind farms are often arranged in regular or near-regular arrays, with relatively clear row and column directions between different turbine rows. Correspondingly, the laying of collection cables must not only meet electrical connection requirements but also constraints related to construction routing, sea area occupation, and subsequent operation and maintenance.

[0061] In related technologies, topology optimization of offshore wind farm collection systems is often treated as a multi-constraint combinatorial optimization problem. For a given number of wind turbines, it is necessary to simultaneously determine the turbine grouping method, the connection order within a group, the cross-row travel method, and the convergence path to the substation. Combined with the selection of submarine cable specifications and length calculations, the cost of candidate topology schemes is evaluated. Without imposing additional spatial structural constraints, related technologies often treat this problem as a general network construction or combinatorial optimization problem, generating candidate schemes through exhaustive search, heuristic search, clustering, minimum spanning tree variants, or hybrid optimization algorithms. These methods can achieve certain results when the number of wind turbines is small or the constraints are weak. However, as the scale of wind turbines increases, the number of candidate connection relationships increases, and spatial constraints strengthen, the number of candidate topology schemes grows rapidly, leading to a significant expansion of the search space.

[0062] However, the aforementioned algorithmic optimization methods face another efficiency constraint in practical engineering applications. In existing engineering practices, topology optimization of power collection systems typically employs a sequential process of "detailed scheme - cost - comparison": for each candidate grouping scheme, a detailed layout design, including all engineering details such as line spacing, safety distances, and bend locations, must be completed before the cost of the scheme can be calculated and compared. Since detailed layout itself is a time-consuming engineering design task, the evaluation cost of a single candidate scheme is extremely high. Engineers can usually only evaluate a few schemes based on experience and cannot systematically traverse the search space. The root of this problem lies in the fact that existing practices couple "grouping topology decision" (i.e., which wind turbines are grouped together) with "detailed path design" (i.e., the specific routing and spacing of cables within a group) into the same evaluation process. In reality, the main factor determining the cost differences between different grouping schemes is the length and cross-sectional grade distribution of the submarine cables (i.e., the stepped cost effect), rather than detailed design parameters such as line spacing and the number of bends. Moreover, the latter is a detailed optimization problem within the same grouping scheme and has little impact on the cost ranking between different grouping schemes.

[0063] Especially in intensive offshore wind farm scenarios, submarine cables cannot be laid arbitrarily between wind turbines; instead, they must be arranged along the array direction of the turbines and the direction of a pre-defined main corridor. This constraint transforms the feasible topology from an arbitrary network connection in the general sense into a structured feasible solution space constrained by both the array structure and the corridor structure. On the one hand, the number of candidate solutions that meet the requirements of corridor-based laying, cable non-crossing, and single-loop mounting restrictions remains large. On the other hand, many heuristic strategies commonly used in free-layout scenarios cannot effectively utilize the internal patterns of feasible solutions in this structured feasible solution space, making it difficult to achieve a balance between search efficiency and result quality. For large-scale offshore wind farms, directly exhaustively enumerating all candidate groups, connections, and routing paths results in high computational complexity, making it difficult to complete full-field topology optimization within an acceptable timeframe. If a general heuristic algorithm is used instead, insufficient utilization of the structured constraints may lead to low solution efficiency or search results deviating from the optimal solution.

[0064] Further analysis reveals that the root cause of the aforementioned technical problems lies not simply in the large number of wind turbines, but in the fact that under the constraint of intensive marine use, the generation, verification, and cost calculation of candidate schemes are simultaneously affected by multiple coupled factors, and the existing single-scheme evaluation mode after detailed design evaluation is incompatible with the requirements of the structured search space. First, whether a wind turbine can be grouped into the same loop depends not only on the capacity limit but also on its arrangement position in the array and whether physical routing constraints are met when connecting across rows. Second, even if several candidate grouping schemes can be formed, the connection order of wind turbines within different schemes, the cross-row connection method, and the main path access method will affect the length of the submarine cable segment and its corresponding cost. Third, between different large grouping schemes, there are often a large number of locally repetitive wind turbine combinations and repetitive intermediate calculation processes. If each candidate scheme is still independently divided, its feasibility verified, and its cost calculated, it will result in a large amount of redundant calculation, further increasing the overall solution time. It is evident that the main shortcomings of the relevant technologies are: a structured computational method that matches the constraints of intensive sea use has not yet been formed, and the reusable patterns inherent in the wind turbine array layout, main path corridors, and repeating substructures cannot be used to compress the search space and reduce the solution complexity. At the same time, the group topology decision-making and detailed path design have not been decoupled, resulting in a large number of candidate solutions not being able to be quickly evaluated and systematically compared.

[0065] For the reasons mentioned above, in the related technologies, in the scenario of topology optimization for offshore wind farm collection systems under the constraint of intensive sea use, there is a technical problem: the search space for feasible topology schemes is too large and the evaluation cost of a single scheme is too high, resulting in excessive exhaustive computational complexity and making it impossible to complete topology optimization within an acceptable engineering timeframe while ensuring the physical feasibility of grouped schemes. This technical problem directly affects the computational efficiency and engineering feasibility of collection system scheme design. If the search complexity cannot be reduced and the evaluation efficiency of a single scheme cannot be improved while ensuring the accuracy of feasibility and cost evaluation, it will be difficult to fully compare multiple candidate collection topologies within the actual engineering design cycle, and it will also be difficult to obtain a superior topology scheme that is adapted to the constraint of intensive sea use.

[0066] The inventive concept of this invention is to transform the topology optimization process of the power collection system under the constraint of intensive sea use into a structured solution process based on wind turbine layout, main path band, terminal point, large group division, small group continuous splitting, cross-row feasibility verification, and loop cost reuse calculation, thereby reducing the solution complexity of feasible topology schemes and solving the technical problem that the exhaustive calculation complexity is too high under the constraint of intensive sea use, and the topology optimization cannot be completed within the acceptable time of the project.

[0067] like Figure 1 and Figure 2 As shown in the figure, this embodiment provides a topology optimization method for offshore wind farm power collection systems considering intensive sea use constraints. This method can be executed by a computing device. The computing device can be a local server, an engineering design terminal, a cluster computing node, or other devices with data processing capabilities.

[0068] The method may include:

[0069] Step S11: Obtain the set of wind turbine coordinates, substation coordinates, maximum number of single-circuit loads, and cost mapping table for submarine cable cross-sections. The cost mapping table records the correspondence between current carrying capacity level and cost per unit length.

[0070] In this embodiment, the acquisition action can be performed by a computing device receiving, reading, extracting, calling, or loading basic input data related to topology optimization. Specifically, the basic input data can come from a wind farm design database, a geographic information system, engineering design documents, manually entered documents, interface transmission results, or historical engineering data. The acquisition can be completed either all at once or in batches.

[0071] In this embodiment, the wind turbine coordinate set is used to characterize the spatial location of each wind turbine within the wind farm. Each element in the wind turbine coordinate set corresponds to the location information of one wind turbine. The location information can be represented using Cartesian coordinates or using latitude and longitude coordinates after projection transformation. To facilitate subsequent identification and sorting according to the wind turbine arrangement, two-dimensional plane coordinates under a unified coordinate system can be used. In addition to the location itself, the wind turbine coordinate set can also include additional information such as wind turbine number, area identifier, equipment capacity, and turbine parameters. However, in this embodiment, it is necessary to at least be able to distinguish the spatial location and identifier of different wind turbines.

[0072] In this embodiment, the booster station coordinates are used to characterize the location of the offshore booster station in the same coordinate system. If the wind farm has multiple collection points or multiple booster stations, the method of this application can be executed separately for each collection point, or the wind turbines can be divided into multiple sub-regions according to the coverage area of ​​the booster station and then processed separately.

[0073] In this embodiment, the maximum number of loads per single circuit is used to limit the maximum number of wind turbines that can be connected to a single collector circuit. This parameter can be determined by electrical design constraints, submarine cable capacity limitations, protection configuration requirements, or engineering experience rules. In this embodiment, the maximum number of loads per single circuit can be input into the calculation device as a pre-set integer parameter.

[0074] In this embodiment, the submarine cable cross-section cost mapping table is used to establish the correspondence between cable carrying capacity and cost per unit length. Specifically, the current carrying capacity level in the mapping table is used to distinguish the current carrying capacity or load-bearing capacity range corresponding to different submarine cable specifications. The cost per unit length is used to characterize the cost of the corresponding submarine cable per unit length. The mapping table can be stored in the form of a table structure, key-value pair structure, database record, or parameter configuration file. It is used to support the computing device to determine the applicable submarine cable specification level and the corresponding cost per unit length based on the number of wind turbines that a certain cable segment needs to support. Specifically, it can be built directly based on the submarine cable cross-section or indirectly based on the current carrying capacity level. It can include only material procurement costs, or it can further include the converted costs of construction, laying, auxiliary materials, etc.

[0075] Step S12: Identify the wind turbine layout based on the wind turbine coordinate set, and construct the main path band connecting each row with the substation.

[0076] In this embodiment, the action of identifying the wind turbine layout can be to determine the arrangement relationship of the wind turbines in the wind farm based on the spatial distribution characteristics of each wind turbine in the wind turbine coordinate set. Specifically, it can include regular array identification, approximately regular array identification, row identification with perturbed coordinates, and local row identification after partitioning.

[0077] In one possible and specific implementation, the computing device can identify multiple wind turbine rows based on the distribution of their longitudinal coordinates. It is understood that the longitudinal and lateral directions are two relative coordinate directions defined relative to the main direction of the current wind farm array, and are not required to be consistent with the absolute direction on a map. Specifically, the computing device can first determine the array's main axis direction based on the overall distribution of the wind turbine coordinates, then use the direction parallel to the main axis as the row direction, and the direction perpendicular to the main axis as the lateral direction. After identifying each row, the computing device can sort the wind turbines within each row according to their lateral coordinates. This sorting establishes the order of the wind turbines within the row, providing a basis for subsequent extraction of candidate feature points, determination of terminal points, and generation of linear sequences. These candidate feature points characterize the representative positions of the main path zone within the corresponding wind turbine row. They can be spatial reference points associated with the row and used to connect the main path corridor. Specifically, they can be the geometric center of the line connecting adjacent wind turbines within the row, the projection point on the centerline of the wind turbines within the row, the corresponding position of the wind turbine in the middle of the row, or a smoothed corridor sampling point.

[0078] In one possible and specific implementation, the computing device can extract the geometric centers of adjacent wind turbines within each row as candidate feature points for that row, based on the sorted wind turbine coordinates. For each row, there may be multiple pairs of adjacent wind turbines, and therefore multiple candidate feature points. The computing device then selects one candidate feature point from each row and combines them to form multiple candidate main paths. Each candidate main path represents a set of corridor reference lines that spans each wind turbine row and extends towards the substation.

[0079] It is understandable that the main path zone is not merely a mathematically defined single line, but rather a path region or set of path centerlines used to constrain the longitudinal convergence direction of submarine cables. It can be a structured corridor traversing multiple wind turbine rows and connecting to a substation, supporting unified organization of cross-row connections and convergence paths. Specifically, it can be a broken line, a strip-shaped area composed of a set of adjacent broken lines, a central path and its tolerance zone, or a trunk path obtained by connecting characteristic points of each row to the substation. In this embodiment, candidate main paths that meet the screening criteria can be considered as the main path zone.

[0080] Step S13: Determine the terminal point based on the relative position of each row of fans and the main path belt, fix the non-terminal point fans of each row into the row group, and assign the terminal points into the corresponding group according to the preset rules, generating multiple group division schemes.

[0081] In this embodiment, the terminal point can be the fan in each row that has the most direct connection to the main path strip and is most suitable as the interface for that row to access the main path strip. It can be a representative fan in that row that performs cross-row connection or corridor access functions. Specifically, the terminal point can be the fan with the smallest horizontal distance from the main path strip, the fan closest to the characteristic point of that row, or a preferred fan among several candidate fans that meet the constraints. In this embodiment, the fan in each row with the smallest horizontal distance from the corresponding characteristic point of the main path strip in that row can be determined as the terminal point of that row.

[0082] In this embodiment, the non-terminal point fan can be any fan within the same row other than the terminal point. The "large group" of this row can be a basic group directly corresponding to a certain row, used to carry the fans that are fixedly assigned to that row. This large group can be an intermediate organizational unit for subsequent continuous subdivision and group generation; it can be a candidate set at a higher level than a group. Specifically, it can be a large group formed by a single row of fans, a large group formed by merging terminal points from multiple rows, or a composite large group composed of fans within a row and terminal points across rows.

[0083] Step S14: Form a linear sequence of wind turbines in each group according to a preset sorting rule, and continuously divide the group based on the linear sequence and the maximum number of loads per loop to generate multiple group schemes.

[0084] In this embodiment, the preset sorting rule can be used to convert a set of wind turbines in a large group that was originally in a two-dimensional space into a one-dimensional ordered sequence. Specifically, the wind turbines in the large group can be sorted using ascending row number as the primary sorting key and ascending horizontal coordinate as the secondary sorting key.

[0085] In this embodiment, the continuous segmentation can be achieved by dividing the wind turbines into several continuous subsequences in a linear sequence, with each continuous subsequence corresponding to a group, and the number of wind turbines in any group not exceeding the maximum load capacity of a single loop. That is, by maintaining sequence continuity, it can group spatially adjacent or nearby wind turbines into the same group. Specifically, it can be equal-length segmentation, unequal-length segmentation, segmentation based on combined enumeration, or segmentation based on constraints, etc.

[0086] In one possible and specific implementation, the computing device can first generate multiple group size combinations based on the maximum number of loads per single loop and the total number of wind turbines within the group. Each group size combination can consist of multiple positive integers, where each positive integer does not exceed the maximum number of loads per single loop, and the sum of the positive integers equals the total number of wind turbines within the group.

[0087] Then, the computing device can arrange the group size combinations in different orders to obtain multiple length sequences. For each length sequence, a continuous wind turbine subsequence of the corresponding length can be extracted sequentially along the aforementioned linear sequence. Each continuous wind turbine subsequence constitutes a group, and multiple groups together constitute a group scheme for the same large group. By traversing different group size combinations and their permutations, multiple group schemes can be generated for the same large group.

[0088] It should be noted that the aforementioned groups can be direct connection units ultimately used for loop cost calculation and topology output. They can be a group of wind turbines that meet the load limit and are sequentially connected. Specifically, they can be groups consisting only of wind turbines in the same row, or cross-row groups including several wind turbines from multiple rows. The group schemes of different large groups are combined to form the set of field-wide group schemes under the corresponding large group division scheme.

[0089] Step S15: Perform feasibility verification on the cross-row groups in the group scheme based on the preset cross-row physical constraints, and eliminate infeasible schemes.

[0090] In this embodiment, the cross-row group can be a group that includes multiple fans in a row. It is understood that, compared to an intra-row group that only includes fans in the same row, a cross-row group needs to address the physical feasibility of the path during the cross-row connection process.

[0091] In this embodiment, the preset cross-row physical constraints can be used to determine whether the cross-row group can form a non-conflicting cable connection path under the main path band constraints. Specifically, it can include the first row integrity constraint, the intermediate terminal row single-point constraint, and the intermediate blank row constraint.

[0092] In one possible and specific implementation, the computing device may perform this verification step only for cross-row groups that include multiple rows of fans, while for groups that only include fans in the same row, they may be directly considered to satisfy the cross-row physical constraints. For each cross-row group, the computing device may sequentially perform the first row integrity constraint verification, the intermediate terminal row single-point constraint verification, and the intermediate blank row constraint verification.

[0093] Specifically, the first-row integrity constraint verification is used to determine whether all the wind turbines in the first row involved in the cross-row group belong to the cross-row group. The purpose is to ensure that when the first row is the starting position of the cross-row connection, there will be no wiring conflicts caused by some wind turbines in the same row being split to other circuits.

[0094] The single-point constraint verification of the intermediate terminal row can be used to determine whether the number of wind turbines in the intermediate terminal rows involved in the cross-row group is always one. More specifically, the intermediate terminal row can be understood as the wind turbine row located between the first and last rows and playing a transitional access role in the cross-row connection link. Constraining this row to retain only one terminal point helps maintain the singularity of the cross-row path and avoids path intersections caused by multiple lateral connections within the same intermediate row.

[0095] The intermediate empty row constraint verification can be used to determine whether the number of fans in the non-terminal rows between the first and last rows is zero. This non-terminal row can be an intermediate fan row that does not serve as a terminal access point in the current cross-row connection structure. Requiring its fan count to be zero is to avoid local detours and intersections when the cross-row connection passes through these fan rows.

[0096] In one possible and specific implementation, single-point constraint verification of intermediate terminal rows and constraint verification of intermediate empty rows can be performed based on the wind turbine layout information of the entire region. That is, when the computing device determines whether a certain cross-row group is feasible, it not only examines the wind turbines contained in the group itself, but also examines the layout of the wind turbine rows that the group crosses in the entire region and the potential coupling relationships with other groups. In this way, global path conflicts can be missed by judging based solely on local group information.

[0097] Step S16: Based on the main path band, calculate the loop cost for each verified group using a broken line path along the row direction and the main path band direction, and determine the cost of each physical segment in conjunction with the cost mapping table.

[0098] In this embodiment, the circuit cost can be the total cost corresponding to the formation of a current collection circuit by the various wind turbines in a certain group according to a determined connection sequence. Specifically, it can be the cumulative cost determined by the cable length of each adjacent connection segment in the group and the cable specifications corresponding to each segment.

[0099] Specifically, the cost of the loop may include only the cost of the submarine cable materials, or it may include costs such as joints, construction, installation, and maintenance. In this embodiment, it may at least include the cable cost determined based on the cable length and submarine cable cross-section cost mapping table.

[0100] In this embodiment, the polygonal path can be used to characterize the permissible path form when a submarine cable connects from one wind turbine to another under the constraint of intensive sea use. It is not the shortest straight line between any two points, but rather a constrained path formed by a combination of path segments along the row direction and path segments along the main path direction. Specifically, it can include straight segments within the same row, direct connection segments between terminal points across rows, and polygonal segments formed by combining alignment segments and main path segments.

[0101] Step S17: Summarize the circuit costs of each group to obtain the total cost of the entire field for various large group division schemes, and output the topology scheme of the collector system with the optimal total cost.

[0102] In this embodiment, the total construction cost can be the sum of the loop construction costs of all feasible subgroups corresponding to a certain large group division scheme. Specifically, it can be the sum of simple construction costs, or it can be a comprehensive cost value formed by adding other evaluation factors.

[0103] In one possible and specific implementation, the computing device can read the cost of all small loops under each large group partitioning scheme, add them up, and obtain the corresponding total cost. Then, the computing device can compare the total costs corresponding to multiple large group partitioning schemes, select the scheme with the lowest total cost as the current optimal candidate scheme, and output the corresponding collector system topology scheme.

[0104] In this embodiment, the output may be: outputting the wind turbine grouping results in the optimal solution; outputting the wind turbine connection sequence in each group; outputting the length and corresponding cost of each physical segment; outputting path data for drawing or further engineering design; or writing the results into a database, generating a report file, or displaying them on an interface, etc.

[0105] In one possible and specific implementation, if the total cost of multiple schemes is similar, the computing device can further introduce auxiliary evaluation indicators for secondary ranking based on the total cost. For example, spatial aggregation index and intra-row continuity index can be calculated separately, and the scheme with more concentrated spatial distribution and better intra-row continuity can be selected as the final output among multiple schemes with similar total costs.

[0106] In one possible and specific implementation, to reduce redundant calculations, the computing device can also identify the same large groups of wind turbines in different large group division schemes before summarizing the total cost. For large groups with the same wind turbine number sequence, the results of continuous segmentation, feasibility verification, and loop cost calculation can be reused. To this end, a unique signature can be generated first based on the wind turbine number sequence sorted by row number and horizontal coordinate within the large group. Then, the sub-group schemes and sub-group loop cost results corresponding to the large groups with the same unique signature are associated and stored. Afterward, when calculating the total cost for each large group division scheme, the corresponding results are directly extracted from the associated storage and summarized.

[0107] In one possible and specific implementation, the computing device can divide the wind field into four computing sub-regions with the main path zone and the horizontal line of the booster station as the boundary, and perform mirror transformation on the wind turbine coordinates and feature points of different sub-regions to map them to the shape of the reference area. Then, the above steps are performed on each sub-region. Finally, the results of each sub-region are summarized to obtain the optimal solution for the entire field.

[0108] In some implementations, the step of identifying the wind turbine layout based on the wind turbine coordinate set and constructing the main path band connecting each row to the substation includes:

[0109] Step S121: Based on the longitudinal coordinate distribution of each fan, identify multiple fan rows and sort the fans in each row according to their lateral coordinates.

[0110] In this embodiment, the vertical coordinate can be the coordinate component of the wind turbine along the column direction of the wind turbine arrangement in the wind farm plane coordinate system, which can correspond to the column direction of the wind turbine arrangement in the wind farm design. The horizontal coordinate can be the coordinate component in the direction orthogonal to the vertical coordinate, which can correspond to the row direction of the wind turbine arrangement.

[0111] In this embodiment, the wind turbine layout identification can be achieved by dividing all the wind turbines in the field into several groups according to the similarity of their longitudinal coordinates, with each group corresponding to a row of wind turbines.

[0112] Specifically, the identification process can employ a density-based clustering algorithm. More specifically, a vertical coordinate distance threshold can be set as the cluster neighborhood radius. This threshold value is related to the row spacing in the wind farm design; for example, it can be between 900 and 1500 meters when the row spacing is 3000 meters. The clustering algorithm traverses each wind turbine, searching for all turbines whose vertical coordinate differences are within the neighborhood radius, centered on the turbine's vertical coordinate value, and assigning these turbines to the same cluster. Turbines not assigned to any existing cluster serve as the starting point for new clusters, continuing the expansion until all turbines in the entire field are assigned to their corresponding clusters. Each cluster corresponds to a row of turbines. To further improve the accuracy of layout identification, the lateral coordinate distribution range of turbines within each cluster can be checked after clustering. If the lateral coordinate distribution range of turbines in a certain cluster is significantly larger than that of other rows, it is possible that the row actually contains two adjacent sub-rows with similar vertical coordinates due to construction deviations. In this case, the neighborhood radius threshold can be lowered to further refine the sub-row.

[0113] Step S122: Based on the sorted wind turbine coordinates, extract the geometric center of adjacent wind turbines in each row as candidate feature points for that row, and generate multiple candidate main paths by combining the candidate feature points of each row.

[0114] In this embodiment, for a sequence of wind turbines in a sorted row, the adjacent wind turbines can be two wind turbines that are immediately next to each other in that row. The geometric center can be the arithmetic mean of the coordinates of the two adjacent wind turbines, specifically the coordinate point formed by the average of the horizontal coordinates and the average of the vertical coordinates of the two wind turbines. The physical significance of the geometric center lies in its location at the midpoint of the line connecting the two wind turbines, representing the gap between adjacent wind turbines in the row. The submarine cable enters the main path corridor from this gap, which avoids interference with the wind turbine foundations and ensures a relatively balanced connection distance between the wind turbines on both sides.

[0115] In this embodiment, the candidate feature points can be reference points used to characterize the possible locations the main path band may pass through in that row. For a row including multiple wind turbines, since there are multiple pairs of adjacent wind turbines, the row can correspond to multiple candidate feature points. The candidate main path can be a broken line formed by selecting one candidate feature point from each row and connecting them sequentially according to the order of the rows. This broken line starts from the first row, passes through the selected feature points of each row in sequence, and finally extends towards the booster station, forming a possible longitudinal corridor centerline.

[0116] In one possible implementation, candidate feature points can be selected not only from the geometric centers of adjacent wind turbines, but also from the coordinates of a wind turbine closer to the substation, other internal points along the line connecting adjacent wind turbines, or the point corresponding to the median of the lateral coordinates of all wind turbines in the row. In another possible implementation, if there are many wind turbines in a row and they are densely distributed, the geometric centers of some adjacent wind turbine pairs can be extracted to reduce the number of candidate feature points.

[0117] In this embodiment, when the computing device generates candidate main paths, it selects a point from the candidate feature points in the first row, a point from the candidate feature points in the second row, and so on, until a point is selected from the last row. The selected feature points in each row are connected in order of row number to form a candidate main path. This process iterates through all possible selection combinations to generate multiple candidate main paths. For a wind farm with multiple rows of wind turbines, if each row has several candidate feature points, the total number of candidate main paths is the product of the number of candidate feature points in each row.

[0118] Step S123: Select candidate main paths that meet the following conditions as main path bands: the lateral offset between adjacent candidate feature points does not exceed a preset threshold, and the lateral offset between the last candidate feature point and the booster station does not exceed a preset threshold.

[0119] In this embodiment, the lateral offset can be the absolute value of the difference between two adjacent rows of candidate feature points in the lateral coordinate direction. This value reflects the lateral turning range of the main path band between two adjacent rows.

[0120] In this embodiment, the preset threshold is a value used to limit the upper limit of the lateral turning angle of the main path strip, to prevent the main path strip from having excessively severe lateral bending, thereby ensuring the engineering feasibility of the main path corridor.

[0121] In this embodiment, the lateral offset between the last row of candidate feature points and the booster station can be the absolute value of the difference between the last row of candidate feature points and the booster station in the lateral coordinate direction, which is used to reflect the lateral alignment relationship between the end of the main path belt and the booster station.

[0122] In one possible implementation, the preset threshold can be set as a multiple of the turbine spacing, for example, 2 to 3 times the turbine spacing. In another possible implementation, the preset threshold can also be determined based on the maximum lateral turning radius allowed by the submarine cable laying process. In yet another possible implementation, the preset threshold can be directly set as a fixed value based on historical engineering experience.

[0123] In this embodiment, when the computing device performs the screening, it calculates the lateral offset between adjacent rows of candidate feature points for each candidate main path. If the lateral offset between any two adjacent rows of candidate feature points exceeds a preset threshold, the candidate main path is eliminated. Then, the lateral offset between the last row of candidate feature points and the booster station is calculated. If this lateral offset exceeds a preset threshold, the candidate main path is also eliminated. After the above two rounds of screening, the remaining candidate main paths are determined as the main path bands. In one specific implementation, if multiple main path bands remain after screening, all of them are retained for subsequent optimization calculations; if only one main path band remains, it is directly used as the sole main path band. In another implementation, the computing device can also further sort the screened main path bands according to the total path length or path smoothness, and select a portion of the main path bands for subsequent optimization processes.

[0124] In some implementations, the steps of determining terminal points based on the relative positions of each exhaust fan and the main path belt, fixing non-terminal point fans in each row to their respective large groups, and assigning terminal points to their corresponding large groups according to preset rules to generate multiple large group division schemes include:

[0125] Step S131: For each row of fans, calculate the horizontal distance between each fan in the row and the characteristic point of the main path in the row, and determine the fan with the smallest horizontal distance as the terminal point of the row, and the remaining fans as the non-terminal point fans of the row.

[0126] In one possible implementation, the computing device reads the lateral coordinates of all fans in each row and the lateral coordinates of the feature points along the main path in that row. It then calculates the absolute value of the difference between the lateral coordinate of each fan and the feature point to obtain the horizontal distance between each fan and the feature point. By comparing the horizontal distances of all fans in the row, the fan with the smallest value is identified as the terminal point of that row. The remaining fans in the row are considered non-terminal point fans. In another possible implementation, if multiple fans have the same minimum horizontal distance to the feature point, the fan with the smaller lateral coordinate, the fan with the larger lateral coordinate, or the fan closer to the substation can be selected as the terminal point, and the remaining fans are considered non-terminal point fans.

[0127] Step S132: Fixedly assign the non-terminal point fans of each row to the corresponding basic group of the row.

[0128] In this embodiment, the basic group can be a group consisting of a certain row as the main body, and the group number corresponds to the row number. Specifically, the row number is an ordinal number assigned to each row of wind turbines in the entire field in descending or ascending order of the longitudinal coordinate, used to identify the relative position of each row in the wind field. For example, the row farthest from the booster station can be numbered as the first row, and the numbers can be increased sequentially towards the booster station. Alternatively, the row closest to the booster station can be numbered as the first row, and the numbers can be increased sequentially towards the direction away from the booster station. Regardless of the numbering direction, the row numbers only need to remain unique and ordered throughout the entire field.

[0129] Understandably, non-terminal point wind turbines in each row can only be assigned to the corresponding basic group within that row; there are no other options. The physical constraint of this rule stems from the requirements of intensive marine cable laying. Specifically, the cables within each row must be laid along the row's arrangement direction and ultimately converge at the row's terminal point. Non-terminal point wind turbines cannot bypass the terminal point of their row to directly connect to the main path segment of other rows; therefore, their assigned group must be consistent with the group to which the row's terminal point belongs. Once non-terminal point wind turbines are fixedly assigned to the basic group of their row, their group assignment no longer changes with variations in the group division scheme.

[0130] Step S133: Assign the terminal points of each row to the basic group corresponding to this row, other basic groups with numbers lower than this row number, or additional groups, where different assignment options correspond to different group division schemes.

[0131] In this embodiment, the additional major group can be a major group independent of all basic major groups, which only includes the terminal points assigned to that group and does not include any non-terminal point wind turbines. In one embodiment, the additional major group is an independent grouping within the aforementioned corresponding major groups. The other basic major groups with numbers lower than the row number can be basic major groups with row numbers lower than the current row number. For example, the terminal point of the 3rd row can be assigned to the basic major group corresponding to the 1st or 2nd row.

[0132] In one possible implementation, the computing device can process the terminal points of each row sequentially from row 1 to row J. For the terminal point in row j, its allocation position includes: the basic major group corresponding to this row, any other basic major group with a number less than j, and an additional major group. Therefore, there are j+1 allocation options for the terminal point in row j. The computing device generates all possible allocation combinations through a recursive or traversal algorithm. For a wind farm with J rows of wind turbines, theoretically, multiple major group partitioning schemes can be generated.

[0133] In another possible implementation, the computing device can also generate the major group partitioning scheme by recursively determining the allocation of the terminal points in the first row, then the allocation of the terminal points in the second row, and so on, up to the Jth row. The allocation choices for each row are independent of each other, and the Cartesian product of the allocation results of all rows constitutes the set of all major group partitioning schemes.

[0134] In another possible implementation, the generated grouping scheme corresponds to a set of large groups, which includes multiple basic large groups and possible additional large groups. Each basic large group includes the non-terminal turbines that are fixedly assigned to that row, as well as the terminal turbines of other rows assigned to that row. Additional large groups include all the terminal turbines of each row assigned to that group. The difference between the different grouping schemes lies in which large group each row of terminal turbines is assigned to; the assignment of non-terminal turbines remains constant.

[0135] In some implementations, the step of forming a linear sequence of wind turbines within each group according to a preset sorting rule, and then continuously dividing the group based on the linear sequence and the maximum load capacity of a single loop to generate multiple group schemes includes:

[0136] Step S141: Arrange the fans in the large group in ascending order of row number as the primary sorting key and ascending order of horizontal coordinate as the secondary sorting key to form a linear sequence.

[0137] In this embodiment, the row number can be the ordinal number of the row determined in the aforementioned steps. It is understood that using the row number as the primary sorting criterion serves the purpose that the main path extends from the row furthest from the substation to the row closest to the substation, and electrical energy converges along the main path from the furthest row to the closest row. After arranging in ascending order by row number, the sequence of fans within the group corresponds to the physical direction of electrical energy flow along the main path; that is, the first part of the sequence corresponds to the upstream fans furthest from the substation, and the second part corresponds to the downstream fans closest to the substation.

[0138] In this embodiment, the ascending order of the lateral coordinates can be achieved by arranging the fans in ascending order of their lateral coordinate values ​​within the same row. The direction of the lateral coordinates is consistent with the laying direction of the cables within the row. Within the same row, after the fans are arranged in ascending order of their lateral coordinates, adjacent fans in the sequence are also physically adjacent within the row. The cables within the row start from the fan with the smallest lateral coordinate (i.e., the one at the beginning of the sequence) and proceed sequentially towards the one with the largest lateral coordinate (i.e., the one at the end of the sequence).

[0139] Understandably, based on the above double-bond sorting, the arrangement order of the wind turbines within the large group in the resulting linear sequence completely corresponds to the actual cable laying path and the direction of power collection: first, they are arranged from farthest to nearest row according to row number, and then within each row, they are arranged according to the horizontal coordinate from the end furthest from the terminal point to the end closest to the terminal point. This linear sequence provides a physically constrained sorting basis for subsequent continuous segmentation.

[0140] In one possible implementation, if a group comprises three fans in the first row, two fans in the second row, and one fan in the third row, with the horizontal coordinates of the fans in the first row being 1000, 3000, and 5000 respectively, the horizontal coordinates of the fans in the second row being 2000 and 4000 respectively, and the horizontal coordinate of the fans in the third row being 3000, the resulting linear sequence after arranging them in ascending order by row number and horizontal coordinate is as follows: fan in the first row with horizontal coordinate 1000, fan in the first row with horizontal coordinate 3000, fan in the first row with horizontal coordinate 5000, fan in the second row with horizontal coordinate 2000, fan in the second row with horizontal coordinate 4000, and fan in the third row with horizontal coordinate 3000. In this sequence, fans in the same row remain continuous and ordered by horizontal coordinate, while fans in different rows are ordered by row number.

[0141] Step S142: Based on the maximum number of loads on a single loop, decompose the total number of wind turbines in the large group into multiple small group size combinations; wherein each combination consists of multiple positive integers not exceeding the maximum number of loads, and the sum of the positive integers is equal to the total number of wind turbines in the large group.

[0142] In this embodiment, the group size combination can be a method of dividing the total number of wind turbines in a large group into several groups, each containing a number of wind turbines. Each positive integer represents the size of a group, i.e., the number of wind turbines included in that group. The maximum number of wind turbines that can be mounted on a single loop can be the upper limit of the number of wind turbines allowed in a group; therefore, each positive integer obtained from the decomposition cannot exceed this upper limit.

[0143] In one possible implementation, the computing device first determines the range of the number of groups. The lower limit of the number of groups is determined by dividing the total number of turbines in the larger group by the maximum load capacity per loop and rounding up, ensuring that all turbines can be accommodated even if each group reaches its maximum capacity. The upper limit of the number of groups is the total number of turbines in the larger group, i.e., the extreme case where each group includes only one turbine. In another possible implementation, the computing device may also directly limit the range of the number of groups based on engineering experience, for example, excluding group sizes that include only one or two turbines.

[0144] After determining the range of possible group sizes, the computing device decomposes the total number of wind turbines within each group into a sum of positive integers for each possible group size, ensuring that each positive integer does not exceed the maximum load capacity of a single loop. To avoid duplication, a decreasing constraint can be used during decomposition, meaning that each subsequent positive integer is no greater than the preceding positive integer, thus generating unique group size combinations. In another possible implementation, the computing device can also use a recursive algorithm or a dynamic programming algorithm to generate all positive integer decomposition results that satisfy the constraints.

[0145] Step S143: Perform full permutations for each group size combination to obtain sequences of various lengths.

[0146] In this embodiment, the length sequence can be a sequence obtained by arranging the positive integers in a combination of group sizes in different orders. Each element in the length sequence corresponds to a group size, and the length of the length sequence is equal to the number of groups. The permutation can be any possible order of all positive integers in a combination of group sizes, with each different permutation corresponding to a different length sequence.

[0147] In one possible implementation, the computing device can perform full permutations of the positive integers included in each group size combination generated in step S142. For example, for a group size combination consisting of three positive integers, the full permutation can yield six length sequences. Each length sequence represents a different group size permutation order, corresponding to a different access center distribution. In another possible implementation, if there are positive integers with the same value in the group size combination, the computing device can remove duplicate length sequences generated by swapping the positions of identical values ​​when generating the full permutation, retaining only the unique permutation result.

[0148] Step S144: For each length sequence, extract a continuous wind turbine subsequence of the corresponding length along the linear sequence to generate a subgroup scheme for the large group.

[0149] In this embodiment, the continuous wind turbine subsequence can be a subset consisting of adjacent wind turbines in a linear sequence, where the wind turbine indices in the sequence are continuous and uninterrupted. The grouping scheme can be a set of several groups formed by dividing a large group, with each group corresponding to a continuous wind turbine subsequence.

[0150] In one possible implementation, the computing device, for each length sequence, starts from the beginning of the linear sequence and, based on the value of the first element in the length sequence, extracts a number of continuous wind turbines to form the first group. Then, based on the value of the second element in the length sequence, it extracts continuous wind turbines starting from the position immediately following the first group to form the second group. This process continues until all continuous wind turbine subsequences corresponding to all elements in the length sequence have been extracted. Thus, all wind turbines within a large group are divided into several groups, where the number of wind turbines in each group is equal to the element value at the corresponding position in the length sequence, and the wind turbines in all groups are continuously distributed in the linear sequence, do not overlap, and are sequentially connected.

[0151] In another possible implementation, if the starting position of the linear sequence corresponds to the fan furthest from the substation, then the access order of the subgroups obtained by truncating the length sequence from front to back corresponds to the confluence order from the far end towards the substation. If the starting position of the linear sequence corresponds to the fan closest to the substation, then the truncating order corresponds to the confluence order from the near end to the far end. The computing device can select the truncating direction according to actual engineering requirements and maintain consistency in direction across all subgroups.

[0152] By traversing all group size combinations generated in step S142 and performing the full permutation in step S143 and the linear truncation in step S144 for each combination, the computing device can generate multiple group schemes for the same large group. Each group scheme corresponds to a different wind turbine grouping method and access center distribution. In one implementation, the computing device retains all generated group schemes for subsequent feasibility verification and cost calculation. In another implementation, the computing device can also pre-screen the group schemes according to preset engineering rules, eliminating schemes containing isolated wind turbines or excessively small groups before proceeding to subsequent steps.

[0153] In some implementations, the step of verifying the feasibility of cross-row groups in the group scheme based on preset cross-row physical constraints and eliminating infeasible schemes includes:

[0154] Step S151: For a multi-row fan group, sequentially perform the first row integrity constraint verification, the intermediate terminal row single-point constraint verification, and the intermediate empty row constraint verification; wherein, the first row integrity constraint verification is used to determine whether all fans in the first row involved in the cross-row group belong to the cross-row group; the intermediate terminal row single-point constraint verification is used to determine whether the number of fans in the intermediate terminal rows involved in the cross-row group is one; the intermediate empty row constraint verification is used to determine whether the number of fans in the non-terminal rows between the first and last rows involved in the cross-row group is zero; wherein, the intermediate terminal row single-point constraint verification and the intermediate empty row constraint verification are performed based on the fan layout information of the entire area.

[0155] In this embodiment, the cross-row group can be a group of fans that includes two or more fan rows. In contrast, an intra-row group is a group that includes only fans from the same fan row. Since the cable connections of an intra-row group do not involve the longitudinal path of the cross-row, intra-row groups do not need to undergo cross-row physical constraint verification. In one embodiment, after generating the group scheme, the computing device first determines whether the fans included in the group involve multiple fan rows. If multiple fan rows are involved, the group is determined to be a cross-row group, and the verification process of this step is triggered. If only a single fan row is involved, this step is skipped or the loop cost calculation step is directly entered.

[0156] In this embodiment, the first row can be the row with the smallest numerical value among the multiple row numbers involved in the cross-row group, that is, the row where the cable path of the group begins to connect to the main path band. The last row can be the row with the largest numerical value among the multiple row numbers involved in the cross-row group, that is, the row where the cable path of the group finally leaves the main path band or continues to extend downstream. The intermediate terminal row can be the row among the multiple row numbers involved in the cross-row group that is between the first row and the last row and performs the function of a terminal point in the group. The non-terminal row can be the row among the multiple row numbers involved in the cross-row group that is between the first row and the last row but does not perform the function of a terminal point in the group.

[0157] In one possible implementation, when the computing device performs the first-row integrity constraint verification, it can first extract the row numbers of all wind turbines involved in the cross-row group and determine the row with the smallest row number as the first row. Then, the computing device can obtain the set of all wind turbines in the first row across the entire region, that is, all wind turbines in the row in the entire wind farm, regardless of which large group or subgroup they belong to. Next, the computing device can check one by one whether all wind turbines in the first row are included in the current cross-row group. If any wind turbine in the first row does not belong to the cross-row group, it is determined that the group does not satisfy the first-row integrity constraint.

[0158] Understandably, as the starting point for cross-row connections, the cables within the first row must be laid continuously from the furthest wind turbine to the terminal point. If a wind turbine in the first row is assigned to another group, the lateral outgoing cable of that turbine will cross the cables laid along the row direction of that group, violating the requirement for parallel and non-crossing cables under intensive marine use conditions.

[0159] In one possible implementation, when the computing device performs single-point constraint verification on intermediate terminal rows, it can first determine the row numbers involved in the cross-row group, excluding the first and last rows. Within these intermediate rows, it further identifies which rows contain the terminal point of the group and defines these rows as intermediate terminal rows. For each intermediate terminal row, the computing device can, based on the wind turbine layout information for the entire region, count the total number of wind turbines in that row across the entire wind farm. If the total number of wind turbines in that row is exactly one, the constraint is considered satisfied. If it is more than one, the constraint is considered not satisfied.

[0160] It is understandable that the intermediate terminal row only serves as a longitudinal junction transfer node on the main path, and there should be no other wind turbines in its row except for the terminal point. Otherwise, the lateral outgoing cables of other wind turbines will spatially cross with the cables laid longitudinally along the main path by this group.

[0161] In one possible implementation, when the computing device performs the intermediate blank line constraint verification, it can first determine all row numbers between the first and last rows of the cross-row group. Among these row numbers, the first row, last row, and intermediate terminal rows are excluded; the remaining row numbers are the non-terminal rows. For each non-terminal row, the computing device, based on the wind turbine layout information for the entire region, calculates the total number of wind turbines in that row across the entire wind farm. If the total number of wind turbines in that row is zero, the constraint is considered satisfied. If any wind turbine exists, the constraint is considered not satisfied.

[0162] In this embodiment, the wind turbine layout information for the entire area can be wind turbine distribution data covering all wind turbine rows in the entire wind farm. Specifically, it can be a pre-constructed global data structure that records the wind turbine composition of each wind turbine row in all major group division schemes. Alternatively, it can be the distribution results obtained by querying the affiliation relationships of each wind turbine row in real time during the verification process.

[0163] In this embodiment, the execution based on the wind turbine layout information of the entire area can be represented as the judgment object of the single-point constraint verification of the intermediate terminal row and the intermediate empty row constraint verification. It is not limited to the wind turbines within the cross-row group currently being verified, but needs to examine the existence of wind turbines in the entire wind farm range involving the row.

[0164] In one possible implementation, the computing device pre-establishes row distribution information for the entire region before performing this step. This information can be indexed by row number, recording which groups occupy each row under the current grouping scheme and the set of fan numbers for each row. When performing single-point constraint verification of intermediate terminal rows and intermediate empty row constraint verification, the computing device can directly query this pre-established information to obtain the number of fans in the corresponding row for judgment. In another possible implementation, the computing device can also traverse all groups in real time during the verification process to count the total number of fans in the target row.

[0165] Step S152: Eliminate the group schemes corresponding to cross-row groups that do not meet any of the above constraint verifications.

[0166] In this embodiment, the elimination can be expressed as removing infeasible group schemes from the candidate set for subsequent cost calculations and scheme aggregation. In one possible implementation, if a group scheme within a large group includes multiple cross-row groups, the entire group scheme is eliminated if any one of these cross-row groups does not satisfy any of the aforementioned constraints. In another possible implementation, if all group schemes within a large group are eliminated, then the large group has no feasible groupings under the current large group partitioning scheme. The computing device can mark the large group as infeasible and exclude it or assign it an infinite cost value when subsequently aggregating the total cost. In yet another possible implementation, the computing device can also perform verification immediately after generating cross-row groups. If the verification fails, the subsequent generation and calculation of that group are directly abandoned to reduce unnecessary calculations.

[0167] In some implementations, the step of calculating the loop cost for each verified group based on the main path band using a broken line path along the row direction and the main path band direction, and determining the cost of each physical segment in conjunction with a cost mapping table, includes:

[0168] Step S161: For two adjacent wind turbines within the group, determine the physical length of the cable connecting the two wind turbines based on their row positions and whether they are terminal points, using the following path calculation methods: If the two wind turbines are in the same row, use the straight-line distance between them as the physical length; if both wind turbines are terminal points and located in different rows, use the straight-line distance between them as the physical length; if at least one of the two wind turbines is not a terminal point and is located in a different row, use the distance between the two wind turbines... The physical segment length is the sum of the alignment section length from one fan to the feature point of the main path in that row, the main path segment length from the feature point of that row to the feature point of another row along the main path, and the alignment section length from another fan to the feature point of the main path in its row. For the last fan in the group, the physical segment length is the sum of the alignment section length from the fan to the feature point of the main path in its row, the main path segment length from the feature point of that row to the feature point of the last row along the main path, and the distance from the feature point of the last row to the substation.

[0169] In this embodiment, the circuit cost can be the total construction cost corresponding to all wind turbines in a group forming a complete collector circuit according to a determined electrical connection sequence. This circuit starts with the wind turbine furthest from the substation, connects each wind turbine in series, and finally converges into the substation.

[0170] In this embodiment, the broken path can be a constrained path that the submarine cable travels from one connection point to the next under the constraint of intensive use of the sea. It can be composed of a combination of a transverse path segment laid along the row direction and a longitudinal path segment laid along the main path strip direction, rather than a free straight line between any two points.

[0171] In this embodiment, the physical segment can be a continuous cable segment between two adjacent connection points in the circuit, with each cable segment carrying a specific number of wind turbine loads, corresponding to a specific current carrying capacity level.

[0172] In one possible implementation, when calculating the physical segment length between two adjacent wind turbines within a group, the computing device first reads the row number and terminal point identification information of the two wind turbines. If the two wind turbines have the same row number, they are determined to be in the same row. In this case, the computing device can directly calculate the straight-line distance between the coordinates of the two wind turbines as the physical segment length. This straight-line distance corresponds to the actual length of the cable laid parallel to the wind turbine array within the row, and is a length measurement of the segment within the row.

[0173] If the two wind turbines are located in different rows, the computing device can further determine whether both turbines are terminal points. If both are terminal points, the computing device can directly calculate the straight-line distance between the coordinates of the two turbines as the length of the physical segment between them. This straight-line distance corresponds to the length of the cable laid along the main path corridor between the two terminal points. Since the terminal points are adjacent to the feature points of the main path, the connecting cable between the two terminal points is actually laid within the main path corridor, without needing to go through the detour of aligning into the main path and then aligning back from the main path. Therefore, a straight-line distance can be used instead of a broken-line distance, avoiding a systematic overestimation of the cost of cross-row connections.

[0174] If the two fans are located in different rows, and at least one of them is a non-terminal point, the computing device can use a broken-line path to calculate the length of the physical segment between them. The alignment segment length can be the distance that the non-terminal fan moves laterally from its location to a feature point on the main path of that row, reflecting the lateral alignment length of the cable within the row from the fan's location to the main path access point. The main path segment length can be the distance that extends longitudinally along the main path from a feature point in one row to a feature point in another row, reflecting the longitudinal laying length of the cable across different rows within the main path corridor. Specifically, the computing device first calculates the alignment segment length from one fan to a feature point on the main path of its row, then calculates the main path segment length along the main path from the feature point in that row, passing through the feature points of each intermediate row to a feature point in another row, and finally calculates the alignment segment length from the other fan to a feature point on the main path of its row. The three lengths are then added together to obtain the total length of the physical segment.

[0175] In one possible implementation, if only one of two adjacent wind turbines is a non-terminal point, then only the alignment section length corresponding to that non-terminal point is calculated, and the alignment section length corresponding to the other terminal point is approximately zero. If both turbines are non-terminal points, then the alignment section lengths corresponding to both turbines are calculated separately and included in the physical section length.

[0176] For the last wind turbine in a group, the computing device calculates the length of its final physical segment to the substation. This final physical segment can be the connecting cable from the last wind turbine in the group to the substation, carrying the total load of all wind turbines in the group. The computing device first calculates the alignment segment length from the last wind turbine to the characteristic point of its row's main path, then calculates the length of the main path segment along the main path from the characteristic point of that row, passing through the characteristic points of each downstream row to the characteristic point of the last row, and finally calculates the straight-line distance from the characteristic point of the last row to the substation. The three lengths are then added together to obtain the final physical segment length. In one implementation, if the row containing the last wind turbine is also the last row, the main path segment length is zero, and the final physical segment length consists only of the alignment segment length and the distance from the characteristic point of the last row to the substation.

[0177] Step S162: Based on the length of each physical segment and the number of wind turbines carried by that physical segment, query the cost mapping table to obtain the corresponding unit length cost, multiply the unit length cost by the corresponding physical segment length to obtain the cost of each physical segment, and sum the costs of each physical segment to obtain the cost of the group loop.

[0178] In this embodiment, the number of wind turbines carried by the physical segment can be the number of wind turbines counted from the location of the physical segment towards the end of the loop, that is, the total wind turbine load that the cable segment needs to transmit.

[0179] In this embodiment, the current carrying capacity rating can be a cable specification rating determined based on the number of wind turbines carried by the physical segment. The more wind turbines a segment carries, the higher the current carrying capacity rating, the larger the required cable cross-section, and the higher the cost per unit length. In one embodiment, the calculation device starts counting segments from the last wind turbine in the loop towards the substation: the physical segment between the last wind turbine and its adjacent upstream turbine carries only the load of the last wind turbine, corresponding to the first current carrying capacity rating. Moving upstream a segment, the physical segment carries the load of two wind turbines, corresponding to the second current carrying capacity rating. This continues until the last physical segment carries the load of all the wind turbines in the group, corresponding to the highest current carrying capacity rating.

[0180] In one possible implementation, the computing device, for each physical segment within the group, first determines the number of wind turbines carried by that segment, i.e., its current carrying capacity level. Then, it looks up the unit length cost corresponding to that current carrying capacity level in the submarine cable cross-section cost mapping table. Next, it multiplies this unit length cost by the actual length of the physical segment to obtain the cost of that segment. This process is repeated for all physical segments within the group, including segments between adjacent wind turbines and the final segment from the last wind turbine to the substation. After calculating the cost of each segment, the costs of all physical segments are summed to obtain the loop cost for the group.

[0181] In another possible implementation, if the cost mapping table is stored in discrete tiers, i.e., only recording the unit cost corresponding to integer current carrying capacity tiers, and the number of wind turbines carried by a certain physical section happens to correspond to a certain tier in the table, then the value can be directly retrieved from the table. If the table does not directly record the cost corresponding to the number of turbines carried, then the unit length cost can be determined by rounding up to the nearest tier, linear interpolation, or piecewise constant extension. In yet another possible implementation, the unit length cost in the cost mapping table may only reflect the cost of submarine cable materials, or it may further include the comprehensive unit cost after conversion of laying construction, joint fabrication, and auxiliary material consumption, but the cost caliber of each tier within the same mapping table remains consistent.

[0182] In some embodiments, prior to the step of summarizing the costs of each subgroup loop to obtain the total cost of the site for various large group division schemes, the method further includes:

[0183] Step A: After sorting the wind turbines in each group by row number and horizontal coordinate, extract the wind turbine number sequence included in the group as the unique signature of the group.

[0184] In this embodiment, the unique signature can be an identifier that uniquely identifies the composition of wind turbines within a large group. Since the composition of a large group depends only on which wind turbines are included in the group, and is independent of the order in which these wind turbines are arranged within the group, a deterministic identifier can be generated by extracting the number sequence of the wind turbines within the large group after sorting them according to a uniform rule.

[0185] In one possible implementation, the computing device can first sort all the wind turbines within the large group by row number in ascending order, and then sort the turbines with the same row number by horizontal coordinate in ascending order. After sorting, the computing device extracts the turbine numbers from each location sequentially and arranges them in the extraction order to form a number sequence. This number sequence is the unique signature of the large group.

[0186] In another possible implementation, the computing device may further convert the sorted number sequence into string form, array form, or hash value form for storage, as long as the conversion method can maintain the distinguishability between different constituent groups.

[0187] In another possible implementation, if there are many fans in a large group, the computing device can also extract a compressed representation of the number sequence after sorting. For example, specific delimiters can be used to connect the numbers to form an identifier string, as long as the identifier string can uniquely restore the original number sequence.

[0188] Step B: Traverse all large group partitioning schemes, perform continuous partitioning, feasibility verification, and loop cost calculation only once for large groups with the same unique signature, and store the resulting small group schemes and corresponding small group loop costs associated with the unique signature.

[0189] In this embodiment, the associated storage can be a data storage structure that establishes a mapping relationship between unique signatures and corresponding calculation results. Specifically, it can be a key-value pair mapping table in memory, an index array, an associated list, or a record table in an external database. The large group with the same unique signature can be a large group consisting of the exact same set of wind turbine numbers in different large group division schemes. Since step S132 fixes the non-terminal point wind turbines of each row into their respective large group, while step S133 only changes the allocation of terminal points, a large number of large groups with completely identical compositions may be generated between different large group division schemes. For example, when the allocation scheme of the terminal points in the 3rd row changes, the large groups composed of non-terminal points in the 4th row and subsequent rows may remain completely unchanged, thus repeating in different schemes.

[0190] In one possible implementation, the computing device can first generate all large group partitioning schemes, or generate each scheme one by one in a recursive order. For each large group in the current large group partitioning scheme, the computing device can generate its unique signature as described in step A. Then, the computing device checks whether the unique signature already exists in the associated storage. If it does not exist, it indicates that the large group is appearing for the first time. The computing device performs continuous partitioning of the large group to generate subgroup schemes, performs feasibility verification on cross-row subgroups, calculates the loop cost of each subgroup that passes verification using a broken line path, and associates the optimal subgroup scheme and its subgroup loop cost with the unique signature in the associated storage. If it already exists, it means that the composition of the large group is exactly the same as the large groups that have been processed before, and its continuous partitioning results, feasibility verification results, and loop cost results have all been calculated. The computing device directly skips the repeated calculation of the large group and does not perform the continuous partitioning, feasibility verification, and loop cost calculation steps again.

[0191] In another possible implementation, the computing device can also employ a delayed computation strategy. Specifically, it first generates all major group partitioning schemes and records the major group signatures contained in each scheme. Then, it counts the frequency of occurrence of each signature, performs a complete continuous partitioning, feasibility verification, and loop cost calculation only on the unique signature that has appeared once, and finally stores the calculation results in association. In yet another possible implementation, the loop cost of the subgroup stored in the associated storage can be the cost of the most cost-effective subgroup among multiple subgroup schemes, or it can be the cost set of all feasible subgroup schemes for the major group, so that it can be selected as needed during subsequent queries.

[0192] Step C: For each group division scheme, retrieve the cost of the small loops corresponding to each major group included in the scheme by querying the associated storage, and summarize the retrieved small loop costs to obtain the total cost of the entire site for the group division scheme.

[0193] In this embodiment, the query association storage can be a process of retrieving the corresponding calculation result in the association storage based on the unique signature of the large group.

[0194] In one possible implementation, the computing device can, for a given grouping scheme, sequentially extract the unique signatures of each major group within that scheme, and use each unique signature as a retrieval key to query the corresponding subgroup loop cost in the associated storage. Since the associated storage has only performed calculations once for major groups with the same unique signature, the query results are always valid, calculated cost data, regardless of whether the major group appears for the first time in the current grouping scheme. The computing device can then sum the subgroup loop costs corresponding to all major groups under that scheme to obtain the total cost for the entire site.

[0195] In another possible implementation, if the records of a large group in the associated storage indicate that there is no feasible subgroup scheme for that large group, the computing device can mark the large group partitioning scheme as an infeasible scheme, assign it a total cost of infinity, or directly remove it from the candidate schemes.

[0196] In another possible implementation, after obtaining the cost of each major group of loops, the computing device can further accumulate the cost details of the physical segments within each group to form a complete cost composition record of the major group division scheme, which can be used for subsequent scheme comparison and engineering analysis.

[0197] In some implementations, the step of summarizing the costs of each group of loops to obtain the total cost of the entire field for various large group division schemes, and outputting the optimal collector system topology scheme in terms of total cost, includes:

[0198] Step S171: Sort the total cost of each scheme and select the scheme with the lowest total cost as the first candidate scheme.

[0199] In this embodiment, the total construction cost can be the sum of the loop construction costs of all groups corresponding to a certain group division scheme. This total construction cost has been obtained in step C by querying the associated storage and summarizing the loop construction costs of the major groups.

[0200] In this embodiment, the first candidate scheme can be the grouping scheme that ranks first after sorting, i.e. has the lowest total cost.

[0201] In one possible implementation, after obtaining the total cost of each of the major group division schemes, the computing device can use a numerical comparison algorithm or a sorting algorithm to arrange the schemes in ascending order of total cost and extract the scheme at the first position in the sorting result as the first candidate scheme.

[0202] In another possible implementation, if two or more schemes have the same total cost and are both the lowest, the computing device can identify all of these schemes as the first candidate schemes and proceed to step S172 for secondary evaluation.

[0203] Step S172: If the difference between the total cost of multiple schemes and the first candidate scheme is less than a preset ratio threshold, then calculate the spatial aggregation degree index and the intra-row continuity index for each of the multiple schemes. Based on the spatial aggregation degree index and the intra-row continuity index, sort the multiple schemes a second time, and take the optimal scheme after the second sorting as the final output power collection system topology scheme.

[0204] In this embodiment, the preset ratio threshold can be a criterion for determining whether the total cost of multiple schemes is sufficiently close. Specifically, this threshold can be set according to the engineering precision requirements, for example, it can be set to one percent, two percent, or a certain fixed ratio coefficient of the total cost of the first candidate scheme. When the ratio obtained by dividing the difference between the total cost of other schemes and the total cost of the first candidate scheme by the total cost of the first candidate scheme is less than the threshold, it is determined that the scheme and the first candidate scheme belong to the same close tier in terms of cost.

[0205] In one possible implementation, the preset ratio threshold can be pre-set by the engineering designer based on the project cost sensitivity and input into the computing device.

[0206] In another possible implementation, the computing device may also automatically determine the threshold based on the statistical characteristics of the distribution of the total cost of the entire project, for example, by taking a certain multiple of the standard deviation of the total cost of the entire project.

[0207] In this embodiment, the spatial aggregation index is used to measure the compactness of the spatial distribution of wind turbines within each group in a scheme. Specifically, this index reflects the relative relationship between the average distance between wind turbines within a group and the average nearest neighbor distance of all wind turbines in the field.

[0208] In one possible implementation, the computing device calculates the pairwise Euclidean distances between all wind turbines within each group in a given scheme, and takes the average as the group's average distance. Then, it calculates the Euclidean distances between each wind turbine and its nearest neighbor for the entire site, and takes the average as the site-wide average nearest neighbor distance. Finally, it compares the group's average distance with the site-wide average nearest neighbor distance multiplied by a normalization coefficient to obtain the group's spatial aggregation index. A smaller index value indicates a more compact spatial distribution of wind turbines within the group, shorter cable runs, and greater convenience for construction and maintenance.

[0209] In another possible implementation, spatial aggregation indexes can also be other spatial compactness measures such as the maximum span of the wind turbines within the group, the area of ​​the enclosing box, or the centroid dispersion.

[0210] In this embodiment, the intra-row continuity index can be used to measure whether the fans in the same row within each group are topologically continuous in a scheme. This index reflects whether the fans in the same row within a group are adjacent in the complete sorting sequence of the row, and whether there are any interruptions caused by the insertion of fans from other groups.

[0211] In one possible implementation, for each group in a scheme, the computing device can extract a subset of the fans in each row for that group. After sorting the fans in this subset according to their horizontal coordinates, it checks whether the index difference between two adjacent fans in the complete sorted sequence within the original row is one. If the index difference is one, the adjacent fan pair is determined to be a consecutive pair. If the index difference is greater than one, it is determined that there are fans from other groups inserted in between, and the adjacent fan pair is a non-consecutive pair. The number of consecutive pairs in the row for that group is divided by the total number of fan pairs in the row for that group to obtain the intra-row continuity index of that group in that row. The intra-row continuity index of all groups in all rows is averaged to obtain the intra-row continuity index of the scheme. The larger the index value, the more continuous the fan grouping in the same row, the more orderly the cable routing, and the more beneficial it is to reducing the risk of construction intersections.

[0212] In another possible implementation, the continuity index within a row can also be determined using a Boolean decision method, that is, it is determined to be continuous only when all the fans in a row of a group are continuously distributed, otherwise it is determined to be discontinuous.

[0213] In one possible implementation, when the computing device performs secondary sorting, it first filters out all schemes whose total cost difference from the first candidate scheme is less than a preset percentage threshold, forming a set of schemes to be evaluated in the second round. Then, the computing device calculates the spatial aggregation degree index and the intra-scheme continuity index for each scheme in this set. Next, the computing device sorts the set of schemes to be evaluated in the second round based on these two indices.

[0214] In one implementation, the spatial aggregation degree index and the intra-outcome continuity index can be weighted and combined into a comprehensive score. The solutions are then sorted from low to high according to the comprehensive score, and the solution with the lowest score is selected as the optimal solution after secondary sorting.

[0215] In another implementation, the schemes can be sorted first by spatial aggregation degree index, and then sorted by intra-row continuity index among the schemes with the same or similar spatial aggregation degree index.

[0216] In another implementation, if only one solution satisfies the condition that the difference between it and the first candidate solution is less than a preset ratio threshold, then that solution is directly output as the optimal solution after secondary sorting.

[0217] In this embodiment, the final output power collection system topology scheme includes the wind turbine composition of each group under the optimal large group division scheme, the connection sequence of the wind turbines in each group, the length of each physical segment and its corresponding current carrying capacity level and cost per unit length, as well as the total cost of the entire site under this scheme.

[0218] In one possible implementation, the computing device writes the topology scheme into an engineering design database in the form of structured data, or outputs it in the form of a report document, or displays it in the form of a graphical topology diagram on the engineering design terminal interface, for engineers to further review and use in subsequent detailed design.

[0219] Furthermore, the step of outputting the optimal power collection system topology scheme in terms of total cost also includes:

[0220] Step S173: Perform lightweight calculations for each main path zone scheme to obtain the total cost of the entire site corresponding to each main path zone scheme; wherein, the lightweight calculations include generating multiple large group division schemes, generating multiple small group schemes, performing feasibility verification and calculating loop cost, and do not perform secondary sorting based on spatial aggregation degree index and intra-row continuity index.

[0221] In this embodiment, the main path zone scheme can be different main path zones constructed and screened according to the aforementioned steps. For any main path zone scheme, the executing entity can generate multiple large group division schemes, generate multiple small group schemes, perform feasibility verification and loop cost calculation based on the position of each row of feature points corresponding to the main path zone scheme, and compare the total cost of the entire field corresponding to each feasible scheme under the main path zone scheme to obtain the total cost of the entire field corresponding to the main path zone scheme.

[0222] In this embodiment, the lightweight computation can be used to quickly compare the costs of different main path zone schemes. Specifically, lightweight computation may include the generation of large group partitioning schemes, small group schemes, feasibility verification, and loop cost calculation processes corresponding to the main path zone schemes, but secondary sorting based on spatial aggregation degree index and intra-row continuity index is not performed at this stage. That is, the lightweight computation stage is mainly used to obtain the comparison results of each main path zone scheme in terms of cost, without further spatial quality optimization for schemes with similar total costs.

[0223] In one possible and specific implementation, for each main path zone scheme, the executing entity can traverse multiple large group division schemes under the main path zone scheme and summarize the loop cost of the corresponding group schemes to obtain multiple total field costs; then, the minimum value among the multiple total field costs is selected as the total field cost corresponding to the main path zone scheme.

[0224] Step S174: Determine the target main path zone scheme based on the total cost of the entire site corresponding to each main path zone scheme.

[0225] In this embodiment, the executing entity can compare the total cost of each main path scheme and determine the main path scheme with the lowest total cost as the target main path scheme. The target main path scheme can be understood as the main path scheme that enters the subsequent complete calculation stage.

[0226] In one possible implementation, if the total cost of multiple main path strip schemes is the same, or the difference is less than a preset range, the implementing entity can determine the target main path strip scheme according to a preset selection rule. The preset selection rule can be any rule suitable for further filtering among multiple main path strip schemes, for example, determined according to the main path strip length, the smoothness of the main path strip, or a preset priority order.

[0227] Step S175: Perform a complete calculation for the target main path scheme to output the final collector system topology scheme.

[0228] In this embodiment, the executing entity can perform a complete calculation for the target main path band scheme. The complete calculation may include: generating multiple large group partitioning schemes based on the target main path band scheme, generating multiple small group schemes, performing feasibility verification, calculating loop costs, and summing the total cost of each scheme. Unlike the aforementioned lightweight calculation, the complete calculation is not only used to obtain cost results but also to form the collector system topology scheme required for the final output.

[0229] In one possible implementation, the complete calculation can also be combined with the aforementioned associated storage, reusing the corresponding sub-group schemes and sub-group loop cost calculation results for large groups with the same unique signature, in order to reduce redundant calculations. Furthermore, when the difference between the total cost of multiple schemes and the first candidate scheme is less than a preset proportional threshold, the executing entity can also perform a secondary sorting of the multiple schemes based on spatial aggregation degree and intra-row continuity indicators, thereby determining the final output power collection system topology scheme.

[0230] In this embodiment, by dividing the processing of multiple main path zone schemes into two stages, lightweight calculation and full calculation, different main path zone schemes can be quickly screened based on the total cost of the entire project. Then, full calculation is performed on the selected target main path zone scheme, thereby improving the overall solution efficiency in the scenario of multiple main path zone schemes while ensuring the accuracy of the final output scheme.

[0231] It should be noted that the deduplication strategy in this embodiment is not a general caching processing method that exists independently of the grouping structure, but rather relies on the two-level grouping structure of this invention to be established. Specifically, since the non-terminal wind turbines in each row are fixedly assigned to the basic group corresponding to that row, and only the allocation of terminal points in each row varies between different grouping schemes, large groups consisting of the same set of wind turbine numbers will repeatedly appear between different grouping schemes. Based on this structural constraint, the aforementioned unique signature can be used to identify large groups that make up the same group, and the results of their continuous segmentation, feasibility verification, and loop cost calculation can be calculated once and reused multiple times. Correspondingly, in traditional free-layout or free-grouping scenarios, since wind turbines can freely belong to different groups, there is no large number of repetitive substructures formed by the fixed assignment of non-terminal points, and therefore, the structural basis for adopting the deduplication strategy of this embodiment is often lacking. In other words, the deduplication strategy in this embodiment is a calculation strategy closely coupled with the large grouping mechanism formed by the fixed assignment of non-terminal points and the allocation of terminal points.

[0232] In one possible implementation, the step of performing a secondary sorting based on spatial aggregation degree and intra-outcome continuity may specifically include:

[0233] For each group, the spatial aggregation index is used as the first calculation factor, and the remainder obtained by subtracting the intra-row continuity index is used as the second calculation factor. The first calculation factor and the second calculation factor are multiplied to obtain the comprehensive spatial quality score of the group.

[0234] For each grouping scheme, the arithmetic mean of the spatial quality scores of all groups included in the scheme is taken as the spatial quality score of the scheme.

[0235] The multiple schemes are sorted in ascending order of spatial quality score, and the optimal scheme after sorting is taken as the final output power collection system topology scheme. The lower the spatial quality score, the more compact the wind turbine spatial distribution of each group in the scheme and the better the continuity within the row.

[0236] In one specific implementation plan, an optimization strategy for the collection system of a large-scale offshore wind farm that takes into account intensive use of the sea is provided.

[0237] like Figure 3 and Figure 4 As shown, the offshore wind farm's power collection system is used to collect the electrical energy generated by each wind turbine generator and transmit it to the offshore substation via submarine cables. The power collection system consists of multiple power collection circuits, each of which connects several wind turbines in series to the offshore substation via submarine cables, thus realizing the collection and transmission of electrical energy.

[0238] Optimizing the current collection system involves optimizing the connection method and topology of the submarine cable to minimize construction costs. In offshore wind farm design, the number of wind turbines carried by different cable segments at different locations within a single current collection loop varies, resulting in different current transmission rates and therefore different cable cross-sectional specifications and costs per unit length. Optimizing cable costs cannot solely rely on reducing the total cable length; it must also consider the price differences between cables with different cross-sections.

[0239] Intensive use of sea area is a crucial constraint in the construction of offshore wind farms. In layouts that do not consider intensive sea area use, submarine cables can be laid freely within the wind farm, with wind turbine groups typically arranged radially, each circuit radiating outwards from the offshore substation. In layouts that consider intensive sea area use, submarine cable placement is constrained: horizontally, cables are laid parallel to the wind turbine array; vertically, cables are concentrated within a main path, forming a longitudinal corridor. The requirement for intensive sea area use stems from relevant policy regulations, which stipulate that, in principle, collector cables within the wind farm should be laid parallel to the wind turbine array and cannot be laid arbitrarily, in order to increase the feasibility of combined utilization of the sea area between wind turbines.

[0240] Topology optimization methods for offshore wind farm collection systems in related technologies are mainly aimed at wind farms that do not consider intensive sea use constraints. These methods often assume that the wind turbine strings are radially distributed, radiating outwards from the booster station, and mainly use clustering algorithms for turbine grouping and line planning.

[0241] In one related technique, an improved fuzzy C-means clustering algorithm and Delaunay triangulation method are used for automatic planning of power collection lines in offshore wind farms. This method divides the wind turbine array into multiple groups using a clustering algorithm, and within each group, triangulation is used to determine the connection paths, thereby achieving automatic generation of power collection lines.

[0242] Another related technology employs an automatic planning method for power collection lines under multi-dimensional constraints. This method includes steps such as intelligent partitioning, cost-optimal planning within a region, global cost-optimal planning across regions, and T-connection path optimization. The intelligent partitioning step is implemented based on a fuzzy clustering algorithm, dividing the wind turbine array into multiple radial regions centered on the substation and limiting the capacity of each region, thereby reducing the complexity of the planning problem.

[0243] In another related technology, a sector-based power collection line planning method is employed. This method uses a polar coordinate system to partition the wind farm into sector regions, forming multiple sub-regions. Within each sub-region, a heuristic optimization algorithm is used to calculate the power collection line path. Specifically, based on the location information of the substation and multiple wind turbines, the method randomly determines the starting wind turbine directly connected to the substation. Starting from this starting turbine, the remaining wind turbines are connected in series according to preset control conditions to obtain the power collection line. This process is repeated multiple times, and the shortest power collection line is selected as the optimal line. This method uses a sector-based scanning method based on the radial distribution assumption.

[0244] Most related technologies assume that submarine cables can be freely laid within the wind farm and that wind turbines are grouped radially to minimize cable length. Therefore, angle-based clustering algorithms or angle-based scanning methods are introduced. However, when submarine cables need to be laid as close together as possible, and can only be laid along the row direction and main path direction, these technologies become almost ineffective or no longer applicable. According to the requirements of intensive marine use, power collection cables within a wind farm should, in principle, be laid parallel to the wind turbine array and should not be laid arbitrarily, in order to increase the feasibility of combined utilization of marine space between wind turbines.

[0245] Since the unit price of submarine cables is related to the deployment plan, the unit price of each cable segment cannot be determined without a defined deployment plan. Therefore, related technologies primarily aim to minimize the cable length. However, the cost of submarine cables depends not only on the cable length but also on the unit price. To achieve the minimum cost, both the total cable length and the length of the more expensive segments must be short. Clustering algorithms, in principle, cannot solve this optimization problem that simultaneously considers length and tiered unit prices.

[0246] Taking a row of 8 wind turbines as an example, under the constraint of the maximum load capacity per single loop, they can be divided into two groups, each containing 4 turbines, or into two groups of 5 and 3 turbines. Submarine cables have minimum size specifications: cables connecting one or two turbines are type 1, cables connecting three turbines are type 2, and cables connecting four turbines are type 3. To simplify calculations, the unit price of the submarine cable can be expressed as a ratio to the unit price of type 1 cable. When the unit price of submarine cables with different cross-sections exhibits different stepped increase patterns, the total length of the two grouping schemes may be the same, but the total cost may differ. The steeper the increase in the unit price of the submarine cable, the greater the economic advantage of uniform grouping over non-uniform grouping. Existing clustering algorithms cannot capture this stepped increase.

[0247] Clustering algorithms for related technologies cannot capture this tiered cost effect. Taking a row of 8 wind turbines with a maximum single-loop load of 5 as an example, it can be divided into two groups, each containing 4 turbines, or into two groups of 5 and 3 turbines. Under one submarine cable unit price model, the unit price of a submarine cable connecting one or two wind turbines is 1, the unit price of a submarine cable connecting 3 wind turbines is 1.2, and the unit price of a submarine cable connecting 4 wind turbines is 1.5. When using two groups, each containing 4 wind turbines, the internal connection sections within each group use 4 submarine cables with a unit price of 1, 2 submarine cables with a unit price of 1.2, and the terminal junction section uses 6 submarine cables with a unit price of 1.5, for a total price of 15.4. When using two groups of 5 and 3 turbines, 4 submarine cables with a unit price of 1 are used, 6 submarine cables with a unit price of 1.2 are used, and 3 submarine cables with a unit price of 1.5 are used, for a total price of 15.7. Under this unit price model, the total cost of the uniform grouping scheme is lower than that of the non-uniform grouping scheme.

[0248] Under another unit price model for submarine cables, the unit price for the cable connecting four wind turbines is 1.7, while the other unit prices remain unchanged. Using two groups, each containing four wind turbines, the total price is 16.6. Using two groups, one with five turbines and the other with three, the total price is 16.3. Under this unit price model, the total cost of the uneven grouping scheme is actually lower than that of the uniform grouping scheme.

[0249] Therefore, the step-increasing unit price of submarine cable sections directly affects the selection of the optimal grouping scheme. Related technologies only optimize for the shortest length and cannot adaptively select the optimal grouping based on the step-increasing unit price pattern, leading to optimization results that deviate from the optimal solution for actual cost.

[0250] The purpose of this implementation plan is to provide a topology optimization method for offshore wind power collection systems. This method can automatically generate topology schemes that meet the requirements of intensive marine use standards, take into account the tiered cost of submarine cables, and has high computational efficiency, thereby achieving dual optimization of offshore wind farm construction costs and marine resource utilization.

[0251] Step 1: Data Initialization and Environment Modeling

[0252] Step 1.1: Input parameter set:

[0253] Wind turbine coordinate set , substation coordinates .

[0254] Cost Mapping Table for Submarine Cable Cross-Section ,in, The number of wind turbines mounted. The cost per unit length of the corresponding submarine cable. This represents the maximum number of devices allowed on a single circuit.

[0255] Layout information: Using a coordinate clustering algorithm and setting a y-coordinate threshold, all wind turbines are automatically identified as J horizontal rows according to their geographical layout.

[0256] Step 1.2: Main path band and coordinate system preset:

[0257] Extract the midpoint of each row of adjacent fans as feature points. ( ).

[0258] From the booster station to each The connecting lines form the "main path zone" or "main path corridor". Submarine cables are laid in a geometrically overlapping manner within the corridor to meet the principle of intensive use.

[0259] There are multiple ways to select the main route strip, which needs to be screened. The selection principle is as follows: the difference between the x coordinates of potential feature points in two adjacent rows should not be greater than A, such as A=2000m. This value is related to the spacing between wind turbines and can be taken as 2-3 times the spacing between wind turbines. The larger this value is, the larger the search space is, but the slower the calculation is.

[0260] Step 2: Divide the entire field into sub-regions and perform coordinate transformation

[0261] Step 2.1: Quadrant division. Using the main path zone and the horizontal line of the booster station as the boundary, the wind field is divided into four independent calculation sub-regions: I (upper left), II (upper right), III (lower left), and IV (lower right).

[0262] Step 2.2: Regional computing power normalization. For regions II, III, and IV, perform coordinate transformation. This allows for the relative positions of wind turbines and the main path feature points in all regions. , booster station All are mapped to the shape of Zone I (top left).

[0263] Step 3: Generate large group partitioning scheme

[0264] For wind turbines within a sub-region, this implementation plan adopts a hierarchical optimization logic of "large group division + small group continuous segmentation".

[0265] The core of the group division lies in determining the "inlet" of each row of fans.

[0266] Step 3.1: Terminal point definition. The wind turbine closest to the main path in each row is defined as the terminal point of that row.

[0267] Step 3.2: Large group generation mechanism.

[0268] Feasibility of crossing rows: The terminal fans in a row can be assigned to fans in the same row or to fan units in the rows above.

[0269] Search space: The allocation and combination of terminal points are generated using a recursive algorithm. For Exhaust fans, considering the flexibility of terminal point allocation, have +1 choice, theoretically possible A large group allocation scheme. Terminal allocation can be:

[0270] Added to an additional large group (Independent of all basic groups);

[0271] Classified into any basic group (Not exceeding the basic group corresponding to its rank number);

[0272] The terminal point of the third row can belong to this row, that is, the third row; it can belong to any of the first two rows; or it can belong to a separate group, which contains only terminal points.

[0273] Large group set: Each solution corresponds to a set. .in Includes all converged to the first The fan number at the terminal point.

[0274] Step 4: Continuously divide the large group into smaller groups.

[0275] After the large group is determined, it needs to be further divided into smaller subgroups within each large group. This implementation scheme adopts a continuous partitioning algorithm based on size constraints, pattern filtering, and linear order preservation. The specific steps are as follows:

[0276] Step 4.1: Generate the total size of the large group partition.

[0277] First, calculate all possible group sizes for n wind turbines within a large group, assuming the maximum load capacity k of the submarine cable is met.

[0278] The range of m values ​​is determined as follows: the range of m values ​​for the number of groups is [[n / k], n].

[0279] Numerical decomposition: For each possible m, use a recursive algorithm to decompose the integer n into the sum of m positive integers, where each integer (group size) satisfies 1. <v<k。

[0280] Descending order constraint and deduplication: To avoid redundant calculations, a descending order constraint is used during numerical decomposition (i.e., the size of the next subgroup is no larger than that of the previous subgroup), generating a unique "partition outline". For example, if n=8 and k=5, then the range of m is [2, 8]. For possible m=3, 8 can be decomposed into [5, 2, 1], [3, 3, 2], etc.

[0281] Step 4.2: Experience-based segmentation pattern filtering

[0282] To ensure the economy and ease of construction of the final topology, and to avoid creating overly fragmented groups (such as loops containing isolated fans), the system introduces a feature pattern filtering mechanism:

[0283] Pattern definition: Define the feature vector to be excluded (Exclude Pattern). For example, [1] means prohibiting groups with a size of 1, and [2, 1] means prohibiting combinations with a size of 2 and a size of 1. For example, pattern filtering [1] can filter out the partition result of [5, 2, 1] to avoid subsequent invalid calculations.

[0284] Filtering logic: Iterate through all partition outlines and remove schemes containing specific unreasonable size combinations. This step effectively reduces the search space for subsequent geometric calculations and improves the engineering usability of the results.

[0285] Step 4.3: Linear Sequential Continuous Segmentation Logic

[0286] For each group of filtered configurations, the final segmentation is performed based on the geometric position of the fan within the row:

[0287] Wind turbine serialization: Arrange the wind turbines in a large group in a linear order from farthest to closest to the main path and from top to bottom.

[0288] Full Length Permutation: Performs a full permutation of the selected scale configuration. For example, for the scale configuration [4, 4, 2], it generates three wind turbine grouping methods: [4, 4, 2], [4, 2, 4], and [2, 4, 4], to cover different access centers.

[0289] Linear slicing: Based on the arranged length sequence, the point sequence is linearly cut to divide the large group into several small groups.

[0290] Step 5: Cost Calculation Model for the Power Collection System

[0291] Step 5.1: Unify mathematical expressions

[0292] For any of the A group (loop) composed of typhoon generators. Its total construction cost Defined as the algebraic sum of the costs of all physical segments within the loop:

[0293] ;

[0294] In the formula, This indicates the number of wind turbines (i.e., the current carrying capacity level) currently supported by the physical segment. Counting begins from the last wind turbine in the loop; for each wind turbine passed, the current carrying capacity level is... Add 1. Indicates the current carrying capacity level as The corresponding cost per unit length of submarine cable (based on a pre-set tiered cost table) Decide). Indicates bearing The geometric length spanned by the physical cable section for typhoon generator power. This length consists of the following three parts according to the "figure-seven" rule: .in, Let S be the total construction cost of loop S, which is the sum of the construction costs of all cables and related facilities required to connect all the wind turbines in this group into a loop. n is the total number of wind turbines in the group. i is the loop variable, representing which physical segment is being calculated.

[0295] Step 5.2: Detailed breakdown of physical segment length

[0296] Inner section The horizontal distance between adjacent fans in the same row.

[0297] Find the correct paragraph If the fan is not the terminal point, move horizontally from the fan to the main path feature point of this row. The distance.

[0298] Main road section : Along the main path, feature points in different rows and The vertical / diagonal distance of movement between them.

[0299] Step 5.3: Typical Application Examples of Length Calculation

[0300] To demonstrate the universality of the formula, three examples covering different engineering scenarios are given below.

[0301] Example 1: Spanning-row connection combination

[0302] Scenario description: The group consists of 4 wind turbines. Located in the first row, This is the terminal point of the 3rd row. This is the terminal point of the 4th row.

[0303] Loop logic: .

[0304] Cost breakdown:

[0305] Section 1 ( ): Capable of carrying the energy of one wind turbine. Cost .in, Represents wind turbine To the fan The straight-line distance. This represents the unit cost of the submarine cable corresponding to the first-level current carrying capacity.

[0306] Section 2 ( ): Carries the energy of 2 wind turbines. The path is as follows: arrive (Find the correct path), then follow the main path... arrive Cost .(Note: for arrive (distance) among them, The main path features are represented in the first row. These represent the feature points of the main path in the second row. This represents the unit cost of the submarine cable corresponding to the second-level current carrying capacity. Represents wind turbine To the main path The distance. Represents the main path band Click Distance between points.

[0307] Section 3 ( ): Carries the energy of 3 wind turbines. The path is along the main path from arrive Cost .in, This represents the unit cost of the submarine cable corresponding to the third-level current carrying capacity. Representative terminal point To the terminal point Distance on the main path band.

[0308] Section 4 ( ): Carries the energy of 4 wind turbines. The path is from... The remaining section of the main path leads to the booster station. Cost. .in, This represents the distance from the substation access point to a certain reference point. Representative terminal point The distance to a certain intermediate reference point (or the entrance to the booster station).

[0309] Example 2: Internal grouping scenario within the same row

[0310] Scenario description: The first row has 10 wind turbines, with a maximum load capacity of... One of the groups is A (6 units).

[0311] Loop logic: The farthest wind turbine in group A (No. 1) is connected to wind turbine No. 6, and No. 6 serves as the terminal point to access the main path.

[0312] Cost calculation: In group A, the cost from segment 1 to segment 2 is... Segments 2 to 3 Use number segments 5 to 6 All main trunk sections connecting to the main path from terminal point 6 and down to the booster station use the same method. .

[0313] Step 6: Optimal Selection

[0314] Step 6.1: Summarize the solutions and calculate all The sum of the costs of all subgroups under the large group scheme.

[0315] Step 6.2: Final output, outputting the topology connections and costs.

[0316] Step 7: Calculate the optimal combined result for Zone I and Zone II

[0317] Step 7.1: Calculate the optimal result for region I. Calculate the optimal result for region I based on steps 3, 4, 5, and 6.

[0318] Step 7.2: Calculate the optimal result for region II.

[0319] Step 7.2.1: Coordinate Transformation Method

[0320] Since Zone I (upper left zone) is the algorithm's baseline zone (i.e., if the wind turbine is on the left side of the main path, it needs to move to the right to merge into it; if the main path is above the booster station, it needs to move downwards to merge into it), the other three zones must be mirrored to completely synchronize their relative positions to the pattern of Zone I.

[0321] Assume the coordinates of the booster station (convergence endpoint) are as follows: The original coordinates of a certain wind turbine are The coordinates of a feature point on the main path are .

[0322] The following are the methods for full coordinate synchronization transformation in the other two regions:

[0323] 1. Conversion method for Region II (upper right region)

[0324] The fan is located on the right side of the main path, above the booster station. A horizontal mirror image of the main path axis is then created, positioning the fan on the left side of the path. Fan: Main path feature points: ; The axis coordinates remain unchanged: ; After the conversion, the lateral displacement of the wind turbine relative to the substation changes from positive to negative. In the algorithm logic, "to the right" ( The command "find the main path (direction)" is physically equivalent to "turn left (direction)" after being restored to the actual coordinates. (Direction) Find the main path.

[0325] 2. Region III (Lower Left Region) Conversion Method

[0326] The fan is located on the left side of the main path, but below the substation. A vertical mirror image of the substation's horizontal line is then created, positioning the fan above the substation. Fan: Simplified to Main path feature points: ; The axis coordinates remain unchanged: ; After the conversion, the downlink direction of the main path ( The direction (after restoration) physically becomes the upward direction ( (Direction), which meets the requirement of the downward convergence of the wind turbine below.

[0327] 3. Region IV (lower right region) conversion method

[0328] The wind turbine is located on the right side of the main path and below the booster station. Perform a center-symmetric mirroring (or perform two mirroring operations, one horizontal and one vertical). Wind turbine: Main path feature points: Map all points from the lower right to the upper left to perfectly match the solution environment of Region I.

[0329] When performing multi-regional integrated solutions, this implementation plan does not simply change the coordinate symbols of the wind turbines, but rather uses the coordinates of the booster stations. Using the origin of the operator, perform spatial transformations on the geometric elements within the region (including wind turbine nodes and discrete points on the main path).

[0330] This synchronous transformation mechanism ensures the physical consistency of the '7-shaped' intensive sea utilization rule across the four quadrants, avoiding the redundant development of pathfinding algorithms for different quadrants and significantly reducing system redundancy.

[0331] Step 7.2.2: Calculate the optimal result for region II based on steps 3, 4, 5, and 6.

[0332] Step 7.3: Add the results of Zone I and Zone II to obtain the optimal result under the main route scheme.

[0333] Step 7.4: Traverse the main path schemes, repeat steps 7.1-7.3, and obtain the optimal result for each main path scheme.

[0334] Step 7.5: Sort all main path schemes in ascending order based on the cost of the optimal scheme to obtain the optimal main path scheme and topology connection scheme.

[0335] Step 8: Calculate the results for Zones III and IV.

[0336] Similarly, based on step 7, the optimal comprehensive solution for zones III and IV is calculated.

[0337] Step 9: Finally, the optimal power collection line topology scheme for the entire field area is obtained.

[0338] According to an embodiment of the present invention, an electronic device is provided; please refer to... Figure 5 The electronic device in this embodiment may include one or more of the following components: a processor, a network interface, memory, non-volatile memory, and one or more application programs, wherein the one or more application programs may be stored in non-volatile memory and configured to be executed by one or more processors, and the one or more programs are configured to perform the methods as described in the foregoing method embodiments.

[0339] According to embodiments of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a computer, causes the computer to perform the method described in any of the above embodiments.

[0340] According to embodiments of the present invention, a computer program product comprising instructions is also provided, which, when executed by a computer, cause the computer to perform a method in any of the above embodiments.

[0341] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A topology optimization method for offshore wind farm power collection systems considering intensive sea use constraints, characterized in that, include: Obtain the set of wind turbine coordinates, substation coordinates, maximum number of single-circuit loads, and cost mapping table for submarine cable cross-sections. The cost mapping table records the correspondence between current carrying capacity level and cost per unit length. The wind turbine layout is identified based on the set of wind turbine coordinates, and a main path band connecting each row to the substation is constructed. The terminal point is determined based on the relative position of each exhaust fan and the main path belt. The non-terminal point fans of each row are fixedly assigned to the row group, and the terminal points are assigned to the corresponding group according to the preset rules, generating a variety of group division schemes. The wind turbines in each group are arranged into a linear sequence according to a preset sorting rule, and then continuously divided based on the linear sequence and the maximum number of loads per loop to generate multiple group schemes; Feasibility verification was performed on cross-row groups in the group scheme based on preset cross-row physical constraints, and infeasible schemes were eliminated. Based on the main path band, the cost of the loop is calculated for each group that has passed the verification by using a broken line path along the row direction and the main path band direction, and the cost of each physical segment is determined by combining the cost mapping table. By summing up the circuit costs of each group, the total cost of the entire field is obtained from various group division schemes, and the optimal power collection system topology scheme with the best total cost is output.

2. The method according to claim 1, characterized in that, The step of identifying the wind turbine layout based on the wind turbine coordinate set and constructing the main path band connecting each row to the substation includes: Based on the longitudinal coordinate distribution of each fan, multiple fan rows are identified, and the fans in each row are sorted according to their lateral coordinates. Based on the sorted wind turbine coordinates, the geometric center of adjacent wind turbines in each row is extracted as the candidate feature point of that row, and multiple candidate main paths are generated by combining the candidate feature points of each row. Candidate main paths that meet the following conditions are selected as main path bands: the lateral offset between adjacent candidate feature points does not exceed a preset threshold, and the lateral offset between the last candidate feature point and the booster station does not exceed a preset threshold.

3. The method according to claim 1, characterized in that, The steps of determining terminal points based on the relative positions of each exhaust fan and the main path, fixing non-terminal point fans in each row into their respective large groups, and assigning terminal points to their corresponding large groups according to preset rules to generate multiple large group division schemes include: For each row of fans, calculate the horizontal distance between each fan in the row and the characteristic point of the main path in the row, and determine the fan with the smallest horizontal distance as the terminal point of the row, and the remaining fans as the non-terminal point fans of the row. The non-terminal point fans in each row are fixedly assigned to the corresponding basic group of that row; The terminal points of each row are assigned to the corresponding basic group, other basic groups with numbers lower than the row number, or additional groups. Different assignment options correspond to different group division schemes.

4. The method according to claim 3, characterized in that, The steps of forming a linear sequence of wind turbines within each major group according to a preset sorting rule, and then continuously dividing the group based on the linear sequence and the maximum load capacity of a single loop to generate multiple group schemes include: Arrange the fans within the large group in ascending order of row number as the primary sorting key and ascending order of horizontal coordinate as the secondary sorting key to form a linear sequence; Based on the maximum number of loads on a single loop, the total number of wind turbines in a large group is decomposed into multiple small group sizes; each combination consists of multiple positive integers not exceeding the maximum load, and the sum of the positive integers is equal to the total number of wind turbines in the large group. By performing full permutations for each group size combination, sequences of various lengths are obtained; For each length sequence, a subsequence of wind turbines of the corresponding length is extracted from the linear sequence to generate a subgroup scheme for that large group.

5. The method according to claim 4, characterized in that, The step of verifying the feasibility of cross-row groups in the group scheme based on preset cross-row physical constraints and eliminating infeasible schemes includes: For cross-row groups of multi-row fans, the following checks are performed sequentially: first-row integrity constraint verification, intermediate terminal row single-point constraint verification, and intermediate empty row constraint verification. The first-row integrity constraint verification determines whether all fans in the first row of the cross-row group belong to that group. The intermediate terminal row single-point constraint verification determines whether each intermediate terminal row of the cross-row group contains only one fan. The intermediate empty row constraint verification determines whether the number of fans in the non-terminal rows between the first and last rows of the cross-row group is zero. The intermediate terminal row single-point constraint verification and the intermediate empty row constraint verification are performed based on the overall fan layout information. Eliminate the group schemes corresponding to cross-row groups that do not meet any of the above constraint verifications.

6. The method according to claim 5, characterized in that, The steps of calculating the loop cost for each verified group based on the main path band using a broken line path along the row direction and the main path band direction, and determining the cost of each physical segment in conjunction with the cost mapping table, include: For two adjacent wind turbines within the group, the physical length of the cable connecting the two wind turbines is determined using the following path calculation methods, based on their relative positions in the row and whether they are terminal points: If the two wind turbines are in the same row, the straight-line distance between them is used as the physical segment length. If the two wind turbines are both terminal points and are in different rows, the straight-line distance between them is used as the physical segment length. If at least one of the two wind turbines is not a terminal point and is in a different row, the physical segment length is the sum of the alignment segment length from one wind turbine to the feature point of the main path in that row, the main path segment length from the feature point of that row to the feature point of another row along the main path, and the alignment segment length from the other wind turbine to the feature point of the main path in its row. For the last wind turbine in the group, the physical segment length is the sum of the alignment segment length from the wind turbine to the feature point of the main path in its row, the main path segment length from the feature point of that row to the feature point of the last row along the main path, and the distance from the feature point of the last row to the substation. Based on the length of each physical segment and the number of wind turbines it supports, the cost per unit length is obtained by querying the cost mapping table. The cost per unit length is then multiplied by the length of the corresponding physical segment to obtain the cost of each physical segment. Finally, the costs of each physical segment are summed to obtain the cost of the group loop.

7. The method according to claim 6, characterized in that, Before the step of summarizing the costs of each sub-group loop to obtain the total cost of the entire site for various large-group division schemes, the method further includes: After sorting the wind turbines within each major group by row number and horizontal coordinate, the sequence of wind turbine numbers included in that major group is extracted as the unique signature of that major group. Iterate through all large group partitioning schemes, perform continuous partitioning, feasibility verification and loop cost calculation only once for large groups with the same unique signature, and store the obtained small group schemes and corresponding small group loop costs associated with the unique signature. For each group division scheme, the cost of the small loops corresponding to each of the major groups included in the scheme is obtained by querying the associated storage, and the obtained small loop costs are summarized to obtain the total cost of the entire site for the group division scheme.

8. The method according to claim 7, characterized in that, The steps of summarizing the costs of each group of loops to obtain the total cost of the entire field for various large group division schemes, and outputting the optimal collector system topology scheme in terms of total cost, include: The total cost of each scheme is ranked, and the scheme with the lowest total cost is selected as the first candidate scheme. If the difference between the total cost of multiple schemes and the first candidate scheme is less than a preset ratio threshold, then the spatial aggregation degree index and the intra-row continuity index are calculated for each of the multiple schemes. Based on the spatial aggregation degree index and the intra-row continuity index, the multiple schemes are sorted a second time, and the optimal scheme after the second sorting is taken as the final output power collection system topology scheme.

9. The method according to claim 8, characterized in that, The steps for determining the optimal total cost of the power collection system topology also include: Lightweight calculations are performed for each main path zone scheme to obtain the total cost of the entire site for each main path zone scheme; wherein, the lightweight calculations include generating multiple large group division schemes, generating multiple small group schemes, performing feasibility verification and calculating loop cost, and do not perform secondary sorting based on spatial aggregation degree index and intra-row continuity index; The target main path scheme is determined based on the total cost of the entire site corresponding to each main path scheme. Perform a full calculation for the target main path scheme to output the final collector system topology scheme.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 9.

Images