A direct-drive wind farm cluster output optimization method based on improved fish eagle optimization algorithm

Abstract

The method for optimizing the output of a direct-drive wind farm cluster based on an improved fish-eagle optimization algorithm includes the following steps: considering the dynamic modulation influence of active power and reactive power on the port impedance characteristics of a direct-drive wind turbine, a frequency coupling impedance model of a grid-side converter of the direct-drive wind turbine is established; based on the established frequency coupling impedance model, an aggregated impedance model of a grid connection point of the direct-drive wind farm is established; under a given active scheduling instruction, the aggregated impedance model established by S2 is subjected to impedance scanning in an average power distribution mode, the maximum peak Nyquist stability criterion is used to analyze the wide-frequency oscillation risk, and the improved fish-eagle optimization algorithm IOOA is introduced to optimize the distribution of active power and reactive power of each direct-drive wind turbine with the maximum safety margin as the target. The method only needs to use the existing power distribution freedom within the station, does not need to modify the hardware or controller of the wind turbine, has small implementation difficulty and low risk, and is suitable for the design optimization of a newly-built wind farm and the performance improvement of a wind farm in operation.

Description

Technical Field

[0001] This invention relates to the field of broadband oscillation suppression in wind farms, and specifically to a method for optimizing the output of a direct-drive wind farm cluster based on an improved Osprey optimization algorithm. Background Technology

[0002] In current technologies, the penetration rate of new energy sources, represented by wind power, in power systems is continuously increasing, and new power systems exhibit significant characteristics of high proportions of renewable energy and power electronic equipment. Direct-drive wind turbines have been widely used in the wind power sector due to their advantages such as high power generation efficiency and low maintenance costs. However, when large-scale wind farms are connected to the grid, the power system exhibits weak grid characteristics, and the resulting broadband oscillation problem has seriously threatened the stable operation of the power system. Therefore, in-depth analysis of the oscillation mechanism of direct-drive wind farms and the proposal of effective suppression strategies are of crucial theoretical and engineering value.

[0003] Impedance analysis is commonly used in broadband oscillation analysis. This method measures the frequency response characteristics of the interconnect ports of the subsystems offline or online, effectively avoiding the order reduction problem of complex state variables within the system. However, existing impedance modeling studies often focus on the impact of converter control parameters on stability, neglecting the dynamic modulation effect of the wind turbine operating conditions, i.e., the real-time output of active and reactive power, on the impedance frequency characteristics.

[0004] Measures to suppress broadband oscillations in wind farms can be mainly divided into three categories: The first category is to achieve rapid suppression of oscillations by optimizing the control parameters of the wind turbine converter. However, in actual engineering, the control parameters of wind turbine units often undergo rigorous factory tuning and type testing. Changing these parameters can easily disrupt the original dynamic balance of the unit, causing unpredictable impacts on the unit's instantaneous protection, fault ride-through, and other dynamic performance characteristics. The second category is to improve the control system, for example, by introducing additional damping control loops on the rotor side and constructing specific feedback to increase target damping to suppress oscillations. The third category is to add additional equipment, such as connecting a static synchronous compensator in series on the grid side to increase the equivalent resistance of the system to suppress wind farm oscillations. However, both of these methods involve hardware upgrades and increased costs. Therefore, how to effectively suppress broadband oscillations without increasing additional hardware investment and without changing the existing control parameters of the unit is an urgent technical problem to be solved. Summary of the Invention

[0005] This invention provides a method for optimizing the output of a direct-drive wind farm turbine cluster based on an improved Osprey optimization algorithm. This method considers the frequency coupling effect, establishes an analytical model of the direct-drive wind turbine impedance that explicitly depends on active and reactive power, and introduces the maximum peak Nyquist stability criterion to analyze the potential broadband oscillation risk of the wind farm under the traditional power average distribution control method. Then, the improved Osprey optimization (IOOA) algorithm is used to optimize the power distribution of the turbine cluster, thereby reshaping the station impedance and effectively suppressing the occurrence of broadband oscillation.

[0006] The technical solution adopted in this invention is as follows: The method for optimizing the output of direct-drive wind farm turbine clusters based on the improved Osprey optimization algorithm includes the following steps: S1: Considering the dynamic modulation effect of active power and reactive power on the port impedance characteristics of the direct-drive wind turbine, establish a frequency coupling impedance model of the grid-side converter of the direct-drive wind turbine. S2: Based on the frequency coupling impedance model established in S1, establish the aggregate impedance model of the grid connection point of the direct-drive wind farm; S3: Under a given active power dispatching command, first use the average power allocation method to perform impedance scanning on the aggregate impedance model established in S2, use the maximum peak value Nyquist stability criterion to analyze the broadband oscillation risk, and then introduce the improved Osprey optimization algorithm IOA to optimize the allocation of active and reactive power of each direct-drive wind turbine with the goal of maximizing safety margin. S4: Perform impedance scanning on the direct-drive wind farm with optimized power distribution to verify the broadband oscillation suppression effect.

[0007] S1 includes: 1.1: Modeling the main circuit of the direct-drive wind turbine. In the time domain, assuming the grid-side converter initially operates under three-phase symmetrical balanced conditions, after injecting positive and negative sequence small-signal disturbances at the grid connection point, due to the slip between the stator and rotor, the frequency of the rotor-side current component will shift with the slip. Taking phase A as an example, its voltage and current in the frequency domain can be specifically described as follows: (1); (2); (3); (4); In the above formula, and The voltage and current of phase a at different frequencies are respectively... f The disturbance component at the location; Indicates the fundamental voltage at frequency f The component at the location; Indicates the positive-sequence perturbation voltage at frequency f The component at the location; This indicates the negative sequence disturbance voltage at frequency f The component at the location; and These are the initial phase angles of the positive-sequence and negative-sequence disturbance voltages, respectively. U 1. U p and U n These represent the amplitudes of the fundamental voltage, positive-sequence perturbation voltage, and negative-sequence perturbation voltage, respectively. I 1. I p and I n These represent the amplitudes of the fundamental current, the positive-sequence perturbation current response, and the negative-sequence perturbation current response, respectively. f 1. f p , f n These are the fundamental frequency, the positive-sequence perturbation frequency, and the negative-sequence perturbation frequency, respectively. The perturbation frequency; Indicates the fundamental current at frequency f The component at the location; This indicates the positive-sequence disturbance current at frequency f The component at the location; This indicates that the negative sequence disturbance current is at a frequency of f The component at the location; , and These are the initial phase angles of the fundamental current, the positive-sequence disturbance current response, and the negative-sequence disturbance current response, respectively. This refers to the active power generated by the direct-drive fan. The reactive power generated by the direct-drive fan; It is an imaginary number; 1.2: A DC voltage loop model for the direct-drive wind turbine is constructed. Ignoring the second harmonic component, convolution processing is performed on each frequency component. The expression for the DC-side voltage in the frequency domain is: (5); In the formula, This refers to the DC-side voltage in the frequency domain. This is the DC side voltage; This is the conjugate value of the fundamental frequency current; s =±j2π f p ; For DC side capacitors; For filtering inductors, This is the DC side current.

[0008] 1.3: Modeling the phase-locked loop (PLL) of the direct-drive wind turbine, with the PLL tracking the phase angle of the grid voltage. θ 1 and small signal disturbance voltage phase angle θ ,Right now θ PLL = θ 1+ θ , θ PLL The phase angle is the output phase angle of the phase-locked loop; in the frequency domain, the small-signal voltage disturbance and the phase-locked loop control... θ The transfer function between them is: (6); In the formula, For frequency f Voltage phase angle; transfer function T PLL ( s )= H PLL ( s ) / [1+ U 1[ f ] H PLL ( s )], H PLL (s) is the transfer function of the phase-locked loop. U 1[ f [The fundamental voltage at a frequency] f The component at the location; 1.4: Modeling the current loop of the direct-drive fan, considering the influence of disturbance voltage on the output angle of the phase-locked loop, and transforming it to the frequency domain, using the convolution theorem, we can obtain: (7); (8); In the above formula, The d-axis frequency domain component of the positive-sequence disturbance current; The q-axis frequency domain component of the positive-sequence disturbance current; 1.5: Modeling the modulation stage of the direct-drive fan. The output signal of the current inner loop controller can be used to obtain the modulation wave through Park. Taking phase a as an example, the modulation signal contains two terms, and the specific expression in the frequency domain is as follows: (9); (10); In the above formula, This is the first modulation signal of phase a; This is the second modulation signal of phase a; This represents the frequency domain component of the d-axis voltage output by the current loop. This represents the frequency domain component of the q-axis voltage output by the current loop. This is the frequency domain representation of the phase-locked loop transfer function; Angular frequency; When the frequency is 0, the current loop outputs the d-axis voltage; When the frequency is 0, the current loop outputs the q-axis voltage; 1.6: Based on the frequency coupling characteristics, the output impedance model of the grid-side converter is represented by a two-dimensional matrix. Combining the DC-side bus voltage with the grid-side converter output voltage, the matrix relationship between the disturbance voltage and the grid-side converter output can be obtained: (11); (12); In the above formula, This is the matrix showing the relationship between the disturbance voltage and the output current; This is the matrix showing the relationship between the disturbance voltage and the output voltage. Indicates resistance; Indicates the dynamic coupling phase on the DC side; Indicates frequency offset coupling phase; Represents coefficient matrix 1; Represents coefficient matrix 2; Represents coefficient matrix 3; Represents coefficient matrix 4; ( P + jQ ) represents complex power; P - jQ ) represents the conjugate of complex power.

[0009] (13); (14); (15); (16); (17); (18); (19); in, I Z Indicates direct current; U 1 represents the fundamental voltage; CZ For DC side capacitors; L p For filtering inductors; P Indicates active power; Represents the voltage loop transfer function; Represents the current loop transfer function; 1.7: The effect of disturbance voltage in the phase-locked loop on the output voltage of the grid-side converter is as follows: (20); In the formula, This is the matrix showing the relationship between the disturbance voltage and the output voltage in the phase-locked loop; The transfer function of the phase-locked loop; Represents the current loop transfer function; This represents the decoupling coefficient.

[0010] 1.8: Based on the data models (1), (5), (7), (8), (9), (10), and (20) of the above steps, the admittance model of the frequency coupling of the grid-side converter can be derived: (twenty one); (twenty two); In the above formula, The admittance matrix represents the frequency coupling of the grid-side converter; Represents the identity matrix; This is the matrix showing the relationship between the disturbance voltage and the output voltage. This is the matrix showing the relationship between the disturbance voltage and the output voltage in the phase-locked loop; This is the matrix showing the relationship between the disturbance voltage and the output current; L m =diag[ sL p , ( s - jω 1) L p ], L m Represented as a diagonal matrix; L p For filtering inductors; sL p The first diagonal element; s - jω 1) L p The second diagonal element; This indicates the effect of voltage disturbances on the output current response of the grid-side converter through the PI controller current loop; Modulation gain; Represents the current loop transfer function; 1.9: The frequency coupling impedance model of the grid-side converter for direct-drive wind turbines is as follows: (twenty three); In the formula, This is the frequency coupling impedance matrix.

[0011] S2 includes: 2.1: A direct-drive wind farm consists of direct-drive wind turbines, box-type substations, collector lines, and main transformers. The collector lines within the farm have a relatively short spatial span, and their parallel admittance has little impact on small disturbance stability. Therefore, they can be simplified into a series impedance model with lumped parameters during modeling.

[0012] definition The high-voltage side reduced impedance includes the parameters of the transformer substation. For the collector line impedance, The equivalent impedance of a single-unit branch; (twenty four); (25); In the above formula, For the transformer's turns ratio, The impedance of the transformer is... The admittance of the transformer substation. The equivalent impedance of a single-unit branch; 2.2: Assuming the main trunk of a direct-drive wind farm has no branch links and no secondary branches, it is... x It consists of several branch roads, each branch road is composed of... n i If the direct-drive wind turbines are connected in parallel, then its first... i Equivalent impedance of the branch for: (26); In the formula, For the first i The equivalent impedance of each branch; For the first i The total active power of the branch circuit; For the first i The total reactive power of the branch circuit; The equivalent impedance of a single-unit branch; For the first i The first branch road j The active power of the typhoon generator; For the first i The first branch road j The reactive power of the typhoon generator; 2.3: Equivalent Impedance of the Entire Wind Farm for: (27); (28); In the above formula, The high-voltage side reduced impedance includes the parameters of the main transformer. The main variable is the ratio. This represents the equivalent admittance of the entire wind farm; The admittance of the main variable; The impedance of the main transformer; The impedance of the grid connection point; The impedance of the power transmission line.

[0013] In step S3, the impedance model of a single unit in a direct-drive wind farm is scanned and verified, and the maximum peak value Nyquist stability criterion is used for analysis. Finally, the IOOA algorithm is introduced for optimization. Specifically, this includes: 3.1: Positive and negative sequence harmonic voltage disturbances with a frequency range of 2.5~1000Hz are sequentially injected at the grid connection point of the direct-drive wind farm, and the voltage and current time-domain waveform data at the grid connection point are collected in real time. The disturbance amplitude is set to 3% of the grid voltage amplitude. Next, characteristic frequency phasors are extracted based on the Fast Fourier Transform (FFT) algorithm. The FFT expression is as follows: (29); In the formula: This refers to the voltage response amplitude. N To analyze the number of points; n As the starting point; T S Indicates the sampling period; f p Indicates the frequency of the disturbance; It is a complex exponential function.

[0014] The calculated response increment is decoupled using the symmetrical component method, and the amplitude and phase characteristics of the positive and negative sequence impedances at specific frequencies are calculated based on the derivation of Ohm's law. Specifically, the symmetrical component method decoupling is as follows: (30); In the formula: , and These are the positive-order, negative-order, and zero-order component phasors, respectively; 'a' is the basic operator. , and These are three phase vectors, a, b, and c, respectively.

[0015] Based on the derived formula of Ohm's law, the amplitude and phase characteristics of positive and negative sequence impedances at a specific frequency are calculated: (31); (32); In the formula: and These are the positive-sequence impedance and the negative-sequence impedance, respectively. and For positive-sequence disturbance voltage and current responses; and The voltage and current responses to negative-sequence disturbances; and The positive sequence impedance magnitude and phase angle; and These represent the negative sequence impedance magnitude and phase angle.

[0016] Finally, the above steps are repeated cyclically within the swept frequency band at a preset step size until the impedance characteristics of the entire frequency band are obtained. For example... Figure 6 As shown in (a).

[0017] 3.2: Introduce the maximum peak value Nyquist stability criterion and prohibit the region radius. With system gain margin Phase margin θ The relationship is as follows: (33); In the formula: It is the reciprocal of the distance from the intersection of the impedance ratio curve and the negative real axis to the origin. θ The angle between the intersection of the impedance ratio curve and the unit circle and the negative real axis; 3.3: The Osprey Optimization Algorithm (OOA) is an intelligent swarm optimization algorithm that simulates the hunting behavior of ospreys and seeks optimal results through a combination of global exploration and local exploitation. A randomly initialized osprey population can be defined as: (34); (35); In the above formula: X This represents the osprey location population matrix. X i Table 1 i The initial position of the osprey. N Indicates population size. m Indicates the number of problem variables. x ij Indicates the first j Among the problem variables, the first one is... i The initial position of the osprey. r ij A random number between 0 and 1. b l,j and b u, jThey represent the first j The upper and lower boundaries of the problem variables; 3.4: The OOA algorithm consists of two phases: development and exploration. Ospreys are ranked according to their fitness values, and those with better fitness are considered as a "school of underwater fish," which can be represented as: (36); In the formula: For the first i The fitness value of the osprey; Indicates the first Only the location of the osprey; For the first Only an osprey; Osprey population size; , They represent the first The and the first The objective function value of the Osprey; This indicates the position of the osprey that achieves the optimal objective function value; For union to be true; 3.5: The Osprey's position is updated as follows: (37); (38); In the above formula: Indicates the first stage of development. j Among the problem variables, the first one is... i The new location for the osprey; Indicates the second stage of development j Among the problem variables, the first one is... i The new location for the osprey; Indicates the first j Among the problem variables, the first one is... i The location of the fish chosen by the osprey; A random number that is either 1 or 2; Indicates the first j The lower bound of each problem variable; Indicates the first j The upper limit of the problem variables; This represents the number of iterations. Indicates the exploration stage j Among the problem variables, the first one is... i The new location for the osprey; 3.6: In the initialization phase of the OOA algorithm, to improve the convergence speed, a chaotic mapping is introduced to initialize the population. Differentiated mappings are applied to different intervals to ensure a uniform distribution of the population in the search space, thereby enabling faster discovery of the optimal solution; (39); In the formula: Indicates the upper bound of the decision variable; Indicates the step size scaling factor; This represents the chaotic state in the (n+1)th iteration; This represents the chaotic state value in the nth iteration; Indicates the control parameters for chaotic mapping; 3.7: Considering the frequency coupling of the direct-drive wind farm grid-connected system, the objective function is defined according to the maximum peak value Nyquist stability criterion: (40); (41); In the above formula: The objective function is... For stability margin; and The penalty factor for active and reactive power is usually taken as 10. 4 ; The reactive power squared penalty factor is set to 10. 2 ; and The active and reactive power generated by a single direct-drive wind turbine, 0 ≤ P k ≤ P max , | Q k |≤ Q max , P max Rated active power of direct-drive fan; Q max The rated reactive power of the fan; its size meets the single-unit capacity constraint. For the entire field's active power dispatch instructions; Number of direct-drive wind turbines at the station; This represents the impedance ratio of the power grid. The radius of the forbidden region in the Nyquist stability criterion for maximum peak value.

[0018] This invention discloses a method for optimizing the output of a direct-drive wind farm cluster based on an improved Osprey optimization algorithm. The technical effects are as follows: 1) This invention establishes an explicit active power dependent model. and reactive power The frequency coupling impedance model of direct-drive wind turbines overcomes the shortcomings of traditional impedance modeling, which ignores the dynamic modulation of impedance characteristics by operating conditions. It can more realistically reflect the impedance frequency characteristics of wind turbines at different output levels, thereby more accurately assessing the broadband oscillation risk of wind farms.

[0019] 2) This invention optimizes the distribution of active and reactive power among the units within a wind farm through the IOOA algorithm, actively reshaping the aggregate impedance at the grid connection point of the station. Without changing the converter control parameters or adding additional damping devices or compensation equipment, it achieves effective suppression of wideband oscillations, significantly reducing the investment cost of suppression measures and the risk of impact on the dynamic performance of the units.

[0020] 3) This invention employs an improved Osprey optimization algorithm (IOOA) that initializes the Osprey population using chaotic mapping. This enhances the uniformity of the initial solution distribution, accelerates the convergence speed of the algorithm towards the optimal solution, and maintains a good balance between global exploration and local development. It can efficiently solve power allocation optimization problems involving multiple units and multiple constraints.

[0021] 4) This invention introduces the maximum peak Nyquist stability criterion and defines a safety margin. The stability requirements are quantified into an optimizable objective function, ensuring that the optimization results can guarantee that the impedance ratio curve of the system is far from the prohibited area under harsh operating conditions such as extremely weak power grids (e.g., short-circuit ratio SCR=1.5), thus effectively enhancing the grid connection stability of wind farms.

[0022] 5) This method only requires the use of the existing power distribution freedom within the wind farm, without the need to modify the wind turbine hardware or controller. It is easy to implement and has low risk. It is suitable for the design optimization of new wind farms and the performance improvement of existing wind farms, and has good engineering promotion value. Attached Figure Description

[0023] The present invention will be further described below with reference to the accompanying drawings and examples; Figure 1 This is a control block diagram for a direct-drive fan.

[0024] Figure 2 This is a schematic diagram of a wind farm structure.

[0025] Figure 3 This is the equivalent circuit diagram for a single branch.

[0026] Figure 4 This is a flowchart of the impedance scan process.

[0027] Figure 5 The Nyquist stability criterion diagram shows the maximum peak value.

[0028] Figure 6 This diagram illustrates the stability criteria for the grid-connected substation system under different short-circuit ratios using an average distribution method.

[0029] Figure 7 The stability criterion diagram for the substation grid-connected system under different short-circuit ratios after optimization. Detailed Implementation

[0030] The technical solution and effects of the present invention will be further described in detail below with reference to the accompanying drawings.

[0031] A method for optimizing the output of a direct-drive wind farm turbine cluster based on an improved Osprey optimization algorithm is characterized by the following steps: Step S1: Based on the topology of the direct-drive fan, perform impedance modeling; Step S2: Based on the single-unit impedance model of a direct-drive wind turbine, take a wind farm as an example to establish an equivalent impedance model of a direct-drive wind farm. Step S3: Impedance scanning is performed on the direct-drive wind farm. The peak-to-peak Nyquist stability criterion is used to analyze wind farms that traditionally use the average power distribution method, revealing a risk of broadband oscillation. Then, the IOOA algorithm is introduced to optimize the power distribution of the wind farm cluster. Step S4: Perform impedance scanning on the wind farm after power allocation optimization to verify the broadband oscillation risk of the site; Preferably, in step S1, an analytical model of the direct-drive wind turbine impedance, which explicitly depends on active and reactive power, is established, and the control block diagram of the direct-drive wind turbine is shown below. Figure 1 As shown; Step S11: Model the main circuit of the direct-drive wind turbine. In the time domain, assuming the grid-side converter initially operates under three-phase symmetrical balanced conditions, after injecting positive and negative sequence small signal disturbances at the grid connection point, due to the slip between the stator and rotor, the frequency of the rotor-side current component will shift with the slip. Taking phase A as an example, the corresponding expressions of its voltage and current in the frequency domain can be specifically described as follows: ; ; ; ; in, U 1. U p and U n These represent the amplitudes of the fundamental voltage, positive-sequence perturbation voltage, and negative-sequence perturbation voltage, respectively. I 1. I p and I n These represent the amplitudes of the fundamental current, the positive-sequence perturbation current response, and the negative-sequence perturbation current response, respectively. f 1. f p , f n These are the fundamental frequency, the positive-sequence perturbation frequency, and the negative-sequence perturbation frequency, respectively. and These are the initial phase angles of the positive-sequence and negative-sequence disturbance voltages, respectively. , and These are the initial phase angles of the fundamental current, the positive-sequence disturbance current response, and the negative-sequence disturbance current response, respectively.

[0032] Step S12: Model the DC voltage loop of the direct-drive fan, ignoring the second harmonic component, and perform convolution processing on each frequency component. The expression for the DC side voltage in the frequency domain is: ; in, I 1 * This is the conjugate value of the fundamental frequency current. s =±j2π f p , C Z For DC side capacitors; L p For filtering inductors, I Z This is the DC side current.

[0033] Step S13: Model the phase-locked loop (PLL) model of the direct-drive wind turbine. The PLL tracks the phase angle of the grid voltage. θ 1 and small signal disturbance voltage phase angle θ ,Right now θ PLL = θ 1+ θ In the frequency domain, small-signal voltage disturbances and phase-locked loop control θ The transfer function between them is: ; Among them, the transfer function T PLL ( s )= H PLL ( s ) / [1+ U 1[ f ] H PLL ( s )], H PLL (s) is the transfer function of the phase-locked loop.

[0034] Step S14: Model the current loop of the direct-drive fan, considering the influence of disturbance voltage on the output angle of the phase-locked loop. Transform to the frequency domain and use the convolution theorem to obtain: ; ; Step S15: Model the modulation circuit of the direct-drive fan. The output signal of the current inner loop controller can be used to obtain the modulation wave through Park. Taking phase a as an example, the modulation signal contains two terms, and the specific expression in the frequency domain is as follows: ; ; Step S16: Based on the frequency coupling characteristics, the output impedance model of the grid-side converter is represented by a two-dimensional matrix. Combining the DC-side bus voltage with the grid-side converter output voltage, the matrix relationship between the disturbance voltage and the grid-side converter output voltage can be obtained: ; ; in: R = I Z 2 / 12 U 1 2 P ; D(s) = I Z 2 - C Z P σ( s - jω 1); X + =3 U 1 2 + sL p ( P + jQ ); X - =3 U 1 2 + sL p ( P - jQ )σ( s - jω 1); Y + =3 U 1 2 + L p ( P + jQ )σ( s - jω 1); Y -=3 U 1 2 + L p ( P - jQ )σ( s - jω 1); σ( s - jω 1)= PH u ( s - jω 1) H i ( s - jω 1) / I Z .

[0035] Step S17, the effect of the disturbance voltage in the phase-locked loop on the output voltage of the grid-side converter is as follows: ; Step S18: Based on the data models of the above steps, the admittance model of the frequency coupling of the grid-side converter can be derived: ; ; In the formula: L m =diag[ sL p , ( s - jω 1) L p ], E It is the identity matrix; I ui This describes the effect of voltage disturbances on the output current response of the grid-side converter through the PI controller current loop.

[0036] Step S19, the impedance model for frequency coupling of the direct-drive wind turbine grid-side converter is: ; Preferably, in step S2, based on the single-unit impedance model of the direct-drive fan, as follows: Figure 2 Taking the actual wind farm shown as an example, the parameters of the direct-drive wind turbine are shown in Table 1. An equivalent impedance model of the direct-drive wind farm is established.

[0037] Step S21, as follows Figure 3 As shown, the definition Z D ( s , P ,Q The high-voltage side reduced impedance includes the parameters of the transformer substation. Z L1 For the collector line impedance, Z u Equivalent impedance for a single-unit branch: ; ; in, Y m1T The admittance of the transformer substation. Z 1T The impedance of the transformer is... k 1 represents the transformer ratio.

[0038] Step S22, direct-drive wind farm backbone link without branches, by 2 It consists of several branch roads, and each branch road is equipped with two ( n i =4) If direct-drive fans are connected in parallel, then their first... i Equivalent impedance of the branch Z T,i ( s , P i , Q j )for: ; Step S23, Equivalent impedance of the entire wind farm Z E ( s , P , Q )for: ; in P , Q for i × j The power matrix, P i They represent the first i The active and reactive power vectors of each wind turbine in the branch circuit.

[0039] Step S24, Equivalent impedance Z at the wind farm grid connection point PCC ( s , P , Q )for: ; ; in: Z C ( s , P, Q The high-voltage side reduced impedance includes the main transformer parameters. Z L2 To collect line impedance, Y E ( s , P , Q Equivalent admittance of the entire wind farm Y m2T Admittance as the main variable Z 2T The impedance of the main transformer. k 2 is the main variable ratio.

[0040] Preferably, in step S3, the impedance model of a single unit in a direct-drive wind farm is scanned and verified, and analyzed using the maximum Nyquist stability criterion. Finally, the IOOA algorithm is introduced for optimization.

[0041] Step S31, impedance scanning process as follows Figure 4 As shown, under ideal operating conditions, the overall power factor is set to 1, and the active power dispatch command for the wind farm is 0.8. P N ( P N The rated power is distributed to each unit in an equal manner. That is, when each wind turbine generates 4MW of active power, an impedance scan is performed on the direct-drive wind farm.

[0042] Step S32 introduces the maximum peak value Nyquist stability criterion, the principle of which is as follows: Figure 5 As shown. This criterion is used in conjunction with the impedance scan results of the wind farm for analysis. The stability criteria for the grid-connected system of the wind farm under different short-circuit ratios with different average distribution methods are as follows: Figure 6 As shown in the figure. It can be seen that as the short-circuit ratio... SCR As the voltage increases, the amplitude-frequency curve of the grid impedance shifts downward, causing the system intercept frequency to shift towards higher frequencies. In the low-frequency range, the system impedance ratio curve follows... SCR As the system moves further away from the forbidden circle, its stability improves. However, in the mid-frequency range, the impedance ratio curve... SCR It decreases and approaches the forbidden circle. When SCR When the impedance ratio drops to 2.5 or below, the impedance ratio curve enters the forbidden circle, and the system faces the risk of oscillation and instability under extremely weak power grid conditions.

[0043] Step S33: With the unit control parameters remaining unchanged, set the active power dispatch command for the wind farm to 0.8. P N ( P N (Rated power), the IOOA algorithm was used to optimize the active and reactive power output of each unit. The optimized power results of each unit are shown in Table 2.

[0044]

[0045] Preferably, in step S4, an impedance scan is performed on the wind farm after power allocation optimization, and the stability criteria of the grid-connected system under different short-circuit ratios are as follows: Figure 7 As shown, in SCR =1.5 and SCR =2.5 Under extremely weak power grid conditions, the Nyquist curve does not enter the forbidden circle and is in the safe and stable region, which proves the effectiveness of the proposed method in suppressing broadband oscillations under extremely weak power grid conditions.

Claims

1. A method for optimizing the output of a direct-drive wind farm turbine cluster based on an improved Osprey optimization algorithm, characterized in that... Includes the following steps: S1: Considering the dynamic modulation effect of active power and reactive power on the port impedance characteristics of the direct-drive wind turbine, establish a frequency coupling impedance model of the grid-side converter of the direct-drive wind turbine. S2: Based on the frequency coupling impedance model established in S1, establish the aggregate impedance model of the grid connection point of the direct-drive wind farm; S3: Under a given active power dispatch command, first use the average power allocation method to perform impedance scanning on the aggregate impedance model established in S2, use the maximum peak value Nyquist stability criterion to analyze the broadband oscillation risk, and then introduce the improved Osprey optimization algorithm (IOOA) to optimize the allocation of active and reactive power of each direct-drive wind turbine with the goal of maximizing safety margin.

2. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 1, characterized in that: It also includes S4: performing impedance scanning on the direct-drive wind farm with optimized power distribution to verify the broadband oscillation suppression effect.

3. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 2, characterized in that: S1 includes: modeling the main circuit model of the direct-drive wind turbine. In the time domain, assuming the grid-side converter initially operates under a three-phase symmetrical balanced condition, after injecting positive and negative sequence small signal disturbances at the grid connection point, due to the slip between the stator and rotor, the frequency of the rotor-side current component will shift with the slip. For phase A, the corresponding expressions of its voltage and current in the frequency domain are described as follows: （1）； （2）； （3）； （4）； In the above formula, and The voltage and current of phase a at different frequencies are respectively... f The disturbance component at the location; Indicates the fundamental voltage at frequency f The component at the location; Indicates the positive-sequence perturbation voltage at frequency f The component at the location; This indicates the negative sequence disturbance voltage at frequency f The component at the location; and These are the initial phase angles of the positive-sequence and negative-sequence disturbance voltages, respectively. U 1. U p and U n These represent the amplitudes of the fundamental voltage, positive-sequence perturbation voltage, and negative-sequence perturbation voltage, respectively. I 1. I p and I n These represent the amplitudes of the fundamental current, the positive-sequence perturbation current response, and the negative-sequence perturbation current response, respectively. f 1. f p , f n These are the fundamental frequency, the positive-sequence perturbation frequency, and the negative-sequence perturbation frequency, respectively. The perturbation frequency; Indicates the fundamental current at frequency f The component at the location; This indicates the positive-sequence disturbance current at frequency f The component at the location; This indicates that the negative sequence disturbance current is at a frequency of f The component at the location; , and These are the initial phase angles of the fundamental current, the positive-sequence disturbance current response, and the negative-sequence disturbance current response, respectively. This refers to the active power generated by the direct-drive fan. The reactive power generated by the direct-drive fan; It is an imaginary number.

4. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 3, characterized in that: A DC voltage loop model of a direct-drive wind turbine is constructed, ignoring the second harmonic component and performing convolution processing on each frequency component. The expression for the DC side voltage in the frequency domain is as follows: （5）； In the formula, This refers to the DC-side voltage in the frequency domain. This is the DC side voltage; This is the conjugate value of the fundamental frequency current; s =±j2π f p ; For DC side capacitors; For filtering inductors, This is the DC side current; A phase-locked loop (PLL) model of a direct-drive wind turbine is constructed, and the PLL tracks the phase angle of the grid voltage. θ 1 and small signal disturbance voltage phase angle θ ,Right now θ PLL = θ 1+ θ , θ PLL The phase angle is the output phase angle of the phase-locked loop; in the frequency domain, the small-signal voltage disturbance and the phase-locked loop control... θ The transfer function between them is: （6）； In the formula, For frequency f Voltage phase angle at point; transfer function T PLL ( s )= H PLL ( s ) / [1+ U 1[ f ] H PLL ( s )], H PLL (s) is the transfer function of the phase-locked loop. U 1[ f [The fundamental voltage at a frequency] f The component at the location; A current loop model of the direct-drive wind turbine is constructed, considering the influence of disturbance voltage on the output angle of the phase-locked loop. Transforming to the frequency domain, and using the convolution theorem, we obtain: （7）； （8）； In the above formula, The d-axis frequency domain component of the positive-sequence disturbance current; The q-axis frequency domain component of the positive-sequence disturbance current; The modulation stage of the direct-drive fan is modeled. The output signal of the current inner loop controller is modulated by Park. For phase a, the modulation signal contains two terms, and the specific expression in the frequency domain is as follows: （9）； （10）； In the above formula, This is the first modulation signal of phase a; This is the second modulation signal of phase a; This represents the frequency domain component of the d-axis voltage output by the current loop. This represents the frequency domain component of the q-axis voltage output by the current loop. This is the frequency domain representation of the phase-locked loop transfer function; Angular frequency; When the frequency is 0, the current loop outputs the d-axis voltage; When the frequency is 0, the current loop outputs the q-axis voltage.

5. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 4, characterized in that: Based on the frequency coupling characteristics, the output impedance model of the grid-side converter is represented by a two-dimensional matrix. Combining the DC-side bus voltage with the grid-side converter output voltage, the matrix relationship between the disturbance voltage and the grid-side converter output is obtained: （11）； （12）； In the above formula, This is the matrix showing the relationship between the disturbance voltage and the output current; This is the matrix showing the relationship between the disturbance voltage and the output voltage. Indicates resistance; Indicates the dynamic coupling phase on the DC side; Indicates frequency offset coupling phase; Represents coefficient matrix 1; Represents coefficient matrix 2; Represents coefficient matrix 3; Represents coefficient matrix 4; ( P + jQ () represents complex power; ( P - jQ () represents the conjugate of complex power; （13）； （14）； （15）； （16）； （17）； （18）； （19）； in, I Z Indicates direct current; U 1 represents the fundamental voltage; C Z For DC side capacitors; L p For filtering inductors; P Indicates active power; Represents the voltage loop transfer function; Represents the current loop transfer function; The effect of the disturbance voltage in the phase-locked loop on the output voltage of the grid-side converter is as follows: （20）； In the formula, This is the matrix showing the relationship between the disturbance voltage and the output voltage in the phase-locked loop; The transfer function of the phase-locked loop; Represents the current loop transfer function; This represents the decoupling coefficient.

6. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 5, characterized in that: Based on the data models (1), (5), (7), (8), (9), (10), and (20) of the above steps, the admittance model of the frequency coupling of the grid-side converter is derived: （21）； （22）； In the above formula, The admittance matrix represents the frequency coupling of the grid-side converter; Represents the identity matrix; This is the matrix showing the relationship between the disturbance voltage and the output voltage. This is the matrix showing the relationship between the disturbance voltage and the output voltage in the phase-locked loop; This is the matrix showing the relationship between the disturbance voltage and the output current; L m =diag[ sL p , ( s - jω 1) L p ], L m Represented as a diagonal matrix; L p For filtering inductors; sL p The first diagonal element; s - jω 1) L p The second diagonal element; This indicates the effect of voltage disturbances on the output current response of the grid-side converter through the PI controller current loop; Modulation gain; Represents the current loop transfer function; The frequency coupling impedance model of the grid-side converter for a direct-drive wind turbine is as follows: （23）； In the formula, This is the frequency coupling impedance matrix.

7. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 6, characterized in that: S2 includes: The spatial span of the power collection lines within the station is relatively short, and their parallel admittance has little impact on small disturbance stability. Therefore, they are simplified to a series impedance model with lumped parameters during modeling. definition The high-voltage side reduced impedance includes the parameters of the transformer substation. For the collector line impedance, The equivalent impedance of a single-unit branch; （24）； （25）； In the above formula, For the transformer's turns ratio, The impedance of the transformer is... The admittance of the transformer substation. The equivalent impedance of a single-unit branch; Assume a direct-drive wind farm backbone with no branches and no secondary branches, consisting of... x It consists of several branch roads, each branch road is composed of... n i If the direct-drive wind turbines are connected in parallel, then its first... i Equivalent impedance of the branch for: （26）； In the formula, For the first i The equivalent impedance of each branch; For the first i The total active power of the branch circuit; For the first i The total reactive power of the branch circuit; The equivalent impedance of a single-unit branch; For the first i The first branch road j The active power of the typhoon generator; For the first i The first branch road j The reactive power of the typhoon generator; Equivalent impedance of the entire wind farm for: （27）； （28）； In the above formula, The high-voltage side reduced impedance includes the parameters of the main transformer. The main variable is the ratio. This represents the equivalent admittance of the entire wind farm; The admittance of the main variable; The impedance of the main transformer; The impedance of the grid connection point; The impedance of the power transmission line.

8. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 6, characterized in that: In step S3, the impedance model of a single unit in a direct-drive wind farm is scanned and verified, and the maximum peak value Nyquist stability criterion is used for analysis. Finally, the IOOA algorithm is introduced for optimization. Specifically, this includes: 3.1: Positive and negative sequence harmonic voltage disturbances are sequentially injected at the grid connection point of the direct-drive wind farm, and the voltage and current time-domain waveform data at the grid connection point are collected in real time. The disturbance amplitude is set to 3% of the grid voltage amplitude. Next, characteristic frequency phasors are extracted based on the Fast Fourier Transform (FFT) algorithm. The FFT expression is as follows: （29）； In the formula: This refers to the voltage response amplitude. N To analyze the number of points; n As the starting point; T S Indicates the sampling period; f p Indicates the frequency of the disturbance; It is a complex exponential function; The calculated response increment is decoupled using the symmetrical component method, and the amplitude and phase characteristics of the positive and negative sequence impedances at specific frequencies are calculated based on the derivation of Ohm's law. Specifically, the symmetrical component method decoupling is as follows: （30）； In the formula: , and These are the positive-order, negative-order, and zero-order component phasors, respectively; 'a' is the basic operator. , and These are the three-phase vectors a, b, and c, respectively. Based on the derived formula of Ohm's law, the amplitude and phase characteristics of positive and negative sequence impedances at a specific frequency are calculated: （31）； （32）； In the formula: and These are the positive-sequence impedance and the negative-sequence impedance, respectively. and For positive-sequence disturbance voltage and current responses; and The voltage and current responses to negative-sequence disturbances; and These are the positive sequence impedance magnitude and phase angle; and The negative sequence impedance magnitude and phase angle; Finally, the above steps are repeated cyclically within the swept frequency band at a preset step size until the impedance characteristics of the entire frequency band are obtained.

9. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 8, characterized in that: S3 further includes: introducing the maximum peak value Nyquist stability criterion, and prohibiting the region radius. With system gain margin Phase margin θ The relationship is as follows: （33）； In the formula: It is the reciprocal of the distance from the intersection of the impedance ratio curve and the negative real axis to the origin. θ The angle between the intersection of the impedance ratio curve and the unit circle and the negative real axis.

10. The method for optimizing the output of a direct-drive wind farm cluster based on the improved Osprey optimization algorithm according to claim 9, characterized in that: In the Osprey Optimization Algorithm (OOA), the random initialization of the osprey population is defined as follows: （34）； （35）； In the above formula: X This represents the osprey location population matrix. X i Table 1 i The initial position of the osprey. N Indicates population size. m Indicates the number of problem variables. x ij Indicates the first j Among the problem variables, the first one is... i The initial position of the osprey. r ij A random number between 0 and 1. b l,j and b u, j They represent the first j The upper and lower boundaries of the problem variables; The OOA algorithm consists of two phases: development and exploration. Ospreys are ranked based on their fitness values, and those with higher fitness are considered a "school of underwater fish," represented as: （36）； In the formula: For the first i The fitness value of the osprey; Indicates the first Only the location of the osprey; For the first Only an osprey; Osprey population size; , They represent the first The and the first The objective function value of the Osprey; This indicates the position of the osprey that achieves the optimal objective function value; For union to be true; The Osprey's position is updated as follows: （37）； （38）； In the above formula: Indicates the first stage of development. j Among the problem variables, the first one is... i The new location for the osprey; Indicates the second stage of development j Among the problem variables, the first one is... i The new location for the osprey; Indicates the first j Among the problem variables, the first one is... i The location of the fish chosen by the osprey; A random number that is either 1 or 2; Indicates the first j The lower bound of each problem variable; Indicates the first j The upper limit of the problem variables; This represents the number of iterations. Indicates the exploration stage j Among the problem variables, the first one is... i The new location for the osprey; In the initialization phase of the OOA algorithm, a chaotic mapping is introduced to initialize the population in order to improve the convergence speed of the OOA algorithm. Differentiated mappings are performed for different intervals to ensure that the population is evenly distributed in the search space so as to find the optimal solution more quickly. （39）； In the formula: Indicates the upper bound of the decision variable; Indicates the step size scaling factor; This represents the chaotic state in the (n+1)th iteration; This represents the chaotic state value in the nth iteration; Indicates the control parameters for chaotic mapping; Considering the frequency coupling of a direct-drive wind farm grid-connected system, the objective function is defined according to the maximum peak value Nyquist stability criterion: （40）； （41）； In the above formula: The objective function is... For stability margin; and Penalty coefficients for active and reactive power; The reactive power squared penalty coefficient; and The active and reactive power generated by a single direct-drive wind turbine, 0 ≤ P k ≤ P max , | Q k |≤ Q max , P max Rated active power of direct-drive fan; Q max The rated reactive power of the fan; its size meets the single-unit capacity constraint. For the entire field's active power dispatch instructions; Number of direct-drive wind turbines at the station; This represents the impedance ratio of the power grid. The radius of the forbidden region in the Nyquist stability criterion for maximum peak value.

Images