Optimization method of power collection system for offshore wind farm considering cable selection

Abstract

A kind of offshore wind farm power collection system optimization method considering cable selection, comprising the following steps: S1, obtaining wind farm basic data, including wind turbine and substation coordinates, wind turbine rated power, multi-specification cable electrical and cost parameters and the like;S2, based on basic data, candidate cable set is generated using clustering partition and distance threshold, candidate set reduction is completed, and cable cross-avoidance constraint set is simultaneously generated;S3, an optimization model with minimum total cost in the whole life cycle is established, double-ring path does not need scene variable, composite ring path needs to integrate scene-driven variable;S4, path acceleration solution is carried out, double-ring uses hot start, cross-constraint simplification, composite ring additionally adds fault set dimension reduction, and initial scheme is obtained;S5, full fault set N-1 check is carried out on initial scheme, illegal scene is included in fault set and is resolved again, until scheme meets N-1 criterion, and finally result is output.The present application realizes integrated optimization of topology and cable selection, and gives consideration to solving quality and efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to offshore wind farm technology, in particular to an offshore wind farm power collection system optimization method considering cable selection. BACKGROUND

[0002] In recent years, driven by the global renewable energy development goal, offshore wind power has become an important direction of energy transformation due to its abundant resources, gradually mature technology, and huge development potential. With the continuous expansion of installed capacity and the gradual extension of wind farm development to deeper waters, the power collection system planning of offshore wind farms will face increasingly high reliability requirements.

[0003] As the key link connecting wind turbines and offshore substations, the investment cost of the power collection system accounts for about 15%-30% of the total construction cost of the wind farm, and the rationality of planning and design directly affects the economy and stability of the system. However, this type of optimization problem has NP-hard and non-convex characteristics, and as the scale of the wind farm expands, the problem complexity rises sharply, making it difficult to obtain a global optimal solution. At the same time, the harsh environmental conditions at sea also result in much higher operation and maintenance costs than onshore wind power. Once the power collection system fails, it may cause serious economic and social consequences. Therefore, how to achieve a reasonable balance between economy and reliability is a problem that needs to be solved urgently.

[0004] The power collection system of an offshore wind farm needs to economically and stably send the dispersed wind turbine power to the offshore substation and collect it for external transmission under the premise of meeting reliability constraints. In addition to the topology structure, the decision of cable selection (cross-section / size / carrying capacity level, etc.) has a decisive impact on the life cycle cost and operability under fault: on the one hand, a larger cross-section can reduce line loss and thermal constraint risk, but will significantly increase the initial investment; on the other hand, a smaller cross-section, although it can reduce construction costs, may trigger a carrying capacity overrun or voltage overrun during fault current redistribution, making the scheme infeasible under the N-1 scenario. Therefore, selecting the appropriate size on each cable, while meeting the N-1 safety criteria, and optimizing the topology-cable selection in coordination, to obtain a high-quality planning scheme that balances economy and reliability, has become a key problem that needs to be solved in engineering practice.

[0005] From the point of view of computational complexity, the optimization problem of the power collection system with cable selection belongs to strong NP difficulty in essence: in addition to the topology binary variables, each candidate edge also introduces multi-specification discrete selection, and the variable scale and the constraint number rapidly increase with the "candidate edge number x specification number x scene number"; if the power flow and capacity constraints of all single-cable fault scenarios are required to be met at the same time, and the engineering implementability constraints such as geometric non-intersection and fan node degree are superimposed, the difficulty and time cost of direct solution will sharply increase with the scale of the station, which becomes a bottleneck restricting industrial landing. Due to the limitations of model complexity and computational difficulty, the current research on power collection system planning considering cable selection is basically limited to relatively simple radial power collection system planning. There is no precedent for the research on the power collection system planning model that integrates cable selection and meets the N-1 criterion.

[0006] In summary, the prior art still has the following deficiencies:

[0007] (1) The research on the power collection system planning based on the N-1 criterion is still weak, and most of the research is still limited to heuristic methods or traditional double-sided ring structures, which cannot achieve higher economic goals;

[0008] (2) Considering cable selection will bring high computational complexity to the model, and currently there is no research that can achieve both N-1 criterion and cable selection;

[0009] (3) In the face of the combination of multiple specifications and multiple scenarios, there is a lack of systematic scale control and computational acceleration mechanism, and it is difficult to stably obtain high-quality solutions on large-scale wind farms.

[0010] It should be noted that the information disclosed in the above background technology 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

[0011] The main purpose of the present application is to overcome the defects in the background art, and to provide a method for optimizing the power collection system of an offshore wind farm considering cable selection.

[0012] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0013] A method for optimizing the power collection system of an offshore wind farm considering cable selection, comprising the following steps:

[0014] S1, obtaining wind farm basic data, including the spatial coordinates of the fan nodes and the substation nodes, the rated power of the fan, the electrical parameters and cost parameters of multiple cable specifications, the system life cycle parameters and the electricity price information;

[0015] S2, based on the basic data, a candidate cable set reduction strategy based on clustering partition and distance threshold is adopted to generate a candidate cable set, and a cable crossing avoidance constraint pair set is identified and generated synchronously, and the model size and the number of crossing constraints are controlled through the candidate set reduction;

[0016] S3, a scene set containing normal scenes and fault scenes is constructed, and an optimization model with the minimum total life cycle cost as the target is established, and the model integrates path set variables and constraints: if it is a double-ring path, cable model selection variables and construction state variables are integrated, power balance constraints based on direct current flow are introduced, ring topology constraints suitable for cable selection are introduced, node degree constraints and cable crossing avoidance constraints are introduced; if it is an N-1 composite ring path, on the basis of the double-ring path, scene-driven operation state variables are additionally integrated, and the full fault set constraint is described through the scene dimension;

[0017] S4, an acceleration solving strategy corresponding to the path is adopted: for the double-ring path, two strategies of hot start initialization based on pre-solution scheme and cable crossing constraint simplification based on incremental loading are adopted; for the N-1 composite ring path, on the basis of the two strategies of the double-ring path, a fault set dimension reduction strategy based on key fault scene screening is additionally adopted; through the strategy acceleration solving, an initial topology and cable selection scheme are obtained;

[0018] S5, the initial scheme is subjected to full fault set N-1 safety verification, and the violation scene is included in the fault set re-solution, wherein the N-1 composite ring path iteratively expands the fault set, and the double-ring path verifies the selection compliance, until the scheme meets the N-1 safety criterion, and finally the topology planning and cable selection result are output.

[0019] Further, in step S3, the cable model selection constraint requires that at most one cable model can be selected for any candidate cable position; if it is an N-1 composite ring path, the scene-driven operation state variable constraint requires that the value of the operation state variable of any candidate cable and any scene is constrained between the sum of all model construction state variables and zero; for any scene predefined as a fault scene, the operation state variable of the corresponding specific cable is forced to be zero; if it is a double-ring path, the scene-driven operation state variable constraint is not required.

[0020] Further, for the double-loop path in step S3, the loop topology constraint of the adaptive cable selection is implemented by introducing binary variables representing the virtual power flow direction and continuous variables representing the node accumulated power, which includes: ensuring that the virtual power flow into and out of each wind turbine node is 1 to form a loop structure; establishing a bidirectional virtual power flow accumulation equation to calculate the accumulated power value of each node in both positive and negative directions; based on the accumulated power value, constructing a constraint condition for each cable that its carrying power needs to meet the N-1 safety criterion, which uses different expressions for cables connecting substations and internal cables of wind farms; the N-1 composite loop path does not require this loop topology constraint.

[0021] Further, for the double-loop path, the specific implementation process of the loop topology constraint of the adaptive cable selection includes:

[0022] Introduce a set of first binary variables to represent the positive direction of virtual power flow on the candidate cable;

[0023] Introduce a set of second binary variables to represent the negative direction of virtual power flow on the candidate cable;

[0024] Constrain the sum of the first binary variables and the second binary variables to be equal to the construction state variable of the cable, to ensure that virtual power flow only exists on the constructed cable, and only one direction of virtual power flow can exist on each cable;

[0025] For each wind turbine node, constrain the sum of all first binary variables on its incident cables and the sum of all second binary variables on its outgoing cables to be equal to 1, and constrain the sum of all first binary variables on its outgoing cables and the sum of all second binary variables on its incident cables to be equal to 1, to force the formation of a closed-loop loop topology structure;

[0026] Introduce a set of first continuous variables to represent the sum of the accumulated downstream wind turbine rated power of each node on the assumed virtual positive power flow path;

[0027] Introduce a set of second continuous variables to represent the sum of the accumulated downstream wind turbine rated power of each node on the assumed virtual negative power flow path;

[0028] Establish a recursive relationship constraint for the first continuous variables and the second continuous variables, so that their values increase along the virtual power flow direction, and the increment is the rated power of the wind turbine node passed through;

[0029] Constrain the lower bound of the first continuous variables and the second continuous variables to be the rated power of the wind turbine itself, and the upper bound to be the capacity of the largest type of cable;

[0030] Based on the first continuous variable and the second continuous variable, a lower limit constraint of the power that each cable needs to carry is constructed: for the cable directly connected to the substation, the power capacity it needs to carry must be no less than the maximum of the first continuous variable and the second continuous variable at the wind turbine node connected to it; for the cable connected to two wind turbine nodes, the power capacity it needs to carry must be no less than the maximum of the first continuous variable and the second continuous variable at the node with the larger maximum, minus a wind turbine rated power.

[0031] The above power capacity requirement constraint is associated with the nominal carrying capacity of the selected cable type, to ensure that the capacity of the selected cable meets the power transmission requirements under all possible fault scenarios.

[0032] Further, the first continuous variable and the second continuous variable are further constrained to be integer multiples of the wind turbine rated power, to conform to the physical reality of discrete power accumulation and improve the efficiency of model solving.

[0033] Further, the power capacity requirement constraint constructed for the cable connected to two wind turbine nodes is linearized by introducing a large enough constant and combining the big M method, to ensure that the constraint is only activated when the cable is actually selected and constructed.

[0034] Further, in step S4, for the N-1 composite loop path, the fault set dimension reduction strategy based on the key fault scenarios includes: in the initial iteration, if it is a single substation scenario, the cable fault scenarios directly connected to the substation node are preferentially included; if it is a multi-substation scenario, there is no need to initially include specific cable fault scenarios; in subsequent iterations, the system overload risk after all candidate cable faults is calculated and sorted based on the line outage distribution factor (LODF), or the fault severity is evaluated by solving a direct current optimization sub-problem with the objective of minimizing the total system overload, and the scenarios with the highest risk are selected to join the main problem fault set, and the iteration is performed until there is no overload under all single cable fault scenarios.

[0035] Further, the cable cross constraint simplification strategy based on incremental loading in step S4 includes: in the solving process, only the cross pairs that may be involved by the cables connected to the substation nodes are initially included in the constraint set; after obtaining the initial scheme, it is checked whether there is an un-constrained cable cross situation; if there is, the corresponding cross constraint is added to the model and re-solved, and through this iterative approach, a feasible solution that meets all cross avoidance requirements is gradually approached.

[0036] Further, the hot start initialization strategy in step S4 includes: if it is a double-loop path, first solve a simplified model without considering cable selection or a relatively fixed topology, and take the optimal solution as the initial solution of the complex optimization model considering cable selection; if it is an N-1 composite loop path, first solve the double-loop optimization model considering cable selection, and take the optimal solution as the initial solution of the N-1 composite loop optimization model considering cable selection; thereby accelerating the solving process.

[0037] A computer program product comprising a computer program which, when executed by a processor, implements the offshore wind farm power collection system optimization method.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] In view of the obvious limitations of the existing offshore wind farm power collection system planning methods: most methods carry out planning within the framework of pre-defined topology (such as double-loop, multi-loop), or use heuristic, meta-heuristic algorithms such as "angle scanning clustering / CMST / CVRP" to determine the topology before capacity configuration, which has good solving efficiency but lacks global optimality guarantee; some methods use mathematical optimization to improve the solution quality in the double-loop topology, but do not systematically couple cable model selection with the N-1 safety criterion in all scenarios, making it difficult to balance reliability and economy; the present application proposes an offshore wind farm power collection system optimization method considering cable selection, the core of which is to construct a scenario-driven planning model based on the N-1 safety criterion: by introducing cable type binary variables, modeling of multi-specification cable selection is realized, and key parameters such as cable cost, resistance, and carrying capacity are uniformly included in the objective function and constraint system (such as constructing carrying capacity constraints for adaptive fault scenarios), and the topology structure and cable model selection of the power collection system are jointly optimized; the model includes two specific cases of "double-loop N-1 with selection" and "N-1 composite loop with selection", and can still achieve a high-quality planning scheme that minimizes the life cycle cost (including construction, operation and maintenance, network loss, and abandoned wind cost) even in the face of the explosive growth of "candidate edges x model x fault scenarios" dimensions under the premise of meeting the N-1 safety criterion.

[0040] In view of the high calculation complexity of the two cable type selection models, in order to adapt to the demand of actual engineering for solving efficiency and the expansibility of the model to larger scale examples, a structured acceleration solving strategy is further provided to better balance the solving efficiency and quality. Specifically, the candidate cable set is reduced by clustering partition and distance screening, and the variable scale is reduced; the fault set is reduced based on key fault scenarios (such as cable fault directly connected with the substation in the single substation scenario and high-risk fault selected by LODF sorting), and the scene calculation amount is reduced; a simplified model (such as a double-sided ring optimization model) is solved first without considering cable type selection or relatively fixed topology structure, and the optimal solution is used as the warm start initial solution to accelerate the model convergence; the cable cross constraint is processed in an incremental loading manner to simplify the constraint system.

[0041] The important innovation points of the present application are embodied in two aspects: one is to build an integrated joint optimization topology and cable type selection framework, breaking the traditional "topology first and then type selection" or "only topology optimization" limitation; the other is to design an efficient acceleration strategy suitable for the main problem, which solves the solving problem in large scale and multiple scenarios.

[0042] Compared with the existing method, the present application has three major advantages: first, high-quality solution is achieved - an optimization method is designed instead of a heuristic algorithm, which can pursue the global optimal solution, and the example comparison results show that the economic and reliability advantages are significant; second, strong expansibility - the model can be naturally extended to cable type selection, multiple substations, large-scale wind farms and other scenarios, and is suitable for different engineering needs; third, balancing solving quality and efficiency - through the cooperation of the core framework and the supporting acceleration strategy, the high quality of the planning scheme is guaranteed, and the solving difficulty and time cost are effectively reduced.

[0043] In summary, the present application ultimately achieves three key goals: first, the cable type selection variable deeply participates in the model construction, realizing the coordinated optimization of topology and type selection; second, the constraint variable is established based on the scene, and the planning model meeting the N-1 safety criterion is constructed; third, an efficient solving method is formed which meets the characteristics of the acceleration strategy, providing reliable technical support for the optimization planning of the offshore wind farm power collection system.

[0044] Other beneficial effects in the embodiments of the present application will be further described in the following. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 The power accumulation diagram for improving the CVRP model in the embodiments of the present application is shown;

[0046] Figure 2 The vector cross product judgment diagram for assisting in judging the cable geometric intersection relationship in the embodiments of the present application is shown;

[0047] Figure 3Flow chart of offshore wind farm power collection system topology planning scheme of the embodiment of the present application;

[0048] Figure 4 Single booster station offshore wind farm wind turbine, substation coordinate and candidate sea cable schematic diagram;

[0049] Figure 5 Sweep+CWS topology planning result chart of Comparative Example 1;

[0050] Figure 6 Double-sided ring topology planning result chart of Comparative Example 2;

[0051] Figure 7 Multi-ring structure topology planning result chart of Comparative Example 3;

[0052] Figure 8 Double-sided ring with selection topology planning result chart of Embodiment 1 of the present application;

[0053] Figure 9 N-1 composite ring with selection topology planning result chart of Embodiment 2 of the present application;

[0054] Figure 10 Double booster station Sheringham Shoal offshore wind farm, substation coordinate and candidate sea cable schematic diagram;

[0055] Figure 11 Double-sided ring topology planning result chart of Comparative Example 4;

[0056] Figure 12 CSI structure topology planning result chart of Comparative Example 5;

[0057] Figure 13 Selection double-sided ring topology planning result chart of Embodiment 3 of the present application;

[0058] Figure 14 Selection N-1 composite ring topology planning result chart of Embodiment 4 of the present application;

[0059] Figure 15 Solution time convergence curve chart of the acceleration solution strategy of the present application;

[0060] Figure 16 General flow chart of offshore wind farm power collection system optimization method considering cable selection. DETAILED DESCRIPTION

[0061] The embodiments of the present application are described in detail below. It should be emphasized that the following description is merely exemplary, and is not intended to limit the scope of the present application and its applications.

[0062] In addition, the terms "first", "second", "third", etc. are used only for descriptive purposes and are not to be construed as indicating or implying relative importance or an ordered ranking of the indicated technical features. Thus, features defined with "first", "second" or "third" can explicitly or implicitly include one or more of such features. In the description of embodiments of the present application, the meaning of "a plurality" is two or more, unless explicitly specified otherwise.

[0063] The present application aims to solve the problem that the cable model selection is not coupled with the full-scenario N-1 safety criterion in the existing offshore wind farm power collection system planning, the calculation complexity is high in multiple specifications and multiple scenarios, and there is a lack of global optimality. The present application proposes a scenario-driven planning model based on the N-1 safety criterion to realize the integrated optimization of topology and cable selection, which is specifically divided into two relatively independent paths: the double-ring path does not need to expand the scenario dimension and the scenario-driven variable, and the N-1 criterion is naturally satisfied by adapting the ring topology constraint of cable selection. The model only integrates cable model selection, construction state variables and basic constraints; the N-1 composite ring path needs to depict the full fault set through the scenario dimension, and the model additionally integrates scenario-driven operating state variables. In order to adapt to the two paths and control the calculation complexity, the present application proposes a step-by-step and path-divided acceleration strategy: based on clustering partition, the candidate cable set is reduced (the cross constraint set size is synchronized to affect the cable, which is completed in the candidate cable set generation stage, laying a foundation for subsequent model solving), the double-ring path adopts hot start initialization based on pre-solution scheme and cable cross constraint simplification based on incremental loading in the solving stage, and the N-1 composite ring path additionally adopts fault set dimension reduction strategy based on key fault scenario screening on this basis. Compared with the existing method, the present application can not only pursue global optimality to realize high-quality solving, but also can be expanded to multiple substations and large-scale scenarios, and the solving quality and efficiency are considered.

[0064] Referring to Figure 16 The embodiment of the present application provides an offshore wind farm power collection system optimization method considering cable selection, which comprises the following steps:

[0065] Step S1, obtaining wind farm basic data, including spatial coordinates of wind turbine nodes and substation nodes, wind turbine rated power, electrical parameters and cost parameters of multiple cable specifications, system life cycle parameters and electricity price information.

[0066] Step S2, based on the basic data, a candidate cable set reduction strategy based on clustering partition and distance threshold is used to generate a candidate cable set, and a cable cross avoidance constraint set is simultaneously identified and generated, so as to control the model size and the number of cross constraints through candidate set reduction.

[0067] In some embodiments, the generation of the candidate cable set in step S2 employs a screening method based on spatial distance and clustering analysis, specifically including: for large-scale wind farm scenarios, first, a clustering algorithm is employed to divide the wind farm area into several sub-zones, and only node pairs located in the same sub-zone are retained as potential candidate cables; then, the Euclidean distance between each pair of nodes corresponding to the potential candidate cables is calculated based on the coordinate information of all nodes, and a distance threshold is set to filter out potential cable connections with a length exceeding the threshold; for non-large-scale wind farm scenarios, candidate cables can be directly screened by calculating the Euclidean distance between nodes and combining the distance threshold, without the need for a clustering sub-zone step; thereby effectively controlling the problem size while maintaining optimization potential.

[0068] Step S3, constructing a scenario set containing normal scenarios and fault scenarios, establishing an optimization model with the minimum total life cycle cost as the target, the model integrating path set variables and constraints: if it is a double-ring path: integrating cable model selection variables, construction state variables, introducing power balance constraints based on direct current flow, ring topology constraints adapted to cable selection, node degree constraints and cable intersection avoidance constraints (operation variables not driven by scenarios); if it is an N-1 composite ring path: on the basis of the double-ring path, additionally integrating scenario-driven operation state variables, and describing the full-fault set constraint through the scenario dimension.

[0069] In some embodiments, the objective function of the optimization model in step S3 is constructed by: cable construction cost, calculated by adding the construction cost of each candidate cable under different models and the product of the corresponding binary selection decision variable; system operation and maintenance cost, calculated by multiplying the cable construction cost by a fixed proportion coefficient, and then uniformly distributing it to the entire system life cycle through the annuity discount formula; system loss cost, based on the direct current flow assumption, the resistance value of each cable under its selected model is multiplied by the square value of the transmission power, and the annual network energy cost is obtained, and then it is extended to the full life cycle through discount calculation; wind curtailment cost, by introducing a large enough penalty coefficient, multiplied by the sum of wind curtailment power of all nodes in all scenarios, to ensure that the wind curtailment amount is zero in the process of minimizing the objective function.

[0070] In some embodiments, the cable model selection constraint in step S3 requires: at most only one cable model can be selected for any candidate cable location; the scenario-driven operation state variable constraint requires: for any candidate cable and any scenario, the value of the operation state variable is constrained between the sum of all model construction state variables and zero; for any scenario predefined as a fault scenario, the operation state variable of the corresponding specific cable is forced to be zero.

[0071] In some embodiments, the power balance constraint based on direct current power flow in step S3 is established by the following procedure: for each wind farm node and each scenario, its power balance equation constrains the sum of all cable power flowing out of the node to be equal to the difference between its wind power and possible curtailed wind power; for each substation node and each scenario, its power balance equation constrains the sum of all cable power flowing into the node to be equal to its power collected to the grid; for each candidate cable and each scenario, its direct current power flow equation is established by introducing node voltage phase angle variables and based on the susceptance value of the selected cable type, while coupling this equation with the cable's selected type and operating state variables using the big M method; the voltage phase angle of any one substation node in the system is set to zero as the reference phase of the entire network; for each candidate cable and each scenario, the absolute value of its transmission power is constrained not to exceed the product of its selected type's current carrying capacity and operating state variable; for each wind farm node and each scenario, its curtailed wind power is constrained between zero and the node's wind power.

[0072] In some embodiments, the ring topology constraint of the adapted cable type selection in step S3 is achieved by introducing binary variables representing virtual power flow direction and continuous variables representing node accumulated power, and the constraint includes: ensuring that the number of virtual power flow flowing into and out of each wind farm node is 1 to form a ring structure; establishing a bidirectional virtual power flow accumulation equation to calculate the accumulated power value of each node in both positive and negative directions; based on the accumulated power value, for each cable, construct its constraint condition that the power it carries needs to meet the N-1 safety criterion, which uses different expressions for cables connecting substations and internal cables of wind farms.

[0073] In further preferred embodiments, the specific implementation process of the ring topology constraint for cable type selection includes: introducing a set of first binary variables to represent the forward direction of virtual power flow on candidate cables; introducing a set of second binary variables to represent the reverse direction of virtual power flow on candidate cables; constraining the sum of the first and second binary variables to be equal to the construction state variable of the cable to ensure that virtual power flow only exists on constructed cables and only one direction of virtual power flow exists on each cable; for each wind turbine node, constraining the sum of the first binary variables on all incident cables and the sum of the second binary variables on all outgoing cables to be equal to 1, and constraining the sum of the first binary variables on all outgoing cables and the sum of the second binary variables on all incident cables to be equal to 1, to force the formation of a closed-loop ring topology; introducing a set of first continuous variables to represent the sum of downstream wind turbine rated powers accumulated by each node on the assumed virtual forward power flow path; introducing a set of second continuous variables to represent the sum of downstream wind turbine rated powers accumulated by each node on the assumed virtual reverse power flow path; establishing recursive relationship constraints for the first and second continuous variables, such that their values increase along the direction of virtual power flow, with the increment being the rated power of the wind turbine node passed through; constraining the lower bound of the values of the first and second continuous variables to be the rated power of the wind turbine itself and the upper bound to be the capacity of the largest type of cable; based on the first and second continuous variables, constructing a power lower bound requirement constraint for each cable: for a cable directly connected to a substation, the power capacity it needs to carry must be no less than the maximum of the first and second continuous variables at the wind turbine node connected to it; for a cable connecting two wind turbine nodes, the power capacity it needs to carry must be no less than the maximum of the first and second continuous variables of the node with the larger maximum value, minus a wind turbine rated power; associating the above power capacity requirement constraint with the nominal current-carrying capacity of the selected type of cable to ensure that the capacity of the selected cable meets the power transmission requirements under all possible fault scenarios.

[0074] In further preferred embodiments, the first and second continuous variables are further constrained to be integer multiples of wind turbine rated power, to conform to the physical reality of discrete power accumulation and improve model solving efficiency.

[0075] In further preferred embodiments, the power capacity requirement constraint constructed for a cable connecting two wind turbine nodes is linearized by introducing a large enough constant and combining the Big-M method to ensure that the constraint is only activated when the cable is actually constructed.

[0076] Step S4, adopt acceleration solving strategy corresponding to the path: for the double-loop path, adopt two strategies of hot start initialization based on pre-solution scheme and cable cross constraint simplification based on incremental loading; for the N-1 composite loop path, in addition to the two strategies of the double-loop path, a fault set dimension reduction strategy based on key fault scenario screening is additionally adopted (because the scenario variable needs to be processed); through the strategy acceleration solving, the initial topology and cable selection scheme are obtained.

[0077] In some embodiments, the fault set dimension reduction strategy based on key fault scenario screening in step S4 includes: in the initial iteration, if it is a single substation scenario, preferentially include the cable fault scenarios directly connected with the substation node; if it is a multi-substation scenario, there is no need to initially include specific cable fault scenarios; in the subsequent iteration, based on the line outage distribution factor (LODF), calculate and sort the system overload risk after all candidate cable faults, or evaluate the fault severity by solving a direct current optimization sub-problem with the objective of minimizing the total system overload, and select a number of scenarios with the highest risk to join the main problem fault set, and iterate until there is no overload under all single cable fault scenarios.

[0078] In some embodiments, the cable cross constraint simplification strategy based on incremental loading in step S4 includes: in the solving process, initially only include the cross pairs that may be involved by the cables connected with the substation node in the constraint set; after obtaining the initial scheme, check whether there is a cable cross situation that is not constrained; if there is, add the corresponding cross constraint to the model and re-solve, and gradually approach the feasible solution that meets all cross avoidance requirements through this iterative method.

[0079] In some embodiments, the hot start initialization strategy in step S4 includes: if it is a double-loop path, first solve a simplified model (such as a double-sided loop topology model, without considering cable selection) that does not consider cable selection or relatively fixed topology, and use the optimal solution as the initial solution of the complex optimization model considering cable selection; if it is an N-1 composite loop path, first solve a double-loop optimization model considering cable selection (as a simplified model), and use the optimal solution as the initial solution of the N-1 composite loop optimization model considering cable selection; thereby accelerating the solving process.

[0080] Step S5, perform N-1 safety check on the initial scheme for the full fault set, and re-solve the fault set for the violation scenarios, wherein only the N-1 composite loop path needs to iteratively expand the fault set, and the double-loop path naturally meets the N-1 due to the topology characteristics, and the check focuses on the compliance of the selection, until the scheme meets the N-1 safety criteria, and the final topology planning and cable selection result is output.

[0081] The offshore wind farm power collection system optimization method considering cable selection provided by the application successfully solves the problem of the coordinated optimization of the topology structure and cable selection of the offshore wind farm power collection system under the premise of meeting the N-1 safety criterion by constructing an optimization model integrating the cable type selection variable, the construction state variable and the scenario-driven operation state variable. The method creatively unifies the multi-specification parameters such as cable cost, resistance, current-carrying capacity and the like into the planning framework, breaks through the limitation of the existing researches which are mostly limited to predefined topology or heuristic methods, and realizes high-quality solution under the pursuit of global optimality. In the face of the explosive growth of the calculation complexity caused by the “candidate edge x type x fault scenario”, the application is matched with systematic acceleration strategies including candidate set reduction, hot start initialization, key fault scenario dimension reduction and incremental constraint loading, which significantly reduces the solution difficulty and time cost, enables the model to have strong expansion capability for processing large-scale wind farms, multiple substations and multi-specification cable scenarios, and finally significantly improves the economy and engineering feasibility of the planning scheme while ensuring high reliability.

[0082] The specific embodiments of the application, algorithm examples and experimental verification thereof are further described below.

[0083] A kind of offshore wind farm power collection system optimization method considering cable selection, constructs the double-sided ring planning model considering cable selection as follows:

[0084] Objective function:

[0085] Suppose a complete weighted directed graph , wherein including transformer node And fan node , Indicate all candidate cables connecting nodes, the cost of each candidate cable In type c is Binary decision variable representing whether candidate cable Is built, indicates that candidate cable Is built when Meanwhile, set the resistance of each candidate cable In type c as The power flow on each cable is represented by variable . Set the wind abandonment amount as , to ensure that the power collection system meets the “N-1” safety constraint, multiply the cost by a maximum number , such as To ensure that the wind abandonment cost item in the objective function is 0. The objective function of the economic planning model of the power collection system meeting the “N-1” safety constraint is constructed as follows:

[0086] (1)

[0087] wherein , , , respectively are the cable construction cost, system operation and maintenance cost, system network loss cost and wind curtailment cost of the power collection system, which can be expressed as:

[0088]

[0089] wherein the construction cost of the cable is the cost of each type of candidate cable multiplied by the sum of the construction decision variables; the operation and maintenance cost of the system can be regarded as the construction cost multiplied by a fixed coefficient; in the system network loss cost, represents the network loss cost of the wind farm in the whole life cycle, is the annual inflation rate. Meanwhile, in the model of the present application, the direct current flow assumption is adopted, and the unit value of each node voltage is 1, so that , therefore the network loss cost of each cable after considering the cable type selection is written as . The wind curtailment cost takes M as a maximum number to ensure that the value of this item is 0.

[0090] Cable type selection constraint:

[0091] Considering the cable type selection, it is required that the cable in each position can only have one type selected at most, and the related constraint is:

[0092] (7)

[0093] Power balance constraint based on direct current flow:

[0094] The model adopts the following form of power transmission planning model based on direct current flow:

[0095] (8)

[0096] (9)

[0097] (10)

[0098] (11)

[0099] (12)

[0100] Constraint (8) is the power balance constraint of i point, which represents the power balance equation of each wind turbine node i; constraint (9) is the power balance constraint of the transformer substation node j; constraint (10) represents the power balance constraint of the cable The constraint (11) is a reference node constraint of the power flow model for the substation node, and the constraint (12) limits the abandoned wind power of each wind turbine node i;

[0101] Improved CVRP ring constraint:

[0102] The traditional double-sided ring power collection system planning often adopts a modeling method based on a CVRP model, and the application improves the same to adapt to the cable selection:

[0103]

[0104] In the constraint (13), and are binary variables representing virtual power flow, wherein represents that the virtual power flows from the node i (j) to the node j (i), and the unidirectional power flow can only exist on the cable that has been constructed; the constraints (14) and (15) limit the inflow and outflow of each wind turbine node to 1, so that the topology is a ring; the constraints (16)-(19) are extensions of the MTZ specific sub-loop elimination constraint, and the variables respectively represent the cumulative power of the node i when the forward (reverse) power flow is considered, and the constraints (20)-(21) limit the value range of .

[0105] Figure 1 A CVRP model example containing bidirectional power accumulation variables is shown. In the example, the path Sub→1→2→3→4→Sub represents a virtual power flow path connecting wind turbines 1, 2, 3 and 4 in the double-sided ring. Under the constraints (16)-(17), the forward accumulated power along the virtual power flow is increased from the of the node 1 to the 4 of the node 4, and correspondingly, the reverse accumulated power is accumulated from the of the node 4 to the 4 of the node 1. The bidirectional accumulated power flow can help better establish the power constraint that meets the N-1 criterion. The following formula gives the number of wind turbines that the i-th cable needs to support under the N-1 criterion , wherein n represents the total number of cables in the ring network, represents the number of wind turbines that the i-th cable needs to support along the ring from the substation node:

[0106] (22)

[0107] ​​The power constraints under N-1 criterion can thus be expressed as follows:

[0108]

[0109] Combining Figure 1 Understanding constraints (23)-(24): for cables directly connected to the substation (e.g. Sub-1), the required current carrying capacity is represented by the maximum of the bi-directional accumulated power at the substation (max(P , )); for cables not connected to the substation (e.g. 1-2), the required current carrying capacity is represented by the maximum of the bi-directional accumulated power at the node topologically farthest from the substation (max(P , )). This forms two big-M constraints (23)-(24) that ensure that only the constructed cables are enforced. The left-hand term represents the maximum current carrying capacity of cable (i,j) for the selected model.

[0110] It is worth noting that the min function carried by the right-hand side of constraint (24) makes it not directly programmable. Given that the output of the wind turbines is often fixed and uniform in the planning of a wind power collection system, the maximum of the bi-directional accumulated power at the node topologically farthest from the substation (max(P , )) for cables not adjacent to the substation (e.g. 1-2) is necessarily smaller than the corresponding value at the node topologically closest to the substation (max(P , )) for the same cable.

[0111] Thus, constraint (24) can be expressed in a more programmable form as follows:

[0112]

[0113] Cable non-crossing and node degree constraints:

[0114] Similarly, the model captures the constraints of cable non-crossing and wind turbine node degree.

[0115] (26)

[0116] where cable pair (AB, CD) belongs to if and only if and as shown in Figure 2 .

[0117] For a ring topology, it is easy to see that the degree of each wind turbine node is equal to 2: ​

[0118] (27)

[0119] Based on the above, the bilateral ring model for cable selection meeting the N-1 criterion is established as follows:

[0120]

[0121] The N-1 composite ring planning model for cable selection considered by the application is as follows:

[0122] The objective function, cable selection constraints, DC power flow constraints, and even cable non-crossing and node degree constraints involved in this part can mostly be used in the related content of the bilateral ring model in terms of principle and form. Here, the parts different from the bilateral ring form are introduced:

[0123] Fault set constraints:

[0124] When modeling the "N-1" fault set, the following is used: to represent the set of all scenarios, and the binary variable is used to represent the candidate cable whether it is normally operated in the scenario t , which is different from the variable representing whether the candidate cable is constructed. represents that the cable t is normally operated in the scenario . The fault set constraint determined by this is:

[0125] (28)

[0126] (29)

[0127] Among them, constraint (28) represents that the value of is between 0 and , that is, the cable is normally operated as a necessary condition for being constructed; constraint (29) is the fault cable constraint, which represents that in the fault scenario , the corresponding preset fault cable cannot be normally operated.

[0128] DC power flow-based power balance constraint:

[0129] The following is used to introduce the DC power flow-based power transmission planning model in the form of scenario variables:

[0130]

[0131] Constraint (30) is the power balance constraint of point i, which represents the power balance equation of each wind turbine node i in each scenario; constraint (31) is the power balance constraint of the substation node j in each scenario; constraint (32) represents that the cable The DC power flow constraint considering phase angle, the constraint (33) is a reference node constraint of the power flow model for the substation node; the constraints (34) and (35) limit the transmission power of each candidate cable and the abandoned wind power of each wind turbine node i respectively;

[0132] Node degree constraint:

[0133]

[0134] In summary, the model of the N-1 composite ring considering cable selection meeting the N-1 criterion is established as follows:

[0135]

[0136] The two models considering cable selection have considerable calculation complexity. Considering the actual size of the solution efficiency of the actual project and the expandability of the proposed model to larger scale examples, the application further proposes the following accelerated solution strategy to better balance the solution efficiency and quality.

[0137] Candidate cable set reduction:

[0138] In the framework of the power collection system planning, the coordinates of the wind turbine and the substation are determined in the micro-siting stage and are input to the model. The candidate cables are generated from these spatial coordinates. At this time, the dimension of the decision variable and the constraint will directly increase with the size of the candidate cable set. It is necessary to use effective means to reasonably reduce the candidate cable set. When facing a larger scale wind farm, the application adopts the method of "partition + candidate cable screening".

[0139] Specifically, after obtaining the coordinate information of the wind farm, the partition to be divided and the number of wind turbines in each partition are confirmed. The Kmeans is used to obtain the centroid position of each partition. The distance matrix of each wind turbine to the centroid is calculated. The distance to the 90th percentile of the capacity limit from each cluster centroid is defined as the "core radius" of each cluster. The wind turbines within the core radius of each cluster are preferentially allocated to the cluster under the premise that the capacity meets the requirement. Then the wind turbines that have not been allocated are backfilled to the nearest cluster that still has excess capacity. Thus, the partition step is completed.

[0140] ​After partitioning, candidate cables are constructed within each cluster without considering inter-zone cables. Specifically, since the proposed method aims to obtain a high-quality power collection system planning scheme, the construction cost term in the objective function is formulated to minimize the Euclidean distance between wind turbines. Connections spanning over long distances are almost impossible to appear in the optimal layout, and thus can be removed by directional screening. The proposed scheme first sets a maximum candidate distance, and cables exceeding the maximum candidate distance are not added to the candidate cable set. If partitioning is not required for smaller-scale wind farms, the proposed method is directly used to construct the reduced cable set.

[0141] Hot-start acceleration:

[0142] In large-scale non-linear, non-convex optimization frameworks, constructing a high-quality initial solution plays a key role in accelerating model convergence and avoiding local optima. A good initial solution can reduce the number of iterations and effectively shorten the solution time. In the framework of the proposed method, a feasible solution is first obtained using the double-loop planning method without cable selection as a hot start for the double-loop planning method considering cable selection. Further, since the proposed double-loop scheme naturally satisfies the N-1 safety criterion, the solution obtained by the double-loop planning method considering cable selection can be used as a hot start for the N-1 composite loop considering cable selection.

[0143] Fault scenario dimension reduction:

[0144] In the planning model, the computational complexity is highly related to the number of fault scenarios. Accurate identification of the most critical emergency scenarios can significantly improve the solution efficiency. The cables close to the substation can be identified as critical fault handling using geometric features, but this principle will have limitations in scenarios considering cable selection. The line outage distribution factor (LODF) is used to calculate the post-fault power flow, and each candidate single-line outage is ranked according to the overload risk.

[0145] First, let be the selected cable set in the current main model planning result, and construct the reduced node admittance matrix by removing the reference bus , assign a fixed direction to each constructed cable , and define its associated vector , where is the standard basis vector of bus i, and let be its admittance. For candidate outage cable k connecting m and n, set . At this time, for the monitored cable a, define its power transfer distribution factor (PTDF) relative to as:

[0146] (37)

[0147] The corresponding line outage distribution factor (LODF) is:

[0148]

[0149] This means the change of power flow per unit on line a caused by removing branch k. Given the base power flow of the master problem , the post-contingency power flow on a under k outage is:

[0150] (39)

[0151] And since , naturally we have Under DC power flow model, these expressions are analytically exact. The post-contingency power flow calculated by LODF is consistent with the result of re-solving the power flow after removing line k, and only one sparse solution is needed each time.

[0152] For each candidate outage , calculate:

[0153] (40)

[0154] where is the transmission capacity of cable a, and then sort all candidate single-line outages in descending order of , and select the top n scenarios as the "critical fault scenarios" of this iteration. Add these critical fault scenarios to the fault scenario set of the master problem, and continue to solve the planning problem iteratively until each cable, single-line outage meets , which means there is no overload scenario, and the resulting planning result meets the N-1 criterion.

[0155] When the scenario contains multiple substations, the power collected at each substation node is a decision variable, and in this case the net injection power of the entire network is not fixed, and the LODF method cannot be used. At this time, for each fault scenario: , solve the following DC optimization subproblem to minimize the overload relaxation under the premise of not rejecting wind, where .

[0156]

[0157] Solving this problem can get the post-contingency power flow and the optimal relaxation , and the optimal objective value , at this time the outages are also sorted in descending order of , and the top n scenarios are added to the scenario set of the master problem, and iterated until all the built network meets , which indicates that no more slack is needed and the design has met the N-1 criterion.

[0158] Cable non-crossing constraint simplification:

[0159] Constraint (26) is effective in preventing cable pairs from crossing, but the large scale of the constraint will introduce significant complexity, which usually dominates in practical planning problems and severely limits the modeling and solving speed. Therefore, when a computational bottleneck is encountered, a feasible approach is to adopt a moderately relaxed For example, only cable pairs containing substation nodes are considered, and tests have shown that, under the planning framework of either a double-loop or an N-1 composite loop, the relaxed can avoid cable pair crossing with a high probability in practical planning processes, and thus become a sufficiently effective benchmark candidate set.

[0160] In the solving process, first, the crossing cable pairs involving substation connections are input to the solver as high-quality constraints. After obtaining the planning result, a crossing check is performed; when a crossing is detected, the constraint set is expanded and iterative refinement is performed. This incremental constraint loading mechanism enables the model to handle higher complexity planning tasks while maintaining the reliability of the constraints.

[0161] Therefore, the overall planning scheme flowchart is as shown in Figure 3 .

[0162] Experimental verification:

[0163] Experiment 1:

[0164] A 36-wind-turbine single-step-up substation wind farm is adopted, and the coordinates of the wind turbines and the substation, and the schematic diagram of the candidate submarine cable are as shown in Figure 4 .

[0165] This experiment compares a plurality of examples, as shown in Table 1.

[0166] Table 1

[0167]

[0168] Planning results: Figure 5 shows the sweep+CWS topology planning result of Comparative Example 1; Figure 6 shows the double-loop topology planning result of Comparative Example 2; Figure 7 shows the multi-loop structure topology planning result of Comparative Example 3; Figure 8 shows the double-loop topology planning result of Example 1 of the present application; Figure 9 shows the N-1 composite loop topology planning result of Example 2 of the present application.

[0169] The cost comparison results (unit: 100 million RMB) are shown in Table 2.

[0170] Table 2

[0171]

[0172] The experimental results show that the cost of the selected double-sided ring model (Example 1) of the application is reduced by 16.5%, 14.7%, and 14.9% compared with the traditional method (Comparative Examples 1-3). The cost of the selected N-1 composite ring model (Example 2) of the application is reduced by 21.4%, 19.7%, and 19.9% compared with the traditional method (Comparative Examples 1-3).

[0173] Experiment 2:

[0174] 88 wind turbines, double voltage boosting stations, and actual Sheringham Shoal offshore wind farm are used, and the wind turbine and substation coordinates, candidate sea cable schematic diagram of the wind farm are as follows Figure 10 .

[0175] A plurality of examples are compared in the experiment, as shown in Table 3.

[0176] Table 3

[0177]

[0178] Planning results: Figure 11 The planning results of the double-sided ring topology of Comparative Example 4 are shown; Figure 12 The planning results of the CSI structure topology of Comparative Example 5 are shown; Figure 13 The planning results of the selected double-sided ring topology of Example 3 of the application are shown; Figure 14 The planning results of the selected N-1 composite ring topology of Example 4 of the application are shown.

[0179] The cost comparison results (unit: million yuan RMB) are shown in Table 4.

[0180] Table 4

[0181]

[0182] The experimental results show that the cost of Example 3 of the application is reduced by 10.6% and 16.9% compared with the traditional method (Comparative Examples 4-5). The cost of Example 4 of the application is reduced by 17.1% and 23.0% compared with the traditional method (Comparative Examples 4-5).

[0183] Acceleration solution effect verification:

[0184] To verify the effectiveness of the four acceleration strategies, the following verification experiments are set up. Comparative Examples 6-9 each selectively disable one of the acceleration strategies, and Example 5 uses all four acceleration strategies, so as to compare the use effects of each strategy:

[0185] All experiments follow the previous 36 wind turbine wind farm and N-1 composite ring model, considering the complexity of the model itself and the difficulty of solving part of the example after disabling the acceleration strategy, this verification part does not consider the cable selection part, and obtains the high-quality feasible solution (Gap <10%) to calculate the solving time of each example, as shown in Table 5.

[0186] Table 5

[0187]

[0188] The convergence curve of the solving time is shown in Figure 15 The results verify the effectiveness of the four acceleration strategies of the embodiment of the application.

[0189] The embodiment of the application also provides a storage medium for storing a computer program, which is executed to perform at least the method as described above.

[0190] The embodiment of the application also provides a control device, including a processor and a storage medium for storing a computer program; wherein the processor is used to execute the computer program to perform at least the method as described above.

[0191] The embodiment of the application also provides a processor, which executes a computer program to perform at least the method as described above.

[0192] The storage medium can be implemented by any type of non-volatile storage device, or a combination thereof. The non-volatile memory 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 application is intended to include but not limited to these and any other suitable type of memory.

[0193] 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 illustrative. For example, the division of the units is merely logical function division, and there can be other division manners 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 or direct coupling or communication connection between the components can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0194] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place or distributed on a plurality of network units; and some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0195] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; and the integrated unit can be implemented in the form of hardware or hardware plus software functional unit.

[0196] 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 including 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 disk or optical disk and various storage program codes.

[0197] Alternatively, the integrated unit of the present application, if implemented in the form of a software function module and sold or used as an independent product, can also be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, and the computer software product 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 disk or optical disk and various storage program codes.

[0198] The methods disclosed in the several method embodiments of the present application can be combined arbitrarily without conflict to obtain new method embodiments.

[0199] The features disclosed in the several product embodiments of the present application can be combined arbitrarily without conflict to obtain new product embodiments.

[0200] The features disclosed in the several method or device embodiments of the present application can be combined arbitrarily without conflict to obtain new method embodiments or device embodiments.

[0201] 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 should not be limited to these descriptions. For those skilled in the art to which the present application belongs, 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 optimizing a power collection system of an offshore wind farm taking into account cable selection, characterized in that, The method comprises the following steps: S1, obtaining wind farm basic data, including spatial coordinates of wind turbine nodes and transformer substation nodes, wind turbine rated power, electrical parameters and cost parameters of various cable specifications, system life cycle parameters and electricity price information; S2, based on the basic data, a candidate cable set reduction strategy based on clustering partition and distance threshold is used to generate a candidate cable set, and a cable crossing avoidance constraint set is simultaneously identified and generated, so that the model size and the number of crossing constraints are controlled through the candidate set reduction; S3, a scene set containing normal scenes and fault scenes is constructed, and an optimization model taking the minimum total cost in the whole life cycle as the target is established, the model integrates path set variables and constraints: if it is a double-sided ring path, the model integrates cable model selection variables and construction state variables, introduces power balance constraints based on direct current flow, ring topology constraints adapted to cable selection, node degree constraints and cable crossing avoidance constraints; if it is an N-1 composite ring path, on the basis of the double-sided ring path, additional scene-driven operation state variables are integrated, and the full fault set constraint is described through the scene dimension; S4, an acceleration solving strategy corresponding to the path is used: for the double-sided ring path, two strategies of hot start initialization based on pre-solution scheme and cable crossing constraint simplification based on incremental loading are used; for the N-1 composite ring path, on the basis of the two strategies of the double-sided ring path, a fault set dimension reduction strategy based on key fault scene screening is additionally used; through the strategy acceleration solving, an initial topology and cable selection scheme are obtained; S5, the initial scheme is subjected to full fault set N-1 safety verification, and the violation scene is included in the fault set re-solution, wherein the N-1 composite ring path iteratively expands the fault set, and the double-sided ring path verifies the selection compliance, until the scheme meets the N-1 safety criterion, and finally the topology planning and cable selection results are output.

2. The offshore wind farm power collection system optimization method according to claim 1, characterized in that, In step S3, the cable model selection constraint requires that at most one cable model can be selected for any candidate cable position; if it is an N-1 composite ring path, the scene-driven operation state variable constraint requires that the value of the operation state variable of any candidate cable and any scene is constrained between the sum of all model construction state variables and zero; for any scene predefined as a fault scene, the operation state variable of the corresponding specific cable is forced to be zero; if it is a double-sided ring path, the scene-driven operation state variable constraint is not required.

3. The offshore wind farm power collection system optimization method according to claim 1 or 2, characterized in that, In step S3, for the double-sided ring path, the ring topology constraint adapted to cable selection is realized by introducing binary variables representing virtual flow directions and continuous variables representing node cumulative power, the constraint includes: ensuring that the number of virtual flow inflow and outflow of each wind turbine node is 1 to form a ring structure; establishing a bidirectional virtual flow cumulative equation to calculate the cumulative power value of each node in the positive and negative directions; based on the cumulative power value, a constraint condition that the carrying power of each cable needs to meet the N-1 safety criterion is constructed for each cable, and different expressions are adopted for the cables connected to the transformer substations and the cables inside the wind farm; the N-1 composite ring path does not require the ring topology constraint.

4. The offshore wind farm power collection system optimization method according to claim 3, characterized in that, For the double-loop path, the specific implementation process of the loop topology constraint of the cable selection includes: a set of first binary variables are introduced to represent the forward direction of the virtual power flow on the candidate cable; a set of second binary variables are introduced to represent the reverse direction of the virtual power flow on the candidate cable; the sum of the first binary variables and the second binary variables is constrained to be equal to the construction state variable of the cable, so as to ensure that the virtual power flow only exists on the constructed cable, and only one direction of the virtual power flow exists on each cable; for each wind turbine node, the sum of the first binary variables on all incident cables and the sum of the second binary variables on all outgoing cables are constrained to be equal to 1, and the sum of the first binary variables on all outgoing cables and the sum of the second binary variables on all incident cables are also constrained to be equal to 1, so as to force the formation of a closed-loop loop topology structure; a set of first continuous variables are introduced to represent the sum of the downstream wind turbine rated power accumulated by each node on the assumed virtual forward power flow path; a set of second continuous variables are introduced to represent the sum of the downstream wind turbine rated power accumulated by each node on the assumed virtual reverse power flow path; recursive relationship constraints are established for the first continuous variables and the second continuous variables, so that the values of the first continuous variables and the second continuous variables increase along the virtual power flow direction, and the increment is the rated power of the wind turbine node passed through; the lower bound of the values of the first continuous variables and the second continuous variables is constrained to be the rated power of the wind turbine itself, and the upper bound is the capacity of the largest type of cable; based on the first continuous variables and the second continuous variables, a power lower bound constraint is constructed for each cable: for the cable directly connected to the substation, the power capacity required to be carried needs to be not less than the maximum value of the first continuous variable and the second continuous variable at the wind turbine node connected; for the cable connecting two wind turbine nodes, the power capacity required to be carried needs to be not less than the maximum value of the maximum value of the first continuous variable and the second continuous variable of the node with the larger maximum value minus a wind turbine rated power; the above power capacity requirement constraint is associated with the nominal current-carrying capacity of the selected type of cable, so as to ensure that the capacity of the selected cable meets the power transmission requirement under all possible fault scenarios.

5. The offshore wind farm power collection system optimization method according to claim 4, characterized in that, The first continuous variables and the second continuous variables are further constrained to be integer multiples of the wind turbine rated power, so as to meet the physical reality of discrete power accumulation and improve the solving efficiency of the model.

6. The offshore wind farm power collection system optimization method according to claim 4, characterized in that, The power capacity requirement constraint constructed for the cable connecting two wind turbine nodes is linearized by introducing a large enough constant and combining the big M method, so as to ensure that the constraint is only activated when the cable is actually selected and constructed.

7. The offshore wind farm power collection system optimization method according to claim 1 or 2, characterized in that, In step S4, for the N-1 compound loop path, the fault set reduction strategy based on the key fault scenarios screening includes: in the initial iteration, if it is a single substation scenario, preferentially include the cable fault scenarios directly connected with the substation node; if it is a multi-substation scenario, there is no need to initially include specific cable fault scenarios; in the subsequent iteration, based on the line outage distribution factor (LODF) to calculate and sort the system overload risk after all candidate cable faults, or by solving the direct current optimization sub-problem with the objective of minimizing the total system overload to evaluate the fault severity, and select a number of scenarios with the highest risk to join the main problem fault set, and iterate until there is no overload under all single cable fault scenarios.

8. The offshore wind farm power collection system optimization method according to claim 1 or 2, characterized in that, The cable cross constraint simplification strategy based on incremental loading in step S4 includes: in the solving process, initially only include the cross pairs that may be involved by the cables connected with the substation node in the constraint set; after obtaining the initial solution, check whether there is an unconstrained cable cross situation; if there is, add the corresponding cross constraint to the model and re-solve, and gradually approach the feasible solution that meets all cross avoidance requirements through this iterative method.

9. The offshore wind farm power collection system optimization method according to claim 1 or 2, characterized in that, The hot start initialization strategy in step S4 includes: if it is a bilateral loop path, first solve a simplified model that does not consider cable selection or relatively fixed topology, and use the optimal solution as the initial solution of the complex optimization model considering cable selection; if it is an N-1 compound loop path, first solve a bilateral loop optimization model considering cable selection, and use the optimal solution as the initial solution of the N-1 compound loop optimization model considering cable selection; thereby accelerating the solving process.

10. A computer program product comprising a computer program, characterized in that, The computer program, when executed by the processor, implements the offshore wind farm power collection system optimization method according to any one of claims 1 to 9. The computer program, when executed by the processor, implements the offshore wind farm power collection system optimization method according to any one of claims 1 to 9.

Images