Method for Constructing Stable Operation Region of Island Multi-Energy Complementary System for Floating Wind Power

Abstract

The present invention discloses a method for constructing the operation stability region of a multi - energy complementary system for islands with floating wind power, including: S1: Using the CPF method, starting from the basic operating state, obtaining the boundary points of the stability region in the power change direction, and mapping them to the two - dimensional power plane of the required section to obtain the boundary points; S2: On the basis of the boundary points, along the direction of the difference between the two power change directions, obtaining the boundary points of the reduced - dimension stability region; repeating S2 multiple times until the required stability region boundary is obtained. The advantages of the present invention compared with the prior art are as follows: providing a method for constructing the operation stability region of a multi - energy complementary system for islands with floating wind turbines, which can be applied in the energy management system of the multi - energy complementary system for islands to guide work such as formulating the system operation strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of power system stability analysis, and specifically refers to a method for constructing the operation stability domain of a multi-energy complementary system for floating wind power on islands. Background Technique

[0002] There are many islands in China, which are widely distributed. It is impossible to install a large number of traditional power generation facilities, resulting in difficulties in the energy and power supply of the islands. Floating wind power is one of the good ways to solve the energy supply of the islands at present. However, the environment it is in changes more violently, and the uncertainty and intermittency are more significant, bringing major challenges to the stability of the island power system. Therefore, it is necessary to study the operation characteristics of floating wind power and, based on its characteristics, study the method for constructing the stability domain of the island power system containing floating wind power, so as to quickly and real-time evaluate the stability margin and provide support for the energy management system of the island power grid.

[0003] In the island power system containing floating wind power, the stability of the system varies greatly at different times and in different environments. This is essentially different from traditional land microgrids, offshore wind power, etc. This is because the operation environment of the multi-energy complementary system on the islands has the characteristics of "three highs and three strengths", namely high temperature, high humidity, high salt fog, heavy precipitation, strong lightning, and strong typhoons. This brings huge challenges to the operation of floating wind power and also brings different operation environments to other facilities in the multi-energy complementary system on the islands, such as wave energy, photovoltaic, and energy storage.

[0004] Although there is a small amount of research work on floating wind power microgrids, none of them have addressed the issue of constructing the stability region of an island multi-energy complementary system with floating wind power. "Research Progress on Key Technologies of Offshore Floating Wind Turbines" and "Technical Comparison of Several Common Wind Power Generation Systems" analyze the characteristics and advantages of the deployment and assembly of floating wind turbines, and clarify that they have greater economic advantages than fixed offshore wind turbines when the water depth exceeds 60 meters, but they do not involve the construction of the system operation stability region. "A New Method for Rapid Search of the Local Boundary of the Static Voltage Stability Region", "Minimum Load Shedding Algorithm Based on the Static Voltage Stability Region in the Active Load Injection Space", and "Mechanism Analysis and Prevention Strategies of Generator Reactive Power Limit Induced Bifurcation" estimate stability by judging the relative position of the current operating point and the stability region boundary. However, the stability boundary is a high-dimensional curve with complex mathematical properties, difficult to obtain, and its expression is hard to acquire. "Preliminary Exploration of Visualization of the Feasible Region of Voltage Stability in Large Power Systems" and "Review of Research on the Safety Region Method of Power Systems" propose a method that starts from the initial operating state, repeatedly uses the continuation power flow method to calculate the stability limits in various power growth directions, and then determines the stability region boundary through these limit points, also known as the fitting method. However, it requires repeated use of the continuation power flow method, resulting in excessive time consumption and difficulty in real-time prediction. "A New Method for Rapid Search of the Critical Point of the Time-Delay Stability Region of Power Systems" proposes a method based on "stepping-tracking", which quickly obtains prediction points by calculating the spatial tangent vector, then corrects the prediction points in combination with the characteristics of the stability region boundary power flow equation, and finally obtains accurate boundary points, seeking a fast solution method for the voltage stability region. However, it does not consider the situation of highly fluctuating injection power, so it is difficult to apply to island power systems dominated by floating wind power.

[0005] All in all, the main deficiencies of the existing technologies are in two aspects: 1) China's first floating offshore wind power platform, the "Three Gorges Leading", was completed in 2021, but it has not been constructed into an island multi-energy complementary system, and naturally there has been no method for constructing the operation stability region of an island multi-energy complementary system with floating wind power. This technology is the first research in this field; 2) The operation stability region assessment methods of traditional land microgrid systems or similar systems cannot be directly applied to island multi-energy complementary systems with floating wind power. The main reason is that traditional methods do not conduct refined modeling of the operation characteristics of floating wind power, and subsequent assessment methods are also not applicable to the operation characteristics of floating wind power. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the above technical defects and provide a method for constructing the operation stability region of an island multi-energy complementary system applicable to floating wind turbines, which can be applied in the energy management system of the island multi-energy complementary system to guide work such as formulating system operation strategies.

[0007] To solve the above technical problems, the technical solution provided by the present invention is: a method for constructing the operation stable region of an island multi-energy complementary system for floating wind power, including the estimation of the operation characteristics of a floating wind turbine, the definition and modeling of the stable region of the island multi-energy complementary system, and the calculation of the stable region;

[0008] Among them, the estimation of the operation characteristics of the floating wind turbine includes using an extreme learning machine to estimate the output characteristics of the floating wind power given a training set, and the calculation of the stable region includes obtaining the boundary of the spatial stable region of the key nodes, specifically:

[0009] S1: Adopt the CPF method, starting from the basic working state, obtain the boundary points of the stable region in the power change direction, and map them to the two-dimensional power plane of the required section to obtain the boundary points;

[0010] S2: On the basis of the boundary points, along the difference direction of the two power change directions, obtain the boundary points of the reduced-dimensional stable region;

[0011] Repeat S2 multiple times until the required stable region boundary is obtained.

[0012] Preferably, the training set is (x i , t i ), and the training equation is:

[0013]

[0014] Where N is the number of samples, x i ∈R n is the independent variable, t i ∈R m is the dependent variable, that is, the data expected to be output by the model, and the number of hidden layer points is K;

[0015] In the formula, ω i =[ω i1 , ···, ω in is the weight between the input layer and the i-th hidden layer; β i =[β i1 , ···, β in is the weight between the output layer and the i-th hidden layer; b i is the deviation value, and y j is the actual output value of the model corresponding to t i ;

[0016] When the error approaches zero:

[0017]

[0018] Preferably, the definition of the stable region of the island multi-energy complementary system is:

[0019] Ω = {y|f(x, y) = 0} (3)

[0020] The x vector is the system operating state variable, specifically the vector composed of the node voltage amplitude V and phase angle θ of the island multi - energy complementary system;

[0021] y is the vector composed of node powers, and f(x, y) is the necessary and sufficient condition for the system related to the variables to remain stable;

[0022] The stability region boundary of the island multi - energy complementary system is formed by saddle - node bifurcation points, and its expression is:

[0023]

[0024] When The properties of saddle - node bifurcation points are reflected at the stability region boundary of the system voltage.

[0025] Preferably, the saddle - node bifurcation points are transformed into eigenvectors, specifically by constructing the following eigen - formula:

[0026]

[0027] Where L is the left eigenvector of the matrix, the eigenvalue is selected as 0, J T (x) is the transpose of J(x), and substituting it into the power flow equation expression is:

[0028]

[0029] Indicates that the power growth direction is d i When it is the stability region boundary point, where d i is the vector representing the power change direction, and λ i is the scalar representing the load margin;

[0030] When the power change direction is d 0 , and the load margin is d 0 The power change relationship between the obtained stability region boundary point as the base point and its adjacent points is obtained:

[0031]

[0032] Where, e i , e j are the vectors of the power directions at nodes i and j of the system, and ηe i +γe j is the difference between the power change directions λ 0 d 0 and λ i d i That is:

[0033] λ 0 d0 +ηe i +γe j =λ i d i (8)

[0034] where e i 、e j are unit vectors of the action nodes of η and γ, with the same length as d 0 . And they are 1 only for the elements corresponding to node i and node j, and 0 for the rest of the elements. When η and γ change gradually, equation (7) describes that the point starts from the base point described by equation (6) and moves gradually along the boundary of the dimensionality reduction stability region, and the trajectory formed by its movement is the boundary of the dimensionality reduction voltage stability region.

[0035] Preferably, it also includes obtaining a curve fitting equation, specifically:

[0036]

[0037] where S m is a point in the system voltage stability region, and the node coordinates are z m =[x m ,L m ,d], x m is a vector composed of the node voltage amplitude V and the phase angle θ, the voltage phase angle θ of the PV node, and the voltage amplitude V and the phase angle θ of the PQ node; L m is the left eigenvector of the power flow Jacobian matrix, d is the power change direction selected to obtain this point when using the CPF method, e is the unit matrix of whether the node has injection power, with the number of rows 2n 1 +n 2 , the number of columns is n, the number of system PV nodes is n 1 , the number of PQ nodes is n 2 , n=n 1 +n 2 , and J(x) represents the power flow Jacobian matrix of the power system;

[0038] Reducing the dimension of matrix e gives:

[0039]

[0040] The coordinates of the operating point are simplified to z m =[x m ,L m ,γ zm ,η zm ;

[0041] The gradient matrix and the spatial tangent vector are perpendicular to each other to obtain:

[0042]

[0043] The matrix is connected in parallel with the column vector s = [0…0, 0, 0, 1] representing the power change direction of the tangent vector to obtain:

[0044]

[0045] where E = [0…0, 0, Δη| zm , where Δη| zm represents the power direction on the one-dimensional projection represented by Δz| zm at node i;

[0046] The spatial tangent vector Δz| zm The solution equation is:

[0047]

[0048] The tangent vector is normalized to obtain Using the step size σ with the formula:

[0049] Find the predicted coordinates of the next operating point

[0050] Preferably, it includes calibration, specifically including constructing a hyperplane in the coordinate space, and the hyperplane passes through and is perpendicular to the tangent vector Δz|z m , and the expression is:

[0051]

[0052] where the intersection point of the hyperplane and the boundary of the voltage stability region is the next actual saddle-node bifurcation point, and the coordinates are denoted as z m+1 , and the calibration formula for solving this point is:

[0053]

[0054] Find z m+1 Take it as the next reference point and continue to find the boundary of the stability region within the range.

[0055] The advantages of the present invention compared with the prior art are:

[0056] (1) Currently, there is no stability region evaluation method designed for the characteristics of island multi-energy complementary systems containing floating wind power, and the present invention fills this gap;

[0057] (2) The stability domain assessment method for similar scenarios such as traditional land microgrids cannot be directly used in the scenarios targeted by the present invention. Although the conventional continuous power flow method (CPF) has high accuracy, it lacks research on the characteristics of floating wind turbines and has a huge amount of calculation. It is difficult to achieve the requirement of rapid assessment of stability margin and cannot be directly used in island power systems containing floating wind power.

[0058] (3) The method of the present invention is highly close to the calculation results of the CPF method, with extremely small errors, and the calculation time is significantly reduced. It is very suitable for the variable environment of island operation and supports real-time online stability domain evaluation;

[0059] The present invention adopts the extreme learning machine method to estimate the output characteristics of floating wind power, optimize the selection of parameters, improve the efficiency and effectiveness of model training, and at the same time, achieve dimensionality reduction construction by obtaining the boundaries of the two-dimensional / three-dimensional spatial stability domain of key nodes, thereby reducing the computational burden and better meeting the needs of real-time monitoring of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of the saddle-node bifurcation point on the PV curve.

[0061] Figure 2 This is a schematic diagram of the solution results for the boundary of the two-dimensional stable domain.

[0062] Figure 3 It is the result of stable domain construction (including the method of the present invention and the traditional CPF method).

[0063] Figure 4 It is a schematic diagram comparing the time consumption of constructing the stable domain boundary by the method of the present invention and the traditional CPF method. DETAILED DESCRIPTION

[0064] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0065] Combined with Figures 1-4 As shown:

[0066] The present invention adopts the method of extreme learning machine to estimate the output characteristics of floating wind power. The specific method is as follows:

[0067] Given a set of training data (x i ,t i ), where N is the number of samples, x i ∈R n is the independent variable, i.e. the input data of the model, t i ∈R m is the dependent variable, i.e. the data expected to be output by the model, and the number of hidden layer points is K. The model trained by this set of data can be expressed by the following equation.

[0068]

[0069] Wherein, ω i =[ω i1 , ···, ω in is the weight between the input layer and the i-th hidden layer; β i =[β i1 , ···, β in is the weight between the output layer and the i-th hidden layer; b i is the bias value, and y j is the actual output value of the model corresponding to t i .

[0070] When the model accuracy reaches a certain level, it can be considered that the error approaches zero, which is expressed by the following equation.

[0071]

[0072] The input weight ω i and the bias value b i of the above method are initially given, and the initial values of these two parameters have a significant impact on the training effect during the specific model training.

[0073] The present invention uses a particle swarm algorithm to calculate and optimize the selection of these two parameters, improving the efficiency and effectiveness of model training.

[0074] Stable region of the island multi-energy complementary system: It is defined as the region composed of the working state points where only the power system can maintain stable operation and not have a collapse accident under the given power system operation state. The expression is:

[0075] Ω={y|f(x,y)=0}(3)

[0076] Wherein, the x vector is the system operation state variable; in the present invention, it is defined as the vector composed of the node voltage amplitude V and phase angle θ of the island multi-energy complementary system; y is defined as the vector composed of node powers, and f(x,y) is the necessary and sufficient condition for the system related to the variables to maintain stability, which is the power flow calculation equation of the island multi-energy complementary system.

[0077] The stable region boundary of the island multi-energy complementary system is formed by saddle-node bifurcation (SNB) points. The expression of the SNB point is:

[0078]

[0079] Wherein, let That is, the properties of the SNB point are reflected at the stable region boundary of the system voltage, that is, the power flow Jacobian matrix of each point is singular.

[0080] Prediction and correction

[0081] According to the characteristics that the boundary of the voltage stability region is the voltage stability critical point and the power flow Jacobian matrix is singular, in order to transform the Jacobian matrix into a vector that can reflect its properties, the eigenvector meets the requirements and is simple to calculate.

[0082] Construct L as the left eigenvector of the matrix, and select the eigenvalue of 0 to reflect the characteristics of the critical stability of the power flow. The characteristic equation:

[0083]

[0084] In the formula, J T (x) is the transpose of J(x).

[0085] Take the above formula as the criterion and substitute it into the power flow equation expression to obtain the following relationship.

[0086]

[0087] This formula represents the boundary point of the stability region when the power growth direction is d i . The boundary of the system voltage stability region is mainly related to the node voltage and power, and the relationship between power and voltage in the power system can be represented by the Jacobian matrix of the power flow equation.

[0088] In formula (6), d i is a vector representing the power change direction, and λ i is a scalar representing the load margin.

[0089] Therefore, based on the boundary point of the stability region obtained when the power change direction is d 0 and the load margin is d 0 , search for the adjacent point of this point on the projection of the system stability region boundary on nodes i and j of the system, and at the same time reflect the power change relationship between this basic point and the adjacent point, which can be reflected by the following expression.

[0090]

[0091] In the formula, e i , e j are vectors representing the power directions on nodes i and j of the system and belong to unit vectors; ηe i +γe j is the difference between the power change directions λ 0 d 0 and λ i d i , that is:

[0092] λ 0 d 0 +ηe i+γe j = λ i d i (8)

[0093] where e i , e j are unit vectors representing the action nodes of η and γ, with the same length as d 0 and being 1 only for the elements corresponding to nodes i and j, and 0 for the remaining elements.

[0094] Equations (6) and (7) are mathematical expressions describing the boundary points of the voltage stability region. By comparing the two equations, it can be found that when η = γ = 0, the two equations are equivalent and both describe the boundary points of the stability region when the power growth direction is d 0 .

[0095] When η and γ change gradually, the points described by Equation (7) start from the basic points described by Equation (6) and move gradually along the boundary of the reduced-dimensional stability region.

[0096] It can be considered that the moving points formed by the points described by Equation (7) are all SNB points, and the trajectory formed by their movement is the boundary of the required reduced-dimensional voltage stability region.

[0097] From this, the principle of obtaining the stability region boundary by this method can be summarized, and the steps are as follows.

[0098] 1) Using the CPF method, starting from the basic operating state, obtain the boundary point S 0 of the stability region when the power change direction is d 1 , and map it to the two-dimensional power plane of nodes i and j required, to obtain the boundary point S' 1 ;

[0099] 2) Based on the boundary point S' 1 , move along the difference direction of the power change direction d 0 and the power change direction d 1 , that is, the direction of ηe i +γe j , to obtain the boundary point S' 2 of the reduced-dimensional stability region;

[0100] 3) Repeat step 2) multiple times until the required stability region boundary is obtained.

[0101] The prediction-correction method has the advantages of fast speed and reliable accuracy in obtaining the boundary of the voltage stability region. However, like the CPF method, there is a problem that it is difficult to obtain the expression of the boundary fitting curve. Therefore, the present invention modifies it on this basis, adds the representation coordinates of the operating points in the stability region, and transforms the prediction-correction method into a problem of solving multiple curve points and then obtaining the curve fitting equation.

[0102] Let a point S in the system voltage stability region m have its node coordinates represented by z m = [x m , L m , d], where x m is a vector composed of the node voltage magnitude V and phase angle θ, including the voltage phase angle θ of the PV node and the voltage magnitude V and phase angle θ of the PQ node; L m is the left eigenvector of the power flow Jacobian matrix; d is the power change direction selected for this point when using the CPF method.

[0103] According to the set node coordinates, the gradient matrix at this point is obtained as follows.

[0104]

[0105] In the formula, e is the identity matrix indicating whether the node has injection power, with 2n 1 + n 2 rows and n columns, where n is the number of PV nodes in the system 1 , and the number of PQ nodes is n 2 , n = n 1 + n 2 . J(x) represents the power flow Jacobian matrix of the power system.

[0106] In this section, only the projection boundary of the voltage stability region boundary on the two-dimensional power plane formed by nodes i and j is sought. Therefore, the matrix e can be reduced in dimension to obtain the reduced-dimensional gradient matrix as follows.

[0107]

[0108] In this regard, the operating point coordinates can also be simplified to z m = [x m , L m , γ zm , η zm .

[0109] According to the perpendicularity of the gradient matrix and the space tangent vector, the following relational expression can be obtained.

[0110]

[0111] This formula can neither solve the tangent vector at this point nor determine the direction of the tangent vector. Therefore, the matrix is paralleled with the column vector s = [0…0, 0, 0, 1] constructed to represent the power change direction of the tangent vector to obtain the following formula.

[0112]

[0113] It can be proved by mathematical calculation that in the formula where represents the power direction on the one-dimensional projection represented by node i.

[0114] According to the constructed matrix equation, the space tangent vector The equation is solved as follows.

[0115]

[0116] Normalize the tangent vector to obtain Then, according to the step size σ, the predicted coordinates of the next operating point can be obtained by the following formula

[0117]

[0118] Based on the power flow equation and the properties of the voltage stability region boundary, is corrected.

[0119] First, construct a hyperplane in the coordinate space. This plane passes through and is perpendicular to the tangent vector . According to matrix calculation and the properties of perpendicular lines, its expression is as follows.

[0120]

[0121] From geometric relationships and the mathematical properties of the system voltage stability region, it can be known that there must be an intersection between the hyperplane constructed by the above formula and the voltage stability region boundary, and this intersection is the next actual SNB point, and its coordinates are denoted as z m+1 .

[0122] The coordinates of this point are obtained by solving the simultaneous equations of the hyperplane expression and the power flow equation based on Newton's method. The correction equation for solving is as follows:

[0123]

[0124] After obtaining z m+1 , take it as the next reference point, and continue to use the above prediction-correction method to obtain the stability region boundary within the required range.

[0125] The specific steps are as follows.

[0126] 1) Set the initial operating point power coordinates (P 0,i , P 0,j ), the tangent direction prediction step size σ, the nodes i and j to be monitored, and set the corresponding power unit vectors e i , e j ;

[0127] 2) Set the initial power growth direction d 0 , and solve for d using the CPF method 0 . The SNB point corresponding to the direction has coordinates (P i , P j ) in the two-dimensional power plane, and can also be expressed as (P 0,i + λ 0 ·ΔP i , P 0,j + λ 0 ·ΔP j ). Then determine the coordinates z 0 of the voltage-related variables x 0 in it and the power flow Jacobian matrix J(x 0 ), and calculate the left eigenvector L 0 when the eigenvalue of the Jacobian matrix is 0. Let z 0 = [x 0 , L 0 , γ 0 , η 0 . Among them, γ 0 = η 0 = 0;

[0128] 3) Based on the SNB point corresponding to the d 0 direction as the base point, calculate the parameters of each item of the matrix in Equation (16), and according to Equation (13), set Δη| z0 = 1; that is, E = [0…0, 0, 0, 1], and solve for the spatial tangent vector 0 at z . Normalize it to get

[0129] 4) Substitute z 0 , the given step size σ and into Equation (3 - 12) to solve for the coordinates of the predicted point

[0130] 5) According to the correction equation (16), using the operating point as the initial operating point, use the Newton method to perform power flow calculation to solve for the SNB point as the correction point z

[0131] = [x 1 , L 1 , γ 1 , η 1 , η 1 . The coordinates of this point in the two-dimensional power plane are expressed as (P 0,i + λ 0 ·ΔP i + γ 1 , P 0,j + λ 0 ·ΔPj +η 1 );

[0132] 6) Let z 0 = z 1 and repeat steps 3) and 4) above to obtain a new z 1 and repeat this step until the z 1 point power coordinates exceed the range to be determined;

[0133] (7) Take the z 0 obtained in 2) as the new base point, and set i.e., E = [0…0, 0, 0, -1], and repeat steps 3) to 6) until the z 1 point power coordinates exceed the range to be determined, and end the process.

[0134] The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0135] The above describes the present invention and its embodiments. Such description is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. Generally speaking, if those of ordinary skill in the art are inspired by it and, without departing from the gist of the present invention, design similar structural modes and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of the present invention.

Claims

1. A method for constructing a static voltage stability domain for an island multi-energy complementary system of floating wind power, characterized in that: It includes the estimation of the operating characteristics of floating wind turbines, the definition and modeling of the static voltage stability domain of the island multi-energy complementary system, and the calculation of the stability domain; The estimation of the operating characteristics of the floating wind turbine generator includes estimating the floating wind power output characteristics using an extreme learning machine given a training set, and the calculation of the stability domain includes obtaining the boundary of the spatial static voltage stability domain for the selected node, specifically: S1: Using the CPF method, starting from the basic working state, the boundary points of the stable domain of the power change direction are obtained, and mapped to the two-dimensional power plane of the nodes i and j to obtain the boundary points; S2: Based on the boundary points, iterative search is performed along the difference vector direction of the power change direction between node i and node j to obtain the boundary points of the stable domain after dimensionality reduction; Repeat S2 several times until the desired stability region boundary is obtained; The stability domain of the island multi-energy complementary system is defined as: Ω={y|f(x,y)=0} (3) The x vector is the system operation state variable, specifically the vector composed of the node voltage amplitude V and phase angle θ of the island multi-energy complementary system; y is the vector of node power, f(x,y) is the necessary and sufficient condition for the variable-related system to remain stable; The boundary of the stable region of the island multi-energy complementary system is formed by the saddle bifurcation point, and its expression is: when The boundary of the system voltage stability region shows the characteristics of saddle node bifurcation point; The saddle bifurcation point is converted into a eigenvector, specifically by constructing the following characteristic formula: Where L is the left eigenvector of the matrix, and the eigenvalue is selected to be 0, J T (x) is the transpose of J(x), which can be substituted into the power flow equation as follows: Indicates that the power growth direction is d i The boundary point of the stable region, where d i is the vector representing the power change direction, λ i is a scalar representing the load margin; When the boundary point of the stable region obtained with the power change direction as d0 and the load margin as d0 is taken as the base point, the power change relationship between the base point and the adjacent points is obtained: Among them, e i 、e j is the vector of the power direction of the system at nodes i and j, ηe i +γe j The power change directions of the two points λ0d0 and λ i d i The difference, that is: λ0d0+ηe i +ge j =λ i d i (8) where e i 、e j is the unit vector of the nodes where η and γ act, with the same length as d0, and only the elements corresponding to nodes i and j are 1, and the rest are 0. When η and γ change gradually, the point described by equation (7) starts from the basic point described by equation (6) and gradually moves along the boundary of the reduced-dimensionality stability domain. The trajectory formed by its movement is the boundary of the reduced-dimensionality voltage stability domain.

2. The method for constructing a static voltage stability domain for an island multi-energy complementary floating wind power system according to claim 1, characterized in that: The training set is (x i ,t i ), the training equation is: Where N is the number of samples, x i ∈R n is the independent variable, t i ∈R m is the dependent variable, that is, the data expected to be output by the model, and the number of hidden layer points is K; Where ω i =[ω i1 ,···,ω in ] is the weight of the input layer and the i-th hidden layer; β i =[β i1 ,···,β in ] is the weight of the output layer and the i-th hidden layer; b i is the deviation value, y j For i The corresponding actual output value of the model; When the error approaches zero:

3. The method for constructing a static voltage stability domain for an island multi-energy complementary floating wind power system according to claim 1, characterized in that: It also includes obtaining the curve fitting equation, specifically: Where S m is a point in the system voltage stability domain, with node coordinates z m =[x m ,L m ,d],x m is the vector formed by the node voltage amplitude V and phase angle θ, the voltage phase angle θ of the PV node and the voltage amplitude V and phase angle θ of the PQ node; L m is the left eigenvector of the power flow Jacobian matrix, d is the power change direction selected for this point when using the CPF method, e is the unit matrix of whether the node has injected power, the number of rows is 2n1+n2, the number of columns is n, the number of PV nodes in the system is n1, the number of PQ nodes is n2, n=n1+n2, J(x) represents the power flow Jacobian matrix of the power system; Reduce the dimension of matrix e to get: The coordinates of the working point are simplified to z m =[x m ,L m ,γ zm ,η zm ]; The gradient matrix and the spatial tangent vector are perpendicular to each other: The matrix In parallel with the column vector s = [0…0,0,0,1] representing the direction of change of the tangent vector power, we get: Where E=[0…0,0,Δη| zm ], where Δη|z m represents Δz| zm The power direction on the one-dimensional projection represented by node i; Space tangent vector Δz| zm The solution equation is: The tangent vector is normalized to get With step size σ, we use: Obtain the predicted coordinates of the next working point 4. The method for constructing a static voltage stability domain for an island multi-energy complementary floating wind power system according to claim 3, characterized in that: Including Correction, specifically includes constructing a hyperplane in the coordinate space, the hyperplane passing through and perpendicular to the tangent vector Δz|z m , the expression is: The intersection of the hyperplane and the voltage stability region boundary is the next actual saddle node bifurcation point, and its coordinates are marked as z m+1 , the correction formula for this point is: Find z m+1 Use it as the next reference point and continue to find the boundary of the stable domain within the range.

Images