Configuration method of non-array wind power plant energy storage system with multidirectional wake effect

Abstract

The invention relates to the technical field of wind power plant frequency support, and discloses a frequency coordination control method and system suitable for a non-array wind power plant in a complex terrain. Through multi-direction wake flow modeling and energy storage dynamic compensation technology, cluster energy storage optimization and complex terrain adaptability design are combined to realize full-period power regulation and control. According to the method, a multi-direction wake flow coupling model is established, and the frequency supporting precision is improved; an energy storage dynamic compensation mechanism is introduced, low-frequency oscillation is suppressed, and the frequency recovery time is shortened; an economic constraint-based cluster energy storage capacity optimization algorithm reduces the energy storage configuration cost; a terrain shielding effect compensation algorithm is developed, wake flow errors are reduced, and complex scene adaptability is enhanced; a full-period power smooth regulation and control strategy is constructed, the frequency recovery rate is larger than or equal to 0.3 Hz / s, and secondary drop is controlled within 0.05 Hz. The frequency supporting capacity and the system stability of the wind power plant under the complex terrain are remarkably improved, meanwhile, economical efficiency and reliability are considered, and the method is suitable for a power system with high-proportion renewable energy access.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind power system optimization energy storage system configuration to support system frequency control, in particular to a non-array wind farm optimization energy storage system configuration frequency support method considering the cumulative wake effect of eight characteristic wind directions. BACKGROUND

[0002] In recent years, with the rapid increase of wind power in the power grid, wind turbines are required to provide auxiliary services including frequency response. Unlike traditional synchronous generators, the output power of which is naturally coupled with the grid frequency, wind turbines, which are connected to the grid through power electronic converters, are decoupled from the grid frequency and cannot automatically respond to system frequency deviation. Therefore, power system operators in various countries have introduced technical specifications to clearly require wind turbine generators to be equipped with frequency support capability and participate in grid frequency regulation services.

[0003] In order to realize the frequency regulation of wind turbines to the system, the commonly used methods at present include frequency response control based on single machine model, multi-machine cooperative frequency support method based on wind turbine kinetic energy recovery, and frequency response method based on energy storage system assisting wind turbines. Most of these methods simplify the entire wind farm as a single machine model, which can only support the system frequency in the scenario where the wind farm adopts an array layout and the wind speed mainly comes from a single direction, and the wind speed change can be predicted by the traditional wake effect model. Because the individual differences in wind speed, operating conditions and geographical location of wind turbines and the electromechanical dynamic characteristics under frequency support are not considered, the single machine model cannot accurately describe the more complex terrain in the region, resulting in deviations in the dynamic modeling of wind turbine frequency, and even in extreme cases, it will exacerbate the frequency drop of the power grid.

[0004] In order to reduce the influence of wind farm wake effect on wind turbine frequency support capability, and to overcome the problems of over-simplified wind farm model and not considering the electromechanical dynamics of wind turbines under frequency support in current control methods, the present application proposes a non-array wind farm optimization energy storage system configuration frequency support method considering multi-directional wake effect. SUMMARY

[0005] The purpose of the present application is to provide a non-array wind farm energy storage system configuration method considering multi-directional wake effect to solve the problems existing in the prior art.

[0006] A non-array wind farm energy storage system configuration method considering multi-directional wake effect, comprising the following steps:

[0007] Step 1), a traditional Jensen wake effect model is established to calculate the wind speed loss of downstream wind turbines;

[0008] Step 2), establish a multi-direction wake effect model;

[0009] Based on the size and direction of the wind speed, the wind direction rose diagram obtained by using fuzzy C-means clustering method is arranged; then a single wind direction wake field model is constructed, and the classical Jensen wake model is modified by introducing the sum of squares superposition model method;

[0010] Step 3), establish a wind turbine frequency support dynamic model, derive the synchronous characteristic curve and frequency support margin;

[0011] Step 4), establish a multi-direction wake field influence evaluation system, calculate the FSM and critical frequency drop value of each wind turbine;

[0012] Step 5), group the wind turbines based on fuzzy K-means clustering to ensure similar frequency support characteristics within the group;

[0013] Step 6), analyze the influence of energy storage system on the frequency support capability of wind turbine, and establish an equivalent mechanical power model;

[0014] Step 7), propose an optimal configuration scheme for the capacity of energy storage system in wind turbine cluster, establish objective function and constraint conditions;

[0015] Step 8), determine the energy storage capacity of each cluster through optimization algorithm to ensure consistency and economy;

[0016] Step 9), verify the role of energy storage system in frequency support and speed recovery stage;

[0017] Step 10), adjust parameters and optimize system performance based on simulation results.

[0018] Further, the establishment of the traditional Jensen wake effect model in step 1) is as follows: assuming that there are N wind turbines in the wind farm, the wake velocity profile generated by the i-th wind turbine can be expressed as:

[0019] v i i = v0(1-δv i i)(1)

[0020] where v0 represents the original (free) wind speed, and δv i represents the total wind speed loss;

[0021] The total wind speed loss caused by multiple upstream wind turbines can be expressed as:

[0022]

[0023] where D j i represents the blade diameter of the j-th wind turbine; C Tjis the thrust coefficient of the jth wind turbine, which is used to describe the nonlinear relationship between the pitch angle and the tip speed ratio; x i -x j represents the distance between the jth upstream wind turbine and the ith downstream wind turbine; k represents the roughness coefficient; is the overlap area between the wake shadow area of the jth wind turbine and the swept area of the ith wind turbine, which can be calculated by the following formula:

[0024] where d ij represents the distance between the center of the downstream wind turbine and the center of the wake effect; L ij represents the distance between the center of the wake effect and the shadow area.

[0025] Further, in step 2), the multi-direction wake effect model is established as follows: for the downstream wind turbine located in the far wake area, the wind speed is represented as follows:

[0026] v x is the wind speed of the downstream wind turbine, x is the horizontal distance between the downstream wind turbine and the upstream wind turbine, r m is the downstream radius of the upstream wind turbine, η is the wake expansion coefficient, C T is the thrust coefficient.

[0027] Based on the size and direction of the wind speed, 8 groups of characteristic wind directions are identified by using the fuzzy C-means clustering method;

[0028] D c = [D c1 D c2 D c3 D c4 D c5 D c6 D c7 D c8 ] (5)

[0029] The characteristic wind speeds corresponding to each wind direction are as follows:

[0030]

[0031] v ij represents the jth characteristic wind speed under the ith characteristic wind direction, and n represents the total mode number of the characteristic wind direction; the wind direction rose diagram obtained by using the fuzzy C-means clustering method;

[0032] Then, a single wind direction wake field model is constructed, and the classical Jensen wake model is modified by introducing a square and superposition model method, and its expression is as follows:

[0033]

[0034] v i and v j represent the wind speed received by the i-th downstream wind turbine and the j-th upstream wind turbine, respectively, τ is the number of upstream wind turbines that have a direct influence on the i-th downstream wind turbine, v ji represents the wind speed of the j-th upstream wind turbine in the wake region flowing to the i-th downstream wind turbine;

[0035] According to the square sum superposition model and the wake effect model, the wind speed received by each wind turbine under the characteristic wind direction is as follows:

[0036] V i is the wind speed set received under the i-th characteristic wind direction. is the wind speed received by the j-th wind turbine under the i-th characteristic wind direction, and ξ is the total number of wind turbines in the wind farm.

[0037] Therefore, the single-direction wind speed receiving model under the multi-direction wake field effect is as follows:

[0038]

[0039] μ ij is a Gaussian membership function, and the expression is as follows:

[0040]

[0041] d j is the actual wind speed direction of the j-th wind turbine, D ci is the distribution expectation value of the i-th characteristic wind direction, and σ is the width of the Gaussian membership function.

[0042] Further, the step 3) is specifically for the wind turbine in the synchronous operation state, and the following formula is used:

[0043] T a is a mechanical time constant, P M and P E are mechanical power and electromagnetic power, D is a damping coefficient, ω r is the rotor speed, and ω r is replaced by ω , and the following is obtained:

[0044] The power angle dynamic equation of the wind turbine is as follows:

[0045]

[0046] ω sf is the output instruction value of the P-f droop frequency regulation control, and ω gis the grid synchronous angular velocity, f0is the grid frequency, in P-f droop control, it can be obtained that:

[0047] w sf -w0= K p (P ref -P E )(14)

[0048] ω0is the rated speed, K p is the droop coefficient, P ref is the active power instruction value of the frequency response output of the load shedding curve, as follows:

[0049] K DL is the load shedding coefficient, the load shedding control measure of the wind power generator aims to ensure that it can release extra power to provide frequency support; during the load shedding operation, the rotor speed is fixed at the overspeed operation point, denoted as A', B' and C';

[0050] The optimal mechanical power expression is:

[0051] K W represents the optimal energy conversion rate of the wind power generator in the maximum power point tracking operation state, which increases with the increase of the wind speed; by substituting the formulas (14)-(16) into (12)-(13), it can be obtained that:

[0052]

[0053] It can be seen from the formulas (17a-b) that the rotor angular dynamics and the grid power angle dynamics are coupled with each other.

[0054] Further, the step 4) is specifically that the synchronization characteristics of the wind power generator are determined by the intersection points of the equations (17a)-(17b), that is, the existence of the intersection points or the existence of the overlapping area between the two curves can ensure that the wind power generator is synchronized with the grid under the conditions of under-frequency or over-frequency;

[0055] The size of the overlapping area can be used as a key index for measuring the frequency support capability of a single wind power generator, which is called frequency support margin (FSM):

[0056]

[0057] Wherein

[0058] The critical frequency drop value is:

[0059]

[0060] Furthermore, in step 5), to better cluster based on the frequency support capability of wind turbines, a fuzzy K-means clustering method is adopted, and its clustering process is as follows:

[0061] First, obtain the original dataset;

[0062] Based on the above characteristic wind direction analysis, a frequency support capacity vector (FSCV) is defined for a single wind turbine:

[0063] FSM k (k = 1, 2, ... 8) is the frequency support margin for the k-th characteristic wind direction. It is the critical frequency drop value of the k-th characteristic wind direction, and γ is the dimensionless normalization coefficient, which is used to make the FSM k (k = 1, 2, ... 8) and The magnitudes are consistent, thus ensuring that the two are comparable;

[0064] For all the wind turbines in the wind farm, the original dataset is:

[0065] A = (A1, A2, ..., A N ) (twenty one)

[0066] Then, initialize the cluster centers and the membership matrix of each wind turbine to the corresponding cluster group;

[0067] Suppose that N wind turbines are divided into K clusters, denoted as C1, C2, ..., C60. K The centers of each cluster are R1, R2, ..., Rn. K The initial membership matrix is ​​denoted as U = [u ij ] N×K , where u ij Let represent the membership function of the i-th wind turbine to the j-th cluster group; furthermore, the fuzzy index is denoted as m > 1; the objective function is defined as follows:

[0068]

[0069] Among them, the Euclidean norm is used to represent A. i With R j The distance between them.

[0070] Finally, update the cluster center / membership matrix;

[0071] Optimal clustering can be achieved by minimizing the objective function equation (22) using the Lagrange multiplier method. Therefore, the membership function and cluster centers can be updated alternately. With the current cluster centers fixed, the membership function is updated as follows:

[0072]

[0073] When the membership function is updated once according to equation (23), the clustering center will be updated as follows:

[0074] With each round of iteration, the objective function J m will gradually decrease, and the iteration process is terminated when the following condition is met:

[0075]

[0076] where ε is the termination threshold;

[0077] Finally, the cluster group to which the i-th wind turbine belongs can be determined by selecting the column with the maximum membership function value in the i-th row in the membership matrix U = [u ij ] N×K .

[0078] Further, step 6) includes dynamic evolution of the rotational speed;

[0079] When a certain wind turbine is allocated an energy storage system with a capacity of P ESS , for the grid synchronization characteristic equation (17), this is equivalent to increasing the mechanical power by P ESS . The electromagnetic power shortage caused by insufficient mechanical power can be compensated by the energy storage system; under this equivalent condition, the dynamic evolution of the rotational speed of the wind turbine is as follows:

[0080]

[0081] When the provided energy storage system power can restore transient balance, the frequency support margin of the energy storage system can be approximately expressed as follows:

[0082]

[0083] where ω1, ω2, η and formula (18) are the same as in formula (18);

[0084] In addition to providing additional power to improve the equivalent frequency support margin of the wind turbine during the frequency recovery phase, the energy storage system can also assist the wind turbine during the rotational speed recovery phase to prevent secondary frequency drop;

[0085] When the rotor kinetic energy of the wind turbine is extracted at time period t0 for frequency support, the rotational speed will continue to decrease to time period t1; during this phase, the maximum energy released can be expressed as:

[0086]

[0087] where ω r0 is the initial rotational speed, and ω rcis the critical speed, J is the moment of inertia. At time period t1, when the speed is about to reach the critical value ω rc (or the grid frequency has recovered), the speed will start to recover, the rotor absorbs energy from the wind to increase the angular speed, resulting in a decrease in power output; at time period t2, the speed reaches the highest value and starts to converge to the initial level;

[0088] At t1, the speed needs to recover; therefore, the energy storage system will release a constant power P Ecom to compensate for the reduced wind turbine power, the calculation formula is:

[0089]

[0090] When the speed increases to the peak value, the power reference value P E1 of the energy storage system should be reduced with the slope of k ramp to ensure the power balance between the load and the resource, in order to reduce the impact on the system frequency, an additional frequency response loop is added, which is related to the frequency deviation after time period t2; finally, the energy storage system power output instruction is P E1 + P Ef .

[0091] Further, the step 7) is specifically that the energy storage systems are allocated to the wind turbine clusters, and each energy storage system will be responsible for providing support for the wind turbines in the same cluster;

[0092] When allocating the energy storage systems, the key consideration is to ensure that the wind turbines in the wind farm have high consistency in frequency support margin, so that all wind turbines can respond to the grid frequency change in a coordinated manner without causing any wind turbine to be disconnected from the grid in the case of serious frequency drop; therefore, the optimal energy storage system capacity configuration scheme is:

[0093]

[0094] wherein α and β represent the weight coefficients of the consistency index and the total cost respectively; σ is the unit power cost of the energy storage system; P ESS,j is the power capacity of the energy storage system allocated to the jth wind turbine cluster; I Coherency is the consistency index, which reflects the similarity of the wind turbines in terms of frequency support margin, and its calculation is based on the difference between the frequency support margin value of each wind turbine and the average frequency support margin value under K wind turbine groups and 8 characteristic wind directions; represents the frequency support margin value of the jth wind turbine under the ith characteristic wind direction; represents the frequency support margin value of the lth wind turbine when equipped with an energy storage system under the ith characteristic direction; represents the frequency support margin value of the ith wind turbine in the ith characteristic direction without the energy storage system; C j represents the total number of wind turbines in the jth wind farm; represents the total budget; represents the power shortage value leading to frequency drop Δf; t 1,l and t 2,l respectively represent the start time and end time of the speed recovery of the lth wind turbine; t 0,l , t 1,l and t 2,l respectively represent the time when the lth wind turbine starts to reduce the speed, recovers the speed and completes the speed recovery (consistent with the time mark in Figure 10 ); represents the time period of the earliest wind turbine in the jth wind farm to start recovering the speed, represents the time period of the last wind turbine in the jth wind farm to complete the speed recovery.

[0095] Further, in order to ensure that the energy storage system can provide sufficient power to recover the speed of the wind turbine, the formula is as follows:

[0096]

[0097] Ensure that the energy storage system always maintains sufficient energy to support frequency stability control

[0098]

[0099] 1=α+β,0<α,β<1(30n).

[0100] Further, in step 8), in the calculation of the consistency index, the integral of the frequency support margin index in the time interval t0 to t1 is adopted to ensure that the wind turbine has a higher consistency level in the whole frequency support process. In order to simplify the solving process of the optimization problem, the consistency index is converted into a discrete form as follows:

[0101]

[0102] Where, Δt is the discrete time step, n l is the total discrete step number, and the calculation formula is n l = (t 1l -t 0l ) / Δt.

[0103] Compared with the prior art, the present application provides a multi-directional wake effect non-array wind farm energy storage system configuration method, which has the following beneficial effects:

[0104] 1. Multi-directional wake modeling: significantly improve the frequency support accuracy and reduce the wake calculation error.

[0105] 2. Energy storage dynamic compensation: enhance grid frequency stability, prevent low frequency off-grid, shorten recovery time.

[0106] 3. Cluster energy storage optimization: reduce energy storage capacity by 35%, balance economy and reliability.

[0107] 4. Complex terrain adaptation: reduce wake shadow error by 42%, improve robustness in complex scenarios.

[0108] 5. Full-cycle regulation: achieve smooth power output, frequency recovery rate increased to 0.3Hz / s, secondary drop controlled within 0.05Hz. BRIEF DESCRIPTION OF DRAWINGS

[0109] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0110] Figure 1 The theoretical basis model of the present application is the traditional Jensen wake effect model

[0111] Figure 2 The non-array wind farm of the present application

[0112] Figure 3 The wind direction rose of the present application

[0113] Figure 4 The wind speed distribution graph of the present application under the condition of considering multi-directional wake effect and not considering multi-directional wake effect

[0114] Figure 5 The wind turbine load reduction curve of the present application

[0115] Figure 6 The wind turbine synchronization characteristic curve of the present application under the condition of fixed wind speed and different grid frequency deviation

[0116] Figure 7 The wind turbine synchronization characteristic curve of the present application under the condition of considering multi-directional wake effect and not considering multi-directional wake effect

[0117] Figure 8 The frequency support margin of the wind turbine of the present application under 8 characteristic wind directions

[0118] Figure 9 The influence of the energy storage system of the present application on the synchronization characteristics of the wind turbine

[0119] Figure 10 Active power and rotor speed of the wind turbine in frequency support of the present application;

[0120] Figure 11 Control scheme of the energy storage system in the process of speed recovery of the present application. DETAILED DESCRIPTION

[0121] The technical solutions in the embodiments of the present application will be described clearly and completely below with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the protection scope of the present application.

[0122] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with the drawings and specific embodiments.

[0123] The frequency support method for optimizing the energy storage system configuration of the non-array wind farm under the influence of the cumulative wake effect of eight characteristic wind directions includes the following steps:

[0124] 1) Establish a traditional Jensen wake effect model, and the detailed content is as follows:

[0125] At present, many wind turbines are installed in vast plain areas in an array layout. In this array layout, the wind speed received by the downstream wind turbine will be directly affected by the upstream wind turbine, so that the working points of the wind turbines in the same wind farm are different. The wind speed reduction effect caused by the upstream wind turbine is called wake effect. One of the most common models reflecting the wake effect is the Jensen model. The Jensen model is based on the assumption that the wake effect will linearly expand along the downstream direction, as shown in Figure 1 The wake velocity profile generated by the ith wind turbine in the wind farm can be expressed as:

[0126] v i =v0(1-δv i ) (1)

[0127] wherein v0 represents the original (free) wind speed, and δv i represents the total wind speed loss. For the downstream wind turbine, the wind speed it receives is not only affected by the wind turbine directly in front, but also affected by other upstream wind turbines. Therefore, the total wind speed loss caused by multiple upstream wind turbines can be expressed as:

[0128]

[0129] wherein D jD represents the blade diameter of the jth wind turbine; C Tj is the thrust coefficient of the jth wind turbine, used to describe the nonlinear relationship between the pitch angle and the tip speed ratio; x i -x j D represents the distance between the jth upstream wind turbine and the ith downstream wind turbine; k represents the roughness coefficient; is the overlap area of the wake shadow area of the jth wind turbine and the swept area of the ith wind turbine, which can be calculated by the following formula:

[0130]

[0131] where d ij D represents the distance between the downstream wind turbine center and the wake effect center; L ij D represents the distance between the wake effect center and the shadow area.

[0132] 2) Establish a multi-direction wake effect model, the details are as follows:

[0133] Figure 2 For the non-array wind farm in the present application, as shown in the figure, the wind farm is distributed on the ridge of the mountain area.

[0134] For the downstream wind turbine located in the far wake area, the wind speed can be represented as follows:

[0135] v x is the wind speed of the downstream wind turbine, x is the horizontal distance between the downstream wind turbine and the upstream wind turbine, r m is the downstream radius of the upstream wind turbine, η is the wake expansion coefficient, C T is the thrust coefficient.

[0136] The present application collects the annual wind speed data of the target wind farm from January 1, 2020 to January 1, 2024, and identifies 8 groups of characteristic wind directions based on the size and direction of the wind speed using the fuzzy C-means clustering method.

[0137] D c = [D c1 D c2 D c3 D c4 D c5 D c6 D c7 D c8 ] (5)

[0138] The characteristic wind speed corresponding to each wind direction is as follows:

[0139]

[0140] v ijVij represents the jth characteristic wind speed under the ith characteristic wind direction, and n represents the total number of characteristic wind directions. Figure 3 The wind direction rose diagram obtained by using the fuzzy C-means clustering method.

[0141] Then the single wind direction wake field model is constructed, and the classical Jensen wake model is modified by introducing the square and superposition model method, and its expression is:

[0142]

[0143] v i and v j represent the wind speed received by the ith downstream wind turbine and the jth upstream wind turbine, respectively, τ is the number of upstream wind turbines that have a direct impact on the ith downstream wind turbine, v ji represents the wind speed of the jth upstream wind turbine in the wake area flowing to the ith downstream wind turbine.

[0144] According to the square and superposition model and the wake effect model, the wind speed received between wind turbines under the characteristic wind direction is as follows:

[0145] V i is the wind speed set received under the ith characteristic wind direction. is the wind speed received by the jth wind turbine under the ith characteristic wind direction, and ξ is the total number of wind turbines in the wind farm.

[0146] Therefore, the single-direction wind speed receiving model under the multi-direction wake field effect is:

[0147]

[0148] μ ij is the Gaussian membership function, and its expression is as follows:

[0149] d j is the actual wind speed direction of the jth wind turbine, D ci is the distribution expectation value of the ith characteristic wind direction, and σ is the width of the Gaussian membership function.

[0150] 3) Establish a wind turbine frequency support dynamic model, the details are as follows:

[0151] For wind turbines in synchronous operation state, the following exists:

[0152] T a is the mechanical time constant, P M and P E are the mechanical power and electromagnetic power, D is the damping coefficient, ω r is the rotor speed, let Substitute ω r , we can get:

[0153] The power angle dynamic equation of the wind turbine is:

[0154]

[0155] ω sf is the output command value of P-f droop frequency regulation control, ω g is the grid synchronous angular velocity, f0 is the grid frequency, in P-f droop control, we can get:

[0156] w sf -w0=K p (P ref -P E ) (14)

[0157] ω0 is the rated speed, K p is the droop coefficient, P ref is the active power command value of the load shedding curve frequency response output, as follows:

[0158] K DL is the load shedding coefficient, the load shedding control measure of the wind turbine aims to ensure that it can release additional power to provide frequency support. The load shedding curve of the wind turbine is shown in Figure 5 , during the load shedding operation, the rotor speed is fixed at the overspeed operating point, denoted as A', B' and C'. The optimal mechanical power expression is:

[0159] K W represents the optimal energy conversion rate of the wind turbine in the maximum power point tracking operating state, which increases with the increase of wind speed. Substitute formulas (14)-(16) into (12)-(13), we can get:

[0160] From formulas (17a-b), it can be seen that the rotor angle dynamics and the grid power angle dynamics are coupled with each other.

[0161] Based on formula (17), under the condition of fixed wind speed and different grid frequency deviations, the synchronous characteristic curve of the wind turbine can be plotted in the P E -ω r plane, as shown in Figure 6 . In Figure 6 , the black curve represents the solution of equation , which describes how the electromagnetic power changes with the mechanical power in order to keep the speed stable at different speeds; the blue curve represents equation The solution, the curve, describes how the electromagnetic power changes with the rotational speed to maintain a stable power angle of the wind turbine. Therefore, to maintain stable rotational speed and power angle of the wind turbine, the wind turbine must operate at the equilibrium point (i.e., Figure 6 The intersection point (in the graph). Simultaneously, as the grid frequency decreases or increases, the blue curve shifts to the left or right relative to the original curve (the solid blue line). It can be clearly observed that when the grid frequency increases, the system always has a transient equilibrium point. However, when the grid frequency decreases to a certain level, the wind turbines may lose synchronization, at which point the two curves no longer intersect (when Δf < f). critical ),like Figure 3 As shown.

[0162] 4) Establish a multi-directional wake field impact assessment system, the details of which are as follows:

[0163] Based on the above discussion, the synchronization characteristics of wind turbines are determined by the intersection of equations (17a) and (17b). That is, the existence of an intersection point or an overlapping region between the two curves can ensure that the wind turbine remains synchronized with the power grid under under-frequency or over-frequency conditions. Figure 6 The black curve in the graph is related to instantaneous wind speed: as wind speed increases, K... W As wind speed increases, the curve rises; conversely, as wind speed decreases, the curve falls. Therefore, considering the multidirectional wake effect, the actual frequency support capability of a wind turbine will be lower than the ideal frequency support capability when the multidirectional wake effect is ignored.

[0164] The impact of multidirectional wake effects on the synchronization characteristics of wind turbines, such as Figure 7 As shown, the multidirectional wake effect affects the synchronization characteristics of wind turbines in two ways.

[0165] On the one hand, the decrease in wind speed causes the mechanical power curve to drop (from the green curve to the red solid line), thus reducing the overlap between the two curves, indicating a reduced margin for system stability. On the other hand, the decrease in received wind speed weakens the wind turbine's ability to withstand grid frequency drops; that is, considering the effects of multidirectional wake effects, previously tolerable grid frequency variations may become unbearable. Therefore, Figure 6 The size of the overlapping region can be used as a key indicator to measure the frequency support capability of a single wind turbine, and is called the frequency support margin (FSM).

[0166]

[0167] in

[0168] The critical frequency drop value is: As can be seen from equation (18), as the wind speed decreases, the overlapping area of the curves decreases, which indicates that the frequency support margin of the system decreases, and thus it is easier to lose synchronism. For the wind farm studied (e.g. shown in Figure 2 ) and its wind speed distribution (e.g. shown in Figure 4 ), the frequency support margin is calculated with and without considering the multi-directional wake effect, and the results are shown in Figure 8 .

[0169] In Figure 8 , the results of the 25 wind turbines shown in Figure 2 are presented, and the results under 8 typical wind directions are compared with and without considering the multi-directional wake effect. When considering the multi-directional wake effect, the FSM of the wind turbine synchronization curve is smaller than that without considering the multi-directional wake effect. The percentage of FSM reduction ranges from 8.5% to 41.2%, indicating that the multi-directional wake effect indeed has a significant impact on the stability margin of the system, and therefore, the multi-directional wake effect should be considered when designing the frequency support scheme of the wind farm.

[0170] 5) Wind turbine clustering based on fuzzy K-means, the details are as follows:

[0171] Before configuring the capacity of the energy storage system in the wind farm, the wind turbines must be divided into different groups according to their similarity. Due to resource and budget constraints, multiple wind turbines need to share one energy storage system; in addition, in order to reduce the complexity of system control, wind turbines sharing the same energy storage system need to have similar frequency support characteristics so that an equivalent model can be used to describe multiple wind turbines. In the process of wind turbine clustering, multiple factors such as internal parameters of wind turbines, geographical location, power capacity, received wind speed, etc. can be considered. However, in this invention, the focus is on the frequency support margin (FSM) of wind turbines under received wind speed. Instead of directly using wind speed as a similarity index, this invention constructs a similarity index based on FSM and wind turbine critical frequency drop value, because they can not only reflect the internal parameters of wind turbines (damping coefficient, load shedding coefficient, time constant), but also reflect the current working state (wind speed and frequency deviation).

[0172] Therefore, in order to better cluster based on the frequency support capability of wind turbines, the fuzzy K-means clustering method is used, and the clustering process is as follows:

[0173] First, the original data set is obtained.

[0174] Based on the above feature wind direction analysis, the frequency support capability vector (FSCV) is defined for a single wind turbine:

[0175] FSM k (k=1,2,...8) is the frequency support margin of the kth characteristic wind direction, is the critical frequency drop value of the kth characteristic wind direction, and γ is a dimensionless normalization coefficient, which is used to make the order of magnitude of k (k=1,2,...8) and uniform, so as to ensure the comparability of the two.

[0176] Therefore, for all wind turbines in the wind farm, the original data set is:

[0177] A=(A1,A2,…,A N ) (21)

[0178] Then, the clustering centers and the membership matrix of each wind turbine to the corresponding cluster group are initialized.

[0179] Suppose that N wind turbines are divided into K cluster groups, denoted as C1,C2,...,C K , and the corresponding centers are R1,R2,...,R K , respectively. The initial membership matrix is denoted as U=[u ij ] N×K , where u ij represents the membership function of the ith wind turbine to the jth cluster group. In addition, the fuzzy index is denoted as m>1. The objective function is defined as follows:

[0180]

[0181] where the Euclidean norm is used to represent the distance between A i and R j .

[0182] Finally, the clustering centers / membership matrix are updated.

[0183] The optimal clustering can be achieved by minimizing the objective function equation (22) through the Lagrange multiplier method. Therefore, the membership function and the clustering center can be updated alternately. When the current clustering center is fixed, the membership function is updated as follows:

[0184]

[0185] When the membership function is updated once according to equation (23), the clustering center will be updated accordingly as follows:

[0186]

[0187] With each round of iteration, the objective function J m will gradually decrease, and the iteration process is terminated when the following condition is met:

[0188]

[0189] where ε is the termination threshold. Finally, the cluster group to which the ith wind turbine belongs can be determined by selecting the column with the maximum membership function value in the ith row of the membership matrix U = [u ij ] N×K .

[0190] 6) Analysis of the impact of energy storage systems on the frequency support capability of wind turbines, with details as follows:

[0191] When wind turbines participate in grid frequency support, they will go through two stages: first, the kinetic energy release stage, during which the rotor speed of the wind turbine decreases to provide more power to the grid; then the speed recovery stage, during which the wind turbine increases the speed to the allowable range by absorbing wind energy, but the output power decreases relatively. Therefore, there may be two different risks in the whole process. In the first stage, if the actual wind speed received by the wind turbine is lower than the expected value, the wind turbine may not be able to maintain synchronization in some specific cases of grid frequency drop, which may lead to disconnection. In the second stage, due to the decrease of wind turbine output power, the grid may experience a secondary frequency drop.

[0192] Therefore, the configuration of energy storage system capacity in wind farms may be an effective measure to improve the stability of frequency control schemes based on wind turbines. Now assume that a wind turbine is allocated a capacity of P ESS energy storage system, then for the grid synchronization characteristic equation (17), it is equivalent to increase the mechanical power by P ESS The shortage of electromagnetic power caused by insufficient mechanical power can be compensated by the energy storage system. Under this equivalent condition, the dynamic evolution of the rotor speed of the wind turbine is as follows:

[0193]

[0194] Correspondingly, in the case of severe frequency drop, Figure 6 the black curve in Figure 9 will move up, thus providing more frequency support margin for the wind turbine, as shown in .

[0195] Figure 9In the diagram, the red curve represents the rotational speed dynamics when the wind speed is the expected value V1, without considering the multi-directional wake effect. When the multi-directional wake effect is considered, the actual wind speed decreases to V1-δV1, causing the rotational speed dynamic curve to shift down to the black curve, resulting in a reduced frequency support margin. Simultaneously, the transient equilibrium point changes from point D to D'. With the additional power provided by the energy storage system, the equivalent rotational speed dynamic curve rises from the black curve to the yellow curve, restoring the transient equilibrium point from D' to D. More importantly, the frequency support margin also increases to a certain level. Assuming the provided energy storage system power can reach... Figure 9 If the transient equilibrium is restored as shown, the frequency support margin of the energy storage system can be approximately expressed as follows:

[0196]

[0197] ω1, ω2, η are the same as in formula (18).

[0198] In addition to providing extra power during the frequency recovery phase to improve the equivalent frequency support margin of wind turbines, energy storage systems can also assist wind turbines during the speed recovery phase to prevent secondary frequency drops. The entire frequency support process is as follows: Figure 10 As shown. During this process, the rotor kinetic energy of the wind turbine is extracted for frequency support in time period t0, and the rotational speed will continue to decrease to time period t1. The maximum energy released during this stage can be expressed as:

[0199]

[0200] Where, ω r0 Let ω be the initial rotational speed. rc Let ω be the critical speed and J be the moment of inertia. In time interval t1, when the speed is about to reach the critical value ω... rc When (or the grid frequency has been restored), the rotational speed will begin to recover, and the rotor will absorb energy from the wind to increase its angular velocity, resulting in a decrease in power output. At time t2, the rotational speed reaches its maximum value and begins to converge back to its initial level.

[0201] To assist wind turbines in frequency support and speed recovery, the energy storage system control scheme throughout the process is as follows: Figure 11 As shown. At time t1, the rotational speed needs to recover. Therefore, the energy storage system will release a constant power P. Ecom To compensate for the reduced wind turbine power, the calculation formula is as follows: When the rotational speed increases to its peak value, the energy storage system power reference value P E1 It should be k rampthe slope is reduced to ensure the power balance between the load and the resource. At the same time, in order to reduce the impact on the system frequency, an additional frequency response loop is added, which is related to the frequency deviation after the time period t2. Finally, the energy storage system power output instruction is P E1 +P Ef

[0202] 7) Optimization of energy storage system capacity in wind turbine cluster, the details are as follows:

[0203] According to the above discussion, the energy storage system can enhance the frequency support capability of a single wind turbine and improve the overall reliability of the system. Therefore, the current problem to be solved is how to determine the capacity of the energy storage system allocated to each wind turbine cluster. In practical application, due to the limitation of investment and maintenance cost, it is not realistic to configure energy storage system for each wind turbine. Therefore, the energy storage system will be allocated to the wind turbine cluster, and each energy storage system will be responsible for providing support for the wind turbines in the same cluster.

[0204] When allocating the energy storage system, the key consideration is to ensure that the wind turbines in the wind farm have high consistency in frequency support margin, so that all wind turbines can respond to the change of grid frequency in the case of serious frequency drop, without causing any wind turbine to be off-grid. Therefore, the proposed optimal energy storage system capacity configuration scheme is:

[0205]

[0206]

[0207] 1 = a + b, 0 < a, b < 1 (30n)

[0208] where a and b represent the weight coefficients of consistency index and total cost respectively; s is the unit power cost of energy storage system; P ESS,j is the power capacity of the energy storage system allocated to the jth wind turbine cluster; I Coherency is the consistency index, which reflects the similarity of wind turbines in frequency support margin, and its calculation is based on the difference between the frequency support margin value of each wind turbine and the average frequency support margin value of K wind turbines and 8 characteristic wind directions; represents the frequency support margin value of the jth wind turbine under the ith characteristic wind direction; represents the frequency support margin value of the lth wind turbine under the ith characteristic direction when equipped with energy storage system; represents the frequency support margin value of the lth wind turbine under the ith characteristic direction when not equipped with energy storage system; C j is the total number of wind turbines in the jth wind turbine cluster; represents the total budget; represents the power shortage value leading to the frequency drop; t 1,l and t 2,l respectively represent the start time and end time of the speed recovery of the lth wind turbine; t 0,l , t 1,l and t 2,l respectively represent the time when the lth wind turbine starts to reduce the speed, recovers the speed and completes the speed recovery (consistent with the time mark in Figure 10 ); represents the time period of the earliest wind turbine in the jth wind turbine group that starts to recover the speed, represents the time period of the last wind turbine in the jth wind turbine group that completes the speed recovery.

[0209] In the above optimization process, equation (30g) aims to ensure that the energy storage system can provide sufficient power to recover the speed of the wind turbine group; and equations (30k)-(30m) aim to ensure that the energy storage system always maintains sufficient energy to support the frequency stability control. In the calculation of the consistency index, the integral of the frequency support margin index in the time interval t0 to t1 is adopted to ensure that the wind turbine group has a higher consistency level in the entire frequency support process. In order to simplify the solving process of the optimization problem, the consistency index is converted into a discrete form as follows:

[0210]

[0211] where, Δt is the discrete time step, n l is the total number of discrete steps, and the calculation formula is n l = (t 1l -t 0l ) / Δt.

[0212] The present application can accurately model the multi-direction wake effect, and improve the frequency support accuracy

[0213] By introducing fuzzy C-means clustering to identify 8 groups of characteristic wind directions, the traditional Jensen wake model is corrected, and a multi-direction wake field model considering the wind direction shielding effect (equations 4-10) is established. Compared with the single wind direction assumption, this model can more truly reflect the wind speed difference between wind turbines in the wind farm, reduce the calculation error of the frequency support margin (FSM) by 8.5%-41.2%( Figure 8 ), and avoid the frequency support deviation caused by the simplification of the wake effect.

[0214] The present application can dynamically compensate the energy storage system, and enhance the frequency stability of the power grid

[0215] The energy storage system releases power to compensate for the mechanical power shortage during the frequency drop stage (equations 26-28), so that the transient equilibrium point of the synchronous characteristic curve is shifted upwards( Figure 9), the FSM is expanded. Simulation shows that the energy storage system can increase the critical frequency drop value (Δf_CFD) from 0.25Hz without energy storage to 0.32Hz, effectively preventing the wind turbine from being disconnected due to low frequency. In the speed recovery stage, the energy storage system suppresses the secondary frequency drop through constant power compensation (formula 30h) and shortens the recovery time to 20 seconds Figure 10

[0216] The application can cluster energy storage optimization configuration, taking into account economy and reliability

[0217] The application proposes an optimization objective function (formula 30a) based on the consistency index (I_Coherency) and unit cost (σ), and divides 25 wind turbines into 5 groups through fuzzy K-means clustering Figure 2 ), each group shares an energy storage system. After optimization, the total capacity of the energy storage is reduced by 35%, while ensuring that the standard deviation of FSM in the group is ≤5%, achieving Pareto optimization of cost and performance. The constraint conditions (formula 30g-m) ensure that the energy storage power (500kW≤P_ESS,j≤2MW) and energy (2-5MWh) are adapted to the load reduction requirements of the wind turbine, and the SOC is maintained in the healthy range of 20%-80%.

[0218] The application can adapt to complex terrain

[0219] The application is aimed at mountainous wind farms Figure 2 ), introduces a Gaussian membership function (formula 10) to integrate the influence of multi-directional wind speed, and reduces the calculation error of wake shadow area by 42% compared with the traditional Jensen model. Combined with the wind direction rose diagram Figure 3 ) and wind speed distribution Figure 4 ), the optimized model maintains FSM>0.15 under 8 characteristic wind directions, which is more than 2 times higher than the conventional method, significantly improving the robustness in complex terrain scenarios.

[0220] The application can dynamically regulate the whole cycle process

[0221] The application controls the rotor kinetic energy release (t0-t1), speed recovery (t1-t2), and SOC balance (after t2) in three stages Figure 11 ), and the energy storage system realizes time sequence optimization of power output. Simulation shows that this strategy can increase the frequency recovery rate from 0.15Hz / s to 0.3Hz / s, and the secondary frequency drop amplitude is controlled within 0.05Hz, meeting the requirements of the grid frequency safety guide.

[0222] The above-described embodiments are only preferred modes of the application and do not limit the scope of the application. Without departing from the design spirit of the application, various modifications and improvements to the technical solutions of the application made by those skilled in the art should fall within the protection scope of the claims of the application.​

Claims

1. A configuration method for a non-array wind farm energy storage system with multi-directional wake effect, characterized in that, Includes the following steps: Step 1) Establish a traditional Jensen wake effect model and calculate the wind speed loss of the downstream wind turbine; Step 2), establish a multi-directional wake effect model; Based on wind speed and direction, a wind rose diagram was obtained by using fuzzy C-means clustering. Then, a single-wind-direction wake field model was constructed, and the classic Jensen wake model was modified by introducing the sum of squares superposition model method; Step 3) Establish a dynamic model of wind turbine frequency support and derive the synchronization characteristic curve and frequency support margin; Step 4) Establish a multi-directional wake field impact assessment system and calculate the FSM and critical frequency drop value of each wind turbine; Step 5) Group the wind turbines based on fuzzy K-means clustering to ensure that the frequency support characteristics within the group are similar; Step 6) Analyze the impact of the energy storage system on the frequency support capability of the wind turbine and establish an equivalent mechanical power model; Step 7) Propose an optimized configuration scheme for the energy storage system capacity in the wind turbine cluster, and establish the objective function and constraints; Step 8) Determine the energy storage capacity of each cluster through an optimization algorithm to ensure consistency and economy; Step 9) Verify the role of the energy storage system in the frequency support and speed recovery phases; Step 10): Adjust parameters based on simulation results to optimize system performance.

2. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 1, characterized in that: Step 1) establishes the traditional Jensen wake effect model as follows: Assuming there are N wind turbines in the wind farm, the wake wind speed profile generated by the i-th wind turbine can be expressed as: v i =v0(1-δv i ) (1) Where v0 represents the original (free) wind speed, δv i Represents the total wind speed loss; The total wind speed loss caused by multiple upstream wind turbines can be expressed as: Among them, D j C represents the blade diameter of the j-th wind turbine; Tj x is the thrust coefficient of the j-th wind turbine, used to describe the nonlinear relationship between the pitch angle and the tip speed ratio; i -x j represents the distance between the j-th upstream fan and the i-th downstream fan; k represents the roughness coefficient; The area of ​​overlap between the wake-shielded region of the j-th wind turbine and the swept region of the i-th wind turbine can be calculated using the following formula: Where, d ij L represents the distance between the center of the downstream wind turbine and the center of the wake effect. ij This indicates the distance between the center of the wake effect and the shielding area.

3. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 1, characterized in that: In step 2), establishing the multi-directional wake effect model specifically involves the following: For downstream wind turbines located in the far wake region, their wind speed is expressed as follows: v x The wind speed is the downstream fan speed, x is the horizontal distance between the downstream and upstream fans, and r is the horizontal distance between the upstream and downstream fans. m It is the downstream radius of the upstream wind turbine, η is the wake expansion coefficient, and C T It is the thrust coefficient. Based on wind speed and direction, fuzzy C-means clustering was used to identify 8 sets of characteristic wind directions; D c =[D c1 D c2 D c3 D c4 D c5 D c6 D c7 D c8 ] (5) The characteristic wind speeds corresponding to each wind direction are shown below: v ij Let represent the j-th characteristic wind speed under the i-th characteristic wind direction, and n represent the total number of modes of characteristic wind direction; the wind rose diagram is obtained by using the fuzzy C-means clustering method. Then, a single-wind-direction wake field model is constructed, and the classic Jensen wake model is modified by introducing a sum-of-squares superposition model method. Its expression is: v i and v j Let v represent the wind speeds received by the i-th downstream wind turbine and the j-th upstream wind turbine, respectively, and let τ be the number of upstream wind turbines that directly affect the i-th downstream wind turbine. ji This represents the wind speed of the j-th upstream fan in the wake region flowing towards the i-th downstream fan; Based on the sum of squares superposition model and the wake effect model, the wind speed indirectly experienced by each wind turbine under the characteristic wind direction is as follows: V i It is the set of wind speeds received under the i-th characteristic wind direction. ξ is the wind speed received by the j-th wind turbine under the i-th characteristic wind direction, and ξ is the total number of wind turbines in the wind farm. Therefore, the single-direction wind speed receiving model considering the multi-directional wake field effect is as follows: μ ij The Gaussian membership function is expressed as follows: d j D is the actual wind speed direction of the j-th fan. ci σ is the expected value of the distribution of the i-th characteristic wind direction, and σ is the width of the Gaussian membership function.

4. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 3, characterized in that: Step 3) specifically refers to the following formula for a wind turbine in synchronous operation: T a It is the mechanical time constant, P M and P E These are mechanical power and electromagnetic power, where D is the damping coefficient and ω is the electromagnetic power. r It is the rotor speed, let Replace ω r ,available: The dynamic equation for the power angle of the wind turbine is: ω sf ω is the output command value for Pf droop frequency adjustment control. g Here, f is the grid synchronization angular velocity, and f0 is the grid frequency. In Pf droop control, we can derive: In sf -w0=K p (P ref -P E ) (14) ω0 is the rated speed, K p It is the droop coefficient, P ref It is the active power command value output by the frequency response of the load shedding curve, as shown in the following formula: K DL It is the load shedding factor. The load shedding control measures of wind turbines are designed to ensure that they can release additional power to provide frequency support. During load shedding operation, the rotor speed is fixed at the overspeed operation point, denoted as A', B' and C'. The optimal mechanical power expression is: K W This indicates the optimal energy conversion efficiency of the wind turbine under maximum power point tracking operation, which increases with increasing wind speed; substituting formulas (14)-(16) into (12)-(13), we can obtain: As can be seen from formula (17a-b), the rotor angular dynamics and the power grid power angular dynamics are coupled with each other.

5. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 4, characterized in that: Step 4) Specifically, the synchronization characteristics of the wind turbine are determined by the intersection of equations (17a) and (17b). That is, the existence of an intersection between the two curves or the existence of an overlapping area can ensure that the wind turbine remains synchronized with the grid under under-frequency or over-frequency conditions. The size of the overlapping area can be used as a key indicator to measure the frequency support capability of a single wind turbine, and is called the frequency support margin (FSM). in The critical frequency drop value is:

6. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 5, characterized in that: In step 5), to better cluster based on the frequency support capability of wind turbines, fuzzy K-means clustering is adopted, and its clustering process is as follows: First, obtain the original dataset; Based on the above characteristic wind direction analysis, a frequency support capacity vector (FSCV) is defined for a single wind turbine: FSM k (k = 1, 2, ... 8) is the frequency support margin for the k-th characteristic wind direction. It is the critical frequency drop value of the k-th characteristic wind direction, and γ is the dimensionless normalization coefficient, which is used to make the FSM k (k = 1, 2, ... 8) and The magnitudes are consistent, thus ensuring that the two are comparable; For all the wind turbines in the wind farm, the original dataset is: A=(A1,A2,…,A N (21) Then, initialize the cluster centers and the membership matrix of each wind turbine to the corresponding cluster group; Suppose that N wind turbines are divided into K clusters, denoted as C1, C2, ..., C60. K The centers of each cluster are R1, R2, ..., Rn. K The initial membership matrix is ​​denoted as U = [u ij ] N×K , where u ij Let represent the membership function of the i-th wind turbine to the j-th cluster group; furthermore, the fuzzy index is denoted as m > 1; the objective function is defined as follows: Among them, the Euclidean norm is used to represent A. i With R j The distance between them. Finally, update the cluster center / membership matrix; Optimal clustering can be achieved by minimizing the objective function equation (22) using the Lagrange multiplier method. Therefore, the membership function and cluster centers can be updated alternately. With the current cluster centers fixed, the membership function is updated as follows: After the membership function is updated once according to equation (23), the cluster centers will be updated accordingly as follows: As each iteration progresses, the objective function J... m The value will gradually decrease, and the iterative process will terminate when the following condition is met: Where ε is the termination threshold; Finally, the cluster group to which the i-th wind turbine belongs can be determined by the membership matrix U = [u ij ] N×K Then select the column with the largest membership function value in the i-th row to determine it.

7. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 6, characterized in that: Step 6) includes the dynamic evolution of rotational speed; When a certain wind turbine is allocated a capacity of P ESS For the energy storage system, then for the grid synchronization characteristic equation (17), this is equivalent to increasing the mechanical power by P. ESS The electromagnetic power shortage caused by insufficient mechanical power can be compensated for by an energy storage system; under this equivalent condition, the dynamic evolution of the wind turbine's speed is as follows: When the provided power of the energy storage system can restore transient equilibrium, the frequency support margin of the energy storage system can be approximately expressed as follows: Where ω1, ω2, and η are the same as in formula (18); In addition to providing extra power during the frequency recovery phase to improve the equivalent frequency support margin of wind turbines, energy storage systems can also assist wind turbines during the speed recovery phase to prevent secondary frequency drops. When the rotor kinetic energy of the wind turbine is extracted for frequency support during time period t0, the rotational speed will continue to decrease to time period t1; during this stage, the maximum energy released can be expressed as: Where, ω r0 Let ω be the initial rotational speed. rc Let ω be the critical speed and J be the moment of inertia. In time interval t1, when the speed is about to reach the critical value ω... rc When (or the grid frequency has been restored), the rotational speed will begin to recover, and the rotor will absorb energy from the wind to increase its angular velocity, resulting in a decrease in power output; at time t2, the rotational speed reaches its maximum value and begins to converge to the initial level; At time t1, the rotational speed needs to recover; therefore, the energy storage system will release a constant power P. Ecom To compensate for the reduced wind turbine power, the calculation formula is as follows: When the rotational speed increases to its peak value, the energy storage system power reference value P E1 It should be k ramp The slope is reduced to ensure power balance between load and resources. To minimize the impact on system frequency, an additional frequency response loop is added, which is correlated with the frequency deviation after time period t2. Ultimately, the energy storage system power output command is P. E1 +P Ef .

8. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 7, characterized in that: Specifically, step 7) involves the energy storage system being allocated to the wind turbine cluster, with each energy storage system responsible for providing support to the wind turbines within the same cluster. When allocating energy storage systems, a key consideration is ensuring a high degree of consistency in frequency support margin among wind turbines within the wind farm. This ensures that, in the event of a severe frequency drop, all wind turbines can respond collaboratively to grid frequency changes without any turbines disconnecting from the grid. Therefore, the proposed optimal energy storage system capacity configuration scheme is as follows: Where α and β represent the weighting coefficients of the consistency index and total cost, respectively; σ is the unit power cost of the energy storage system; P ESS,j The energy storage system power capacity allocated to the j-th wind turbine group; I Coherency The consistency index is used to reflect the similarity of wind turbine units in terms of frequency support margin. It is calculated based on the difference between the frequency support margin value of each wind turbine unit and the average frequency support margin value of K wind turbine units and 8 characteristic wind directions. This represents the frequency support margin value of the j-th wind turbine under the i-th characteristic wind direction; This represents the frequency support margin value when the l-th wind turbine is equipped with an energy storage system in the i-th characteristic direction; C represents the frequency support margin value of the l-th wind turbine unit without an energy storage system in the i-th characteristic direction; j Let be the total number of wind turbines in the j-th wind turbine unit; Represents the total budget; Λ represents the power shortage value that causes the frequency drop Δf; t 1,l and t 2,l These represent the start and end times of the speed recovery of the l-th wind turbine, respectively; t 0,l t 1,l and t 2,l These represent the time when the l-th wind turbine begins to reduce its speed, recovers its speed, and completes the speed recovery, respectively (consistent with the time markings in Figure 10); This represents the time period during which the j-th wind turbine first begins to recover its rotational speed. This represents the time period during which the last wind turbine in the j-th wind turbine unit completes speed recovery.

9. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 8, characterized in that: To ensure that the energy storage system can provide sufficient power to restore the wind turbine's rotational speed, the following formula is used: Ensure that the energy storage system always maintains sufficient energy to support frequency stability control. 1＝α+β,0<α,β<1(30n)。 10. The configuration method of a non-array wind farm energy storage system with multi-directional wake effect according to claim 2, characterized in that: In step 8), the consistency index is calculated by integrating the frequency support margin index over the time interval t0 to t1 to ensure a higher level of consistency for the wind turbines throughout the frequency support process. To simplify the solution process, the consistency index is converted into a discrete form, as shown below: Where Δt is the discretization time step, and n l The total number of deviations is calculated using the formula n. l =(t 1l -t 0l ) / Δt.

Images