Offshore wind farm collection system loss pre-computation and large-scale solution acceleration method

Abstract

A kind of offshore wind farm power collection system loss pre-calculation and large-scale solution acceleration method, comprising: constructing initial mathematical model considering cable construction cost and life cycle loss cost, and then converting nonlinear loss term into linear cost coefficient through network loss pre-calculation subproblem, to obtain mixed integer linear programming model easy to solve;Based on the location of wind turbine and substation, the initial feasible solution of hot start that meets the engineering constraints is constructed by using clustering partition and intra-sector routing strategy;Based on distance threshold, potential connections are screened, connections exceeding the threshold in the hot start solution are supplemented to the candidate set to form the final candidate cable set;Combined with the candidate set and the hot start solution, the solver is used to solve the optimization model efficiently, and the planning scheme is output.The present application significantly improves the solution efficiency and engineering practicability of super large scale offshore wind farm power collection system planning under the premise of accurately depicting network loss economy through model linearization and systematic solution acceleration design.

Description

TECHNICAL FIELD

[0001] The present application relates to offshore wind farm power collection system loss calculation, in particular to offshore wind farm power collection system loss precalculation and large-scale solution acceleration method. BACKGROUND

[0002] Offshore wind power is developing rapidly, and the problem of power collection and transmission within offshore wind farms is increasingly prominent. The power collection system, as a key engineering system connecting wind turbines and offshore booster stations, accounts for 15% to 30% of the total project cost. It is not only an important part of project investment and operating costs, but also a key link affecting project feasibility and system operation efficiency.

[0003] From the perspective of system planning, offshore wind farm power collection system planning not only needs to be optimized in terms of construction cost, but also must consider power loss cost, operation reliability, construction feasibility and other requirements in the whole life cycle. Essentially, it is a combinatorial optimization problem with a large number of binary decision variables and nonlinear constraints: on the one hand, the selection of line direction and topology determines how different wind turbine powers are collected to the booster station; on the other hand, cable selection and capacity allocation determine the line loss level and long-term operation cost. When the number of wind turbines reaches dozens or even hundreds, all possible wind turbine-wind turbine, wind turbine- booster station connection combinations will generate an extremely large set of candidate lines, making the problem size expand rapidly with the site size.

[0004] In existing research and engineering practice, there are usually two technical routes to design the power collection system topology and cable configuration. One is the "quasi-experience design" method dominated by engineering experience and rule base, such as giving a feasible topology according to zoning, branching structure or empirical formula, and then making local adjustments. This method is simple to implement and easy to integrate with engineering processes, but it is often difficult to systematically trade off line loss and life cycle cost, and has limited ability to describe complex constraints and multi-objective compromise. The other is the "optimization design" method represented by mathematical programming and intelligent optimization algorithms, including various models based on mixed integer linear programming (MILP), mixed integer quadratic programming (MIQP), heuristic and metaheuristic algorithms. These methods can theoretically handle economy, loss and constraint conditions at the same time, but in practical applications they generally face two structural problems:

[0005] The first is the nonlinearity introduced by the loss modeling. The line loss in the collection system is usually related to the line resistance and the square of the active power passing through. If the quadratic term of the loss is directly included in the objective function or constraint, an MIQP or more general nonlinear integer programming is formed. Such models are difficult to solve and strongly dependent on the solver, and it is difficult to guarantee a high-quality solution within an acceptable engineering time in a large-scale wind farm scenario. Therefore, some works choose to simplify or even ignore the loss, only considering the capacity upper limit in the constraint; others use rough linearization, scenario equivalence, and other approximation methods. These compromises improve the solving efficiency, but inevitably lose the fine characterization of the full-life loss cost, resulting in a deviation between the optimization results and the actual operation economy.

[0006] The second is the out-of-control of the candidate line and the solving scale in large-scale scenarios. If all potential connections between wind turbines and substations are included in the candidate set, the number of line variables and related constraints in the model will grow quadratically. At the same time, if the engineering constraints such as non-intersection of lines, node degree range, and multiple parallel lines are also explicitly considered, the search space of the solver in the branch and bound will expand dramatically. In existing research, some works try to reduce the scale by setting a distance threshold, pre-dividing the area, or using heuristics to generate candidate arcs, but they usually lack a systematic approach closely coupled with mathematical programming models, and lack a clear theoretical and repeatable application of the integrated technology path from candidate generation to solving acceleration. Without high-quality initial solutions and reasonable candidate set control, even if the linear model form is maintained, the solving performance of mixed integer programming on the scale of hundreds of wind turbines is still unsatisfactory.

[0007] For large-scale MILP / MIQP problems, the warm-start technique is widely recognized as an effective acceleration method. The basic idea is to construct a set of feasible initial solutions with better objective values before optimization, which are used as reference by the solver in the branch and bound process, thereby shortening the search path and improving the solving efficiency. In the field of offshore wind farm collection system planning, common initial solution construction methods include radial layout based on minimum spanning tree and partitioned connection scheme based on simple clustering. However, these methods are often designed around traditional radial or simple ring topology and do not systematically adapt to actual engineering constraints of the collection system, such as node degree requirements of "one-in-one-out" or "two-in-one-out", cable non-intersection constraints, parallel laying constraints, etc. At the same time, there is a lack of an overall acceleration framework that is linked to the construction of candidate cable sets.

[0008] Overall, the prior art mainly has the following deficiencies in the direction of "loss modeling + large-scale solving": first, there is a lack of a method for converting cable construction cost and discounted full life cycle network loss cost into linear unit length cost coefficient, making it difficult to balance between accurately accounting for network loss and maintaining model efficiency; second, there is a lack of a systematic candidate cable set size control strategy for large-scale offshore wind farms, making it difficult to significantly reduce the number of candidate arcs while ensuring model optimality; third, there is a lack of a warm start construction method and a solution acceleration framework closely coupled with mathematical programming models, and it is still not possible to obtain a high-quality collection system optimization scheme stably and efficiently in a few hundred wind turbine scale offshore wind farms. Therefore, there is an urgent need for a new collection system planning technology that can accurately reflect the economic efficiency of line loss while maintaining the linear or near-linear structure of the overall model, and can achieve efficient optimization and solution of large-scale offshore wind farm collection systems by reasonably constructing the candidate line set and high-quality initial solution.

[0009] It should be noted that the information disclosed in the above background section is only for understanding the background of the present application, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0010] The main purpose of the present application is to overcome the defects in the above background art, and to provide a method for pre-computing and accelerating large-scale solution of offshore wind farm collection system loss.

[0011] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0012] A method for pre-computing and accelerating large-scale solution of offshore wind farm collection system loss, comprising the following steps:

[0013] S1, for the offshore wind farm collection system planning problem, an initial mathematical model considering cable construction cost and full life cycle loss cost is constructed;

[0014] S2, by solving the network loss pre-computation sub-problem, the nonlinear loss term in the initial mathematical model is converted into a linear cost coefficient, and a mixed integer linear programming model is obtained;

[0015] S3, based on the location information of wind turbines and substations, a strategy combining clustering partition and intra-sector routing is adopted to construct a warm start initial feasible solution that meets engineering constraints;

[0016] S4, based on a pre-set distance threshold, potential connections between wind turbines are screened, and connections in the warm start initial feasible solution that exceed the threshold are supplemented to the candidate set to construct a final candidate cable set for solving the mixed integer linear programming model;

[0017] S5, based on the final candidate cable set and the thermal start initial feasible solution, using a solver to optimize the solution of the mixed integer linear programming model, output the power collection system planning scheme.

[0018] A computer program product, the computer program is executed by the processor to realize the offshore wind farm power collection system loss precomputation and large-scale solution acceleration method.

[0019] The present application has the following beneficial effects:

[0020] The present application provides a kind of offshore wind farm power collection system loss precomputation and large-scale solution acceleration method, effectively solve the solving efficiency problem caused by the introduction of model nonlinearity due to power loss quadratic term in traditional mathematical optimization, and the explosion of candidate arc number of large-scale wind farm.This method converts power loss into linear coefficient and integrates into cost calculation through precomputation, thereby maintaining the linear or approximate linear structure of the overall model under the premise of accurately reflecting the line loss economy, significantly improve the solving efficiency of optimization model.

[0021] Specifically, on the one hand, the proposed network loss precomputation method can accurately depict the network loss term under the assumption of DC power flow, and by converting the full life cycle network loss cost and one-time construction cost into linear unit length cost coefficient, it realizes the maintenance of model efficient solving characteristics while giving global optimization scheme considering economy and engineering practice. On the other hand, for large-scale wind farm scene, by combining distance screening and thermal start layout, and designing a thermal start strategy combining clustering partition and intra-fan routing, the number of candidate arcs can be systematically reduced and high-quality initial feasible solution can be provided. This not only significantly improves the solving efficiency and convergence speed of mathematical programming model when facing hundreds of wind turbine scale, but also enhances the scalability of the method in large-scale scenarios, thereby realizing efficient and stable solution of super large offshore wind farm power collection system planning problem.

[0022] Other beneficial effects in the embodiments of the present application will be further described in the following. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 The overall flowchart of the offshore wind farm power collection system loss precomputation and large-scale solution acceleration method of the present application.

[0024] Figure 2 The wind turbine and substation coordinates and candidate sea cable of the 42-turbine wind farm are shown in the figure.

[0025] Figure 3A The 42-turbine wind farm power collection system planning result schematic diagram of conventional MIQP model (comparative example 1) is shown in the figure.

[0026] Figure 3BThis is a schematic diagram of the planning results of the power collection system of a 42-wind-turbine wind farm in the MILP model (Example 1) after pre-calculation of network losses.

[0027] Figure 4 The coordinates of the wind turbines and substations for wind farms with 125 wind turbines and 300 wind turbines, as well as a schematic diagram of candidate submarine cables, are provided.

[0028] Figure 5 This is a schematic diagram of the solution gap-time curves for Comparative Example 2, Comparative Example 3, and Example 2 in a wind farm scenario with 125 wind turbines.

[0029] Figure 6 This is a schematic diagram of the solution gap-time curves for Comparative Example 3 and Example 2 in a wind farm scenario with 300 wind turbines. Detailed Implementation

[0030] The embodiments of the present invention will be described in detail below. It should be emphasized that the following description is merely exemplary and not intended to limit the scope and application of the present invention.

[0031] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of the present invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0032] This invention aims to solve the problem of low solution efficiency caused by nonlinear power loss modeling and massive candidate connections in the planning of offshore wind farm collection systems. It proposes an integrated optimization method that combines pre-calculation linearization of network loss with clustering-based hot start acceleration. By pre-converting loss costs into linear coefficients to maintain model efficiency, and combining partitioning strategies and distance screening to construct high-quality initial solutions and a simplified candidate set, this method achieves efficient and stable solutions to ultra-large-scale wind farm planning problems while ensuring the economic efficiency and engineering feasibility of the scheme.

[0033] See Figure 1 This invention provides a method for accelerating the pre-calculation and large-scale solution of losses in offshore wind farm collection systems, comprising the following steps:

[0034] Step S1: For the planning problem of offshore wind farm power collection system, construct an initial mathematical model that considers cable construction costs and total life cycle loss costs.

[0035] In some embodiments, the initial mathematical model constructed in step S1 is a mixed integer quadratic programming model or a mixed integer nonlinear programming model, and the objective function of the model includes a network loss cost term proportional to the square of the line transmission power, which is determined by the current-power relationship under the DC power flow assumption.

[0036] Step S2: convert the nonlinear loss term in the initial mathematical model into a linear cost coefficient through network loss pre-computation sub-problem solving, to obtain a mixed integer linear programming model.

[0037] In some embodiments, the network loss pre-computation sub-problem solving in step S2 includes: for line sections carrying different numbers of wind turbines, pre-computing and comparing the total cost per unit length of all available cable models; the total cost per unit length is composed of the sum of the unit length construction cost and the linearized term of the full life cycle network loss cost considering discount, wherein the linearized term of the network loss cost is calculated based on the square relationship between the cable transmission power and the DC power flow assumption; selecting the cable model with the minimum total cost per unit length and its cost value as the optimal cost coefficient under the corresponding number of wind turbine loads, and integrating it into the objective function of the main optimization model, thereby converting the nonlinear programming problem containing the power square term into a linear programming problem.

[0038] In some embodiments, the solving process of the network loss pre-computation sub-problem is independent of the iteration process of the main optimization model, and is a pre-processing step completed once before optimization solving.

[0039] Step S3: based on the location information of wind turbines and substations, an initial feasible solution for hot start is constructed by adopting a strategy combining clustering partition and intra-sector routing, which satisfies the engineering constraints.

[0040] In some embodiments, the construction of the initial feasible solution for hot start in step S3 includes: taking the substation as the reference point, dividing the wind turbine nodes into several fan-shaped regions according to the preset incoming line degree and sector capacity; in each fan-shaped region, an initial topology is generated by using a monotonic connection method based on fixed direction, specifically: calculating the projection coordinates of each wind turbine node in the region in the selected connection direction and its orthogonal direction, and sorting the nodes in the lexicographic order of the projection coordinates, and then connecting the sorted nodes in turn to form a monotonic path; according to the number of wind turbines carried downstream of each line section in the initial topology, matching the cable model that satisfies the capacity constraint and has the optimal cost, and calculating the total cost of the scheme; traversing the division methods with different wind turbine nodes as the sector scanning starting points, and selecting the scheme with the lowest total cost as the initial feasible solution for hot start.

[0041] In some embodiments, the fixed direction-based monotone connection method specifically comprises: selecting a connection direction and its unit orthogonal vector in the sector; calculating the signed projection of the position vector of each wind turbine node in the sector on the connection direction and its orthogonal vector to obtain a set of two-dimensional projection coordinates; sorting all wind turbine nodes according to the lexicographic order of the projection coordinates; and connecting the sorted wind turbine node sequence in turn to form a path with monotonically non-decreasing coordinates in the selected connection direction, which naturally satisfies the cable non-crossing and node "one-in-one-out" degree constraints.

[0042] In some embodiments, the initial topology generated by the sector routing strategy satisfies the engineering constraints of the offshore wind farm power collection system, and the engineering constraints at least include: each line connection satisfies the node degree limit of "one-in-one-out" or "two-in-one-out", and the line does not cross in spatial layout.

[0043] Step S4: filtering the potential connections between wind turbines based on a preset distance threshold, and supplementing the connections in the hot start initial feasible solution that exceed the threshold to the candidate set to construct a final candidate cable set for solving the mixed integer linear programming model.

[0044] In some embodiments, the construction of the final candidate cable set in step S4 comprises: first, setting a maximum candidate distance threshold, and only including the connection arcs between wind turbines and between wind turbines and substations with a Euclidean distance less than the threshold into the initial candidate set; then, supplementing all line arcs in the hot start initial feasible solution generated in step S3 whose connection length exceeds the maximum candidate distance threshold to the initial candidate set; and the initial candidate set and the supplemented line arcs together constitute the final candidate cable set.

[0045] Step S5: based on the final candidate cable set and the hot start initial feasible solution, using a solver to optimize and solve the mixed integer linear programming model, and outputting a power collection system planning scheme.

[0046] In some embodiments, in step S5, the hot start initial feasible solution is used as the initial input of the branch and bound process of the mixed integer linear programming solver to accelerate convergence when the solver is used for optimization and solving.

[0047] The proposed method for accelerating the pre-calculation and large-scale solution of losses in offshore wind farm collection systems cleverly transforms nonlinear loss terms into linear cost coefficients through network loss pre-calculation. This maintains the linear structure and efficient solution characteristics of the model while accurately considering the economic benefits of losses throughout the entire life cycle. Furthermore, by combining a hot-start construction method based on clustering partitioning and intra-sector routing, as well as a candidate set construction strategy that integrates distance screening and hot-start solution supplementation, the problem scale in large-scale scenarios is effectively controlled, and high-quality initial solutions are provided. Thus, in the planning of ultra-large offshore wind farms with hundreds of wind turbines, a globally optimized solution that is both economical and engineering feasible can be obtained stably and efficiently within an acceptable engineering timeframe.

[0048] The following further describes specific embodiments of the present invention, algorithm examples, and experimental verification.

[0049] A method for pre-calculating and accelerating large-scale solution of losses in offshore wind farm collection systems is proposed. The core of this method is to address the efficiency challenges in traditional mathematical optimization, such as the nonlinearity of the quadratic term in power loss leading to model nonlinearity and the explosion of candidate arcs in large-scale wind farms. Specifically, it includes pre-calculating power losses to convert them into linear coefficients that are integrated into cost calculations to maintain the linearity of the model; employing a combination of distance filtering and hot-start layout to reduce the number of candidate arcs; and designing a hot-start strategy that combines clustering partitioning and intra-sector routing to provide high-quality initial solutions. While ensuring planning feasibility, this method significantly improves the computational efficiency of large-scale wind farm collection system planning, enabling efficient optimization solutions for ultra-large wind farms.

[0050] Network loss pre-calculation subproblem

[0051] Offshore wind power collection system planning is a large-scale, NP-hard planning problem. The line loss term in the objective function is the most complex, significantly impacting the model's solution efficiency. As the number of wind turbines in an offshore wind farm increases, the size of the solution model and the number of candidate submarine cables will grow significantly. Therefore, employing an efficient method for characterizing the network loss term can effectively improve the model's solution efficiency and scalability to large-scale scenarios. Considering the nodes in the power collection system... voltage amplitude , This indicates the per-unit value, therefore in cables Current flowing through , For the active power transmitted on cable (i,j), the active power loss can be calculated using the DC power flow assumption. By approximation, we have .in Select the resistance per unit length of submarine cable type t for this line. Denote the engineering length of the submarine cable (i,j) as... Under the assumption of DC power flow, the objective function expression of the model, which includes one-time construction costs and network loss costs, can be written as:

[0052] (1)

[0053] (2)

[0054] in For the candidate submarine cable set, For line selection, This is a collection of alternative cable models; Model Construction cost per unit length of submarine cable Let be a binary variable indicating whether to use the type t submarine cable on arc (i,j). A value of 1 indicates that type t is selected, and a value of 0 indicates otherwise. The discount factor for converting total lifecycle network losses into costs is as follows: This refers to the annual utilization hours of the wind farm. The annual discount rate is 10%. For the operating years of the wind farm, The above model represents the objective function under the assumption of DC power flow. However, the introduction of active power loss transforms the model into a MIQP problem, posing significant challenges to solution efficiency and scalability in large-scale wind farm scenarios.

[0055] In this invention, the power of the current collection system is considered to be cumulatively added in units of wind turbines as it is transmitted along the cable. Let the rated power of a single wind turbine be... At this time, the transmission power on a certain submarine cable carrying k wind turbines is Given the freedom in cable selection and the limitation of choosing from only a pre-defined range of cable types, the sub-problem of minimizing the sum of construction cost and discounted network loss cost under the condition of supporting k wind turbines can be solved first at the single-line scale. Specifically, the cable sub-problem for supporting k wind turbines takes the following form:

[0056] (3)

[0057] in This refers to the rated power of the fan. Let be the upper limit of the capacity of submarine cable of type t. Solve using constraint equation (3) to ensure the carrying capacity... Typhoon generator capacity constraints Under the premise of the specified model, the sum of the minimum construction cost and the total life-cycle network loss cost considering the discount rate. This is then integrated into the objective function. Therefore, the quadratic term of the loss no longer appears explicitly in the master programming model; instead, it is... The objective function is introduced as a constant parameter. Let... Let be a binary variable for whether candidate arc (i, j) carries k wind turbines, equal to 1 if it carries k wind turbines, otherwise equal to 0. Then the total objective function of the collection system can be rewritten as a linear MILP form:

[0058] (4)

[0059] Cluster-based hot start method and candidate cable set construction method

[0060] Considering that the model's candidate cable set size increases with the number of wind turbines, in the case of large-scale deep-sea wind farms, the model will have very high solution complexity. This section proposes a cluster-based hot start method and a candidate cable set construction method based on distance screening and clustering results, aiming to accelerate the solution of large-scale offshore wind farm planning problems.

[0061] Cluster-based hot start construction

[0062] Hot start technology can significantly accelerate convergence and improve solution efficiency by providing a better initial solution in the optimization algorithm. The present invention adopts a hot start idea of "clustering and partitioning first, then connecting within the region". First, the number of sectors and the number of wind turbines per sector are pre-set according to the substation incoming line degree and candidate cable capacity. Then, taking a substation (step-up station) as the reference point and a certain wind turbine as the scanning starting point, the wind turbines are scanned and assigned to sectors in order of increasing polar angle. When the sector capacity quota is full, the wind turbine is assigned to a new sector until all wind turbines are assigned. Then, according to the partitioning situation, feasible lines are constructed within the partitioning, and cable models that meet the current-carrying capacity constraints are matched according to the number of downstream wind turbines. Finally, the total cost of the planning scheme is calculated, and the scheme with the lowest cost starting from all wind turbine nodes is selected as the hot start scheme. The above construction method based on clustering and partitioning can efficiently give an initial feasible hot start scheme.

[0063] In the conventional radial collection system intra-regional connection method, it is often treated as a minimum spanning tree problem, seeking the smallest topology with connected weighted edges. This problem has mature algorithms that can be solved accurately in polynomial time, such as Kruskal's algorithm and Prim's algorithm, with time complexities of and However, the actual radial collection system planning problem often imposes restrictions on the degree of nodes, requiring the planning scheme to meet the engineering constraints of "two-in-one-out" or "one-in-one-out". In this case, the minimum spanning tree method may not give a scheme that meets the engineering requirements. Meanwhile, the hot start stage only provides an initial feasible solution for the planning problem, mainly emphasizing feasibility rather than the quality of the solution. This invention proposes a sector direction-based monotonic connection method, which can efficiently give an intra-regional connection hot start scheme that meets the most stringent "one-in-one-out" standard and does not cross:

[0064] A fixed connection direction is defined within each sector S. Let the unit vector orthogonal to it be... For each fan in the sector Calculate the signed projection and according to the first back lexicographical sorting Obtain the sorting results And connect them sequentially to obtain a monotonic path within the region. This path is in the direction The upper part is monotonically non-decreasing, therefore for any non-connected edge... , , its in Projection range of direction and They do not overlap; even for adjacent blocks Also because The increasing relationship of directions still satisfies the non-intersection rule. In summary, the sector direction monotonic connection method used in this invention naturally satisfies the cable non-crossing principle and the "one in, one out" engineering constraint. The complexity enables the efficient construction of a hot start scheme.

[0065] Construction of candidate cable sets based on distance filtering and clustering results

[0066] Considering that the planning model aims to obtain the optimal power collection system planning scheme, and the objective function seeks to minimize the Euclidean distance of the connections between wind turbines in terms of construction and network loss costs, excessively long connections between wind turbines are unlikely to appear in the optimal layout and can be removed through targeted filtering. A maximum candidate distance (e.g., 5 times the wind turbine spacing) can be set first; cables exceeding this candidate distance are not included in the candidate cable set, thus effectively controlling the model size without affecting the optimality of the solution.

[0067] While using appropriate distances to filter candidate cable sets can effectively reduce model planning, the clustering hot-start method described in Section 1 often scans unevenly distributed fan-shaped areas when dealing with irregularly shaped or large-scale wind farm scenarios. In such cases, the heuristic connection method may construct some exceptionally long cables. Directly increasing the filtering distance to adopt this hot-start scheme would significantly increase the model size. This invention adopts a construction method that first filters the cable set using appropriate candidate distances, and then adds cables exceeding the maximum candidate distance from all hot-start schemes to the candidate cable set. This effectively utilizes the hot-start scheme without incurring excessive computational costs.

[0068] Experimental verification

[0069] Experiment 1: A 42-wind-turbine wind farm in China is used to verify the effect of network loss pre-calculation.

[0070] The coordinates of wind turbines and substations of the wind farm and the schematic diagram of candidate sea cables are shown in Figure 2 .

[0071] In this experiment, the conventional MIQP model (Comparative Example 1) and the MILP model after network loss pre-calculation (Embodiment 1) are used for solving, and the comparison results are shown in Table 1.

[0072] Table 1

[0073]

[0074] Under this experiment, the model without network loss pre-calculation has an optimal gap of 8.18% after 500 s of solving time, and still has a certain space for cost optimization. After using the network loss pre-calculation model, the global optimal solution is obtained in only 2.8 s, and the efficiency is significantly improved compared with the model without network loss pre-calculation.

[0075] Experiment 2: The effectiveness of the clustering hot start and candidate cable set construction method proposed in the application is verified in two larger-scale wind farms, each with 125 wind turbines and 300 wind turbines. The coordinates of wind turbines and substations of the wind farm and the schematic diagram of candidate sea cables are shown in Figure 4 . Figure 4 (a) is a scene of 125 wind turbines, Figure 4 (b) is a scene of 300 wind turbines. Figure 4 The purple line segment in represents the candidate sea cable obtained by distance screening, and the red line segment represents the expanded candidate sea cable obtained by the hot start solution.

[0076] In the scene of 125 wind turbines, the following three comparative examples are set to verify the effectiveness of the method of the application: Comparative Example 2: A suitable distance screening is used, but the hot start is not used.

[0077] Comparative Example 3: The hot start based on the clustering method is used, but the screening distance is expanded , which will lead to the growth of the model size.

[0078] Embodiment 2: The hot start based on clustering and the construction method of candidate cable expansion proposed in the application are used.

[0079] In the scene of 125 wind turbines, the solving effects of Comparative Examples 2 and 3 and Embodiment 2 are shown in Table 2.

[0080] Table 2

[0081]

[0082] Accordingly, the Gap-time curves of Comparative Example 2, Comparative Example 3 and Example 2 are shown in Figure 5

[0083] It can be seen that Comparative Example 2 needs to spend a certain time in the solution space to find a feasible solution because it does not use hot start; Comparative Example 3 quickly explores at the beginning of the solution because it provides an initial feasible solution, but the model size is increased due to the expansion of the distance screening value, which affects the branch and bound exploration efficiency, resulting in that the efficiency of obtaining a high-quality solution is not improved compared with Comparative Example 2; and Example 2 (the method of the present application) implements hot start at a very low computational cost, and the solution effect is significantly improved compared with Comparative Example 2 and Comparative Example 3.

[0084] In the scenario of 300 wind turbines, the Gap-time curves of Comparative Example 3 and Example 2 are shown in Figure 6

[0085] In this super large scale scenario, Comparative Example 2 without hot start cannot obtain a feasible solution within the given solution time limit (30000s), and Comparative Example 3 has an initial feasible solution, but the Gap is almost impossible to decrease due to the large size. Example 2 of the present application can successfully solve the problem and obtain a high-quality feasible solution (Gap<1%) in an acceptable engineering time in the super large scale scenario.

[0086] In summary, the present application proposes a loss pre-computation and large-scale solution acceleration method for an offshore wind farm power collection system, and the important innovative contributions and design points are as follows: on the one hand, an accurate and efficient network loss pre-computation method is designed, which converts the non-linear loss term into a linear cost coefficient under the assumption of direct current flow, so that the network loss cost in the whole life cycle can be accurately described while keeping the model efficiently solvable; on the other hand, a cooperative acceleration framework is proposed, which combines clustering partition, intra-fan routing and distance screening, and through constructing a high-quality hot start solution that meets the engineering constraints and dynamically optimizing the candidate cable set, the solution efficiency and scalability of the model in large-scale scenarios are significantly improved. Compared with the prior art, the present application has the following significant technical advantages: not only the economic accuracy and mathematical solution efficiency of the model are considered, but also the global optimization scheme considering the construction and loss cost can be output, and through the systematic candidate set control and hot start strategy, the challenge of solution size explosion in the scale of hundreds of wind turbines is effectively addressed, and efficient and stable optimization of the super large offshore wind farm power collection system within an acceptable engineering time is realized.

[0087] The embodiment of the present application also provides a storage medium for storing a computer program, which is executed to perform at least the method described above.

[0088] ​​The embodiment of the present application further provides a control device, comprising a processor and a storage medium for storing a computer program; wherein the processor is used to execute the computer program and at least execute the method as described above.

[0089] The embodiment of the present application further provides a processor, which executes a computer program and at least executes the method as described above.

[0090] The storage medium can be implemented by any type of nonvolatile storage device, or a combination thereof. The nonvolatile storage device can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a ferromagnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc or a compact disc read-only memory (CD-ROM). The magnetic surface memory can be a disk memory or a tape memory. The storage medium described in the embodiment of the present application is intended to include, but is not limited to, these and any other suitable types of memory.

[0091] In several embodiments provided by the present application, it should be understood that the disclosed system and method can be implemented in other manners. The described device embodiments are merely schematic, for example, the division of the units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling between the components can be indirect coupling or direct coupling through some interface, device or unit, and can be electrical, mechanical or other forms.

[0092] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purposes of the embodiment solutions.

[0093] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be separately as a unit, or two or more units can be integrated in one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software function unit.

[0094] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program executes the steps of the above-mentioned method embodiments when executed; and the foregoing storage medium includes mobile storage equipment, read-only memory (ROM), random access memory (RAM), magnetic disc or optical disc and various storage program codes.

[0095] Alternatively, the integrated unit of the present application can be stored in a computer readable storage medium if it is realized in the form of a software function module and sold or used as an independent product. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The foregoing storage medium includes mobile storage equipment, ROM, RAM, magnetic disc or optical disc and various storage program codes.

[0096] The methods disclosed in the several method embodiments of the present application can be combined arbitrarily without conflict to obtain new method embodiments.

[0097] The features disclosed in the several product embodiments of the present application can be combined arbitrarily without conflict to obtain new product embodiments.

[0098] The features disclosed in the several method or device embodiments of the present application can be combined arbitrarily without conflict to obtain new method or device embodiments.

[0099] The above is a further detailed description of the present application in combination with specific preferred embodiments, and the specific implementation of the present application cannot be limited to these descriptions. For those skilled in the art, without departing from the concept of the present application, a number of equivalent substitutions or obvious modifications can be made, and the performance or use is the same, which should be regarded as belonging to the protection scope of the present application.

Claims

1. A method for loss pre-computation and large-scale solution acceleration of an offshore wind farm power collection system, characterized in that, The method comprises the following steps: S1, for the power collection system planning problem of offshore wind farm, an initial mathematical model considering cable construction cost and life cycle loss cost is constructed; S2, the nonlinear loss term in the initial mathematical model is converted into a linear cost coefficient by solving the network loss precalculation subproblem, and a mixed integer linear programming model is obtained; S3, based on the position information of wind turbines and substations, a strategy combining clustering partition and intra-sector routing is adopted to construct a hot start initial feasible solution that meets the engineering constraints; The step S3 of constructing the hot start initial feasible solution comprises: taking the substation as the reference point, dividing the wind turbine nodes into several fan-shaped regions; in each fan-shaped region, a monotonic connection method based on fixed direction is adopted to generate an initial topology, specifically: the projection coordinates of each wind turbine node in the region in the selected connection direction and its orthogonal direction are calculated, and the nodes are sorted in the lexicographic order of the projection coordinates, and the sorted nodes are connected in turn to form a monotonic path; according to the number of wind turbines carried by each line segment in the initial topology, the cable model that meets the capacity constraint and has the optimal cost is matched for the line segment, and the total cost of the scheme is calculated; from the division modes taking different wind turbine nodes as the scanning starting points, the scheme with the lowest total cost is selected as the hot start initial feasible solution; the initial topology generated by the intra-sector routing strategy meets the engineering constraints of the offshore wind farm power collection system, and the engineering constraints at least include: each line connection meets the node degree limit of "one in and one out" or "two in and one out", and the lines do not cross in spatial layout; S4, based on a preset distance threshold, potential connections between wind turbines are screened, and connections exceeding the threshold in the hot start initial feasible solution are supplemented to a candidate set to construct a final candidate cable set for solving the mixed integer linear programming model; S5, based on the final candidate cable set and the hot start initial feasible solution, the mixed integer linear programming model is optimized and solved by using a solver, and a power collection system planning scheme is output.

2. Offshore wind farm power collection system loss pre-computation and large-scale solution acceleration method according to claim 1, characterized in that, The initial mathematical model constructed in step S1 is a mixed integer quadratic programming model or a mixed integer nonlinear programming model, and the objective function includes a network loss cost term proportional to the square of the line transmission power, and the power square term is determined by the current-power relationship under the DC power flow assumption.

3. The offshore wind farm power collection system loss precomputation and large-scale solution acceleration method according to claim 1, characterized in that, The network loss precalculation subproblem solving in step S2 comprises: For line segments carrying different numbers of wind turbines, the unit length total cost of all available cable models is precalculated and compared; The unit length total cost is composed of the unit length construction cost and the life cycle network loss cost linearization term considering discount, wherein the network loss cost linearization term is calculated based on the square relationship between the DC power flow assumption and the cable transmission power; The cable model with the minimum unit length total cost and its cost value are selected as the optimal cost coefficient corresponding to the number of wind turbines carried, and are integrated into the objective function of the main optimization model, so that the nonlinear programming problem containing the power square term is converted into a linear programming problem.

4. The offshore wind farm power collection system loss precomputation and large-scale solution acceleration method according to claim 1, characterized in that, The solution process of the network loss pre-computation sub-problem is independent of the iteration process of the main optimization model, and is a preprocessing step completed once before optimization solving.

5. The offshore wind farm power collection system loss precomputation and large-scale solution acceleration method according to claim 1, characterized in that, In step S3, the substation is taken as a reference point, and the fan nodes are divided into several sector-shaped areas according to the preset incoming line degree and sector capacity; The division mode with different fan nodes as the scanning starting point of the sector is traversed, and the scheme with the lowest total cost is selected as the hot start initial feasible solution.

6. The offshore wind farm power collection system loss precomputation and large-scale solution acceleration method according to claim 5, characterized in that, The fixed direction-based monotone connection method specifically includes: Selecting a connection direction and its unit orthogonal vector in the sector; Calculating the signed projection of the position vector of each fan node in the sector on the connection direction and its orthogonal vector to obtain a set of two-dimensional projection coordinates; Sorting all fan nodes according to the lexicographic order of the projection coordinates; Connecting the sorted fan node sequence in turn to form a path with monotonically non-decreasing coordinates in the selected connection direction, which naturally satisfies the degree constraints of cable non-intersection and node "one-in-one-out".

7. The offshore wind farm power collection system loss precomputation and large-scale solution acceleration method according to claim 1, characterized in that, The construction of the final candidate cable set in step S4 includes: Setting a maximum candidate distance threshold, and only connecting arcs between fans and between fans and substations with a Euclidean distance less than the threshold are included in the initial candidate set; All line arcs in the hot start initial feasible solution generated in step S3 with a connection length exceeding the maximum candidate distance threshold are supplemented into the initial candidate set; The initial candidate set and the supplemented line arcs together constitute the final candidate cable set.

8. The offshore wind farm power collection system loss precomputation and large-scale solution acceleration method according to claim 1, characterized in that, In step S5, when the solver is used for optimization solving, the hot start initial feasible solution is taken as the initial input of the branch and bound process of the mixed integer linear programming solver to accelerate convergence.

9. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the offshore wind farm power collection system loss pre-computation and large-scale solution acceleration method according to any one of claims 1 to 8. The computer program, when executed by a processor, implements the offshore wind farm power collection system loss pre-computation and large-scale solution acceleration method according to any one of claims 1 to 8.

Images